- Remove ENV_VARS_BY_VERSION[23] entry: adding optional env vars
does not require a config version bump (deep-merge handles it)
- Replace change-detector test (assert _config_version == 23) with
invariant test (assert positive int)
- Add TinyFish case to setup.py missing_browser_hint
- Add TINYFISH_BROWSER_TIMEOUT to set_config_value allowed keys
- Add contributor simantak-dabhade to AUTHOR_MAP
Adds TinyFish (tinyfish.ai) as a cloud browser provider alongside
Browserbase, Browser Use, and Firecrawl. Sessions are created via a
simple POST that returns a CDP websocket URL.
- tools/browser_providers/tinyfish.py — TinyFishBrowserProvider
- tools/browser_tool.py — register in _PROVIDER_REGISTRY
- hermes_cli/tools_config.py — add to onboarding provider picker
- hermes_cli/config.py — TINYFISH_API_KEY env var entries
- hermes_cli/nous_subscription.py — browser label + feature state
- website/docs — document env vars and setup
Based on PR #6329 by @simantak-dabhade.
Enable OpenRouter's response caching feature (beta) via X-OpenRouter-Cache
headers. When enabled, identical API requests return cached responses for
free (zero billing), reducing both latency and cost.
Configuration via config.yaml:
openrouter:
response_cache: true # default: on
response_cache_ttl: 300 # 1-86400 seconds
Changes:
- Add openrouter config section to DEFAULT_CONFIG (response_cache + TTL)
- Add build_or_headers() in auxiliary_client.py that builds attribution
headers plus optional cache headers based on config
- Replace inline _OR_HEADERS dicts with build_or_headers() at all 5 sites:
run_agent.py __init__, _apply_client_headers_for_base_url(), and
auxiliary_client.py _try_openrouter() + _to_async_client()
- Add _check_openrouter_cache_status() method to AIAgent that reads
X-OpenRouter-Cache-Status from streaming response headers and logs
HIT/MISS status
- Document in cli-config.yaml.example
- Add 28 tests (22 unit + 6 integration)
Ref: https://openrouter.ai/docs/guides/features/response-caching
When send_message tool is called from inside a running gateway, the
_run_async bridge spawns a worker thread with a separate event loop.
send_weixin_direct then reuses the live adapter's aiohttp session
which was created on the gateway's main loop. aiohttp's TimerContext
checks asyncio.current_task(loop=session._loop) and sees None because
we're executing on the worker thread's loop → raises 'Timeout context
manager should be used inside a task'.
Fix: skip the live-adapter shortcut when the session belongs to a
different event loop, falling through to the fresh-session path.
Point users to xAI's custom voices feature — clone your voice in the
console, paste the voice_id into tts.xai.voice_id. No code changes
needed; the existing TTS pipeline already handles arbitrary voice IDs.
- config.py: link to xAI custom voices docs in voice_id comment
- setup.py: prompt accepts custom voice IDs during xAI TTS setup
- tts.md: short section linking to xAI console and docs
When resolve_provider_client() passes explicit_api_key for OpenRouter auxiliary
tasks, _try_openrouter() now accepts and honors this parameter instead of
silently ignoring it and falling back to OPENROUTER_API_KEY env var.
Root cause: _try_openrouter() had no explicit_api_key parameter, so even
when callers wanted to pass a runtime credential pool key, it could not be used.
Fix:
- Add explicit_api_key: str = None parameter to _try_openrouter()
- Prioritize explicit_api_key over pool key and env var
- Update resolve_provider_client() call site to pass explicit_api_key
Regression coverage:
- Test that explicit_api_key is passed to OpenAI client when provided
- Test that fallback to OPENROUTER_API_KEY still works when explicit_api_key is None
Closes#18338
Two mitigations for the CLOSE_WAIT accumulation reported against QQ Bot
+ Feishu on macOS behind Cloudflare Warp.
1. Shared httpx.Limits helper (gateway/platforms/_http_client_limits.py).
Every long-lived platform adapter now constructs httpx.AsyncClient
with max_keepalive_connections=10 and keepalive_expiry=2.0, vs httpx's
default of unbounded keepalive pool and 5.0s expiry. On macOS/Warp the
default 5s window let idle keepalive sockets sit in CLOSE_WAIT long
enough for seven persistent adapters (QQ Bot, WeCom, DingTalk, Signal,
BlueBubbles, WeCom-callback, plus the transient Feishu helper) to
compound to the 256-fd ulimit. Tunable via
HERMES_GATEWAY_HTTPX_KEEPALIVE_EXPIRY and
HERMES_GATEWAY_HTTPX_MAX_KEEPALIVE env vars.
2. whatsapp.send_typing aiohttp leak. The call was
'await self._http_session.post(...)' with no 'async with' and no
variable capture — the ClientResponse went out of scope unclosed,
holding its TCP socket in CLOSE_WAIT until GC. Fixed by wrapping in
'async with'. This was the only bare-await aiohttp leak in the
gateway/tools/plugins tree per audit; all other aiohttp sites use
the context-manager pattern correctly.
The underlying reporter also saw Feishu SDK (lark-oapi) connections in
CLOSE_WAIT — those are inside the SDK and out of our direct control, but
tightening httpx keepalive across adapters reduces the aggregate pool
pressure regardless of which individual adapter leaks.
Snapshot Content-Type and body while the client context is still
active so pooled connections fully release on exit. Previously the
read happened after `async with httpx.AsyncClient(...)` returned —
which works today only because httpx eagerly buffers non-streaming
responses; a future refactor to `.stream()` would silently read-
after-close.
Part of the #18451 connection-hygiene audit. Salvage of #18502.
Regression from the silent config→env bridge. The bridge at module import
time is correct for max_turns (unconditional overwrite), but every other
agent.*, display.*, timezone, and security bridge key was guarded by
'if X not in os.environ' — so a stale .env entry from an old 'hermes setup'
run would shadow the user's current config.yaml indefinitely.
Symptom: agent.max_turns: 500 in config.yaml, HERMES_MAX_ITERATIONS=60
in .env from an old setup, and the gateway silently capped at 60
iterations per turn. Gateway logs confirmed api_calls never exceeded 60.
Three changes:
1. gateway/run.py: drop the 'not in os.environ' guards for all agent.*,
display.*, timezone, and security.* bridge keys. config.yaml is now
authoritative for these settings — same semantics already in place
for max_turns, terminal.*, and auxiliary.*. Also surface the bridge
failure (previously 'except Exception: pass') to stderr so operators
see bridge errors instead of silently falling back to .env.
2. gateway/run.py: INFO-log the resolved max_iterations at gateway
start so operators can verify the config→env bridge did the right
thing instead of chasing a phantom budget ceiling.
3. hermes_cli/setup.py: stop writing HERMES_MAX_ITERATIONS to .env in
the setup wizard. config.yaml is the single source of truth. Also
clean up any stale .env entry left behind by pre-fix setups.
Regression tests in tests/gateway/test_config_env_bridge_authority.py
guard each config→env key against the 'stale .env shadows config' bug.
When a provider's credential pool has a single entry in 429-cooldown,
resolve_provider_client returns None and AIAgent.__init__ raises a
misleading RuntimeError suggesting the API key is missing — even when
valid fallback_providers are configured.
This patch makes __init__ iterate the fallback chain before raising,
mirroring the existing in-flight fallback logic in the request loop.
If a fallback resolves, the agent initializes against it and sets
_fallback_activated=True so _restore_primary_runtime can pick the
primary back up after cooldown.
Closes#17929
* fix(gateway): config.yaml wins over .env for agent/display/timezone settings
Regression from the silent config→env bridge. The bridge at module import
time is correct for max_turns (unconditional overwrite), but every other
agent.*, display.*, timezone, and security bridge key was guarded by
'if X not in os.environ' — so a stale .env entry from an old 'hermes setup'
run would shadow the user's current config.yaml indefinitely.
Symptom: agent.max_turns: 500 in config.yaml, HERMES_MAX_ITERATIONS=60
in .env from an old setup, and the gateway silently capped at 60
iterations per turn. Gateway logs confirmed api_calls never exceeded 60.
Three changes:
1. gateway/run.py: drop the 'not in os.environ' guards for all agent.*,
display.*, timezone, and security.* bridge keys. config.yaml is now
authoritative for these settings — same semantics already in place
for max_turns, terminal.*, and auxiliary.*. Also surface the bridge
failure (previously 'except Exception: pass') to stderr so operators
see bridge errors instead of silently falling back to .env.
2. gateway/run.py: INFO-log the resolved max_iterations at gateway
start so operators can verify the config→env bridge did the right
thing instead of chasing a phantom budget ceiling.
3. hermes_cli/setup.py: stop writing HERMES_MAX_ITERATIONS to .env in
the setup wizard. config.yaml is the single source of truth. Also
clean up any stale .env entry left behind by pre-fix setups.
Regression tests in tests/gateway/test_config_env_bridge_authority.py
guard each config→env key against the 'stale .env shadows config' bug.
* fix(gateway): shutdown + restart hygiene (drain timeout, false-fatal, success log)
Three issues observed in production gateway.log during a rapid restart
chain on 2026-05-02, all fixed here.
1. _send_restart_notification logged unconditional success
adapter.send() catches provider errors (e.g. Telegram 'Chat not found')
and returns SendResult(success=False); it never raises. The caller
ignored the return value and always logged 'Sent restart notification
to <chat>' at INFO, producing a misleading success line directly
below the 'Failed to send Telegram message' traceback on every boot.
Now inspects result.success and logs WARNING with the error otherwise.
2. WhatsApp bridge SIGTERM on shutdown classified as fatal error
_check_managed_bridge_exit() saw the bridge's returncode -15 (our own
SIGTERM from disconnect()) and fired the full fatal-error path,
producing 'ERROR ... WhatsApp bridge process exited unexpectedly' plus
'Fatal whatsapp adapter error (whatsapp_bridge_exited)' on every
planned shutdown, immediately before the normal '✓ whatsapp
disconnected'. Adds a _shutting_down flag that disconnect() sets
before the terminate, and _check_managed_bridge_exit() returns None
for returncode in {0, -2, -15} while shutting down. OOM-kill (137)
and other non-signal exits still hit the fatal path.
3. restart_drain_timeout default 60s → 180s
On 2026-05-02 01:43:27 a user /restart fired while three agents were
mid-API-call (82s, 112s, 154s into their turns). The 60s drain budget
expired and all three were force-interrupted. 180s covers realistic
in-flight agent turns; users on very-long-reasoning models can still
raise it further via agent.restart_drain_timeout in config.yaml.
Existing explicit user values are preserved by deep-merge.
Tests
- tests/gateway/test_restart_notification.py: two new tests assert INFO
is only logged on SendResult(success=True) and WARNING with the error
string is logged on SendResult(success=False).
- tests/gateway/test_whatsapp_connect.py: parametrized test for
returncode in {0, -2, -15} proves shutdown-time exits are suppressed;
separate test proves returncode 137 (SIGKILL/OOM) still surfaces as
fatal even when _shutting_down is set.
- _check_managed_bridge_exit() reads _shutting_down via getattr-with-
default so existing _make_adapter() test helpers that bypass __init__
(pitfall #17 in AGENTS.md) keep working unmodified.
Two narrow fixes for long pasted messages silently disappearing:
1. _expand_paste_references: replace path.exists() + read_text() with
try/except (OSError, IOError). Closes the TOCTOU window where a paste
file deleted between check and read raised FileNotFoundError, bubbled
up through process_loop's outer except, and silently dropped the
user's input. Failures now return the placeholder text and log a
warning.
2. process_loop outer except: logger.warning() instead of print().
prompt_toolkit's TUI swallows stdout, so 'Error: …' was invisible
to the user. Logged errors are discoverable via hermes logs.
Dropped the larger interrupt_queue→pending_input drain that was part of
the original PR — that's a separate class of input-drop (in-progress
interrupt handling) unrelated to the paste-file TOCTOU reported in the
issue, and worth its own review.
Salvage of #17939.
Discord's per-command name limit is 32 chars. When two skill slugs
share the same first 32 chars (or a skill slug clamps onto a reserved
gateway command name), only the first seen wins — the second is
dropped from the /skill autocomplete. The old behavior incremented a
``hidden`` counter silently, so skill authors had no way to discover
the drop short of noticing their skill was missing from the picker.
Not an actively-biting bug today (no collisions on the default catalog
as of 2026-05), but a landmine the moment someone ships a skill with a
long name. The earlier series in #18745 / #18753 / #18754 dropped the
other silent data-loss paths in the Discord /skill collector; this one
lights up the last remaining one.
Fix: promote ``_names_used`` from a set to a dict keyed by the clamped
name, mapping to the source cmd_key (or a ``"<reserved>"`` sentinel
for names inherited via ``reserved_names``). On collision, log a
WARNING naming both sides — the winner, the loser, the clamped name,
and what to rename.
Two phrasings:
* skill-vs-skill — "both clamp to X on Discord's 32-char command-name
limit; only the winner appears in /skill. Rename one skill's
frontmatter ``name:`` to differ in its first 32 chars."
* skill-vs-reserved — "collides with a reserved gateway command name;
the skill will not appear in /skill. Rename the skill's frontmatter
``name:``."
Tests: three cases in
``tests/hermes_cli/test_discord_skill_clamp_warning.py`` —
skill-vs-skill collision (warning names both cmd_keys + clamped prefix),
skill-vs-reserved collision (warning uses the distinct phrasing), and a
no-collision negative (zero warnings emitted).
Covers PR #18224 fix for issue #18187 — when DiscordAdapter.connect() is
called a second time without an intervening disconnect(), the previous
commands.Bot must be closed before a new one is created. Otherwise both
websockets stay connected to Discord's gateway and both fire on_message,
producing double responses with different wording.
When DiscordAdapter.connect() is called during reconnect, it creates a new
commands.Bot client without closing the previous one. The old client's
websocket remains connected to Discord's gateway, causing both to fire
on_message for every incoming event — resulting in double responses.
Fix: before creating a new Bot instance, check if a previous client exists
and close it. This ensures only one websocket connection is active at any
time.
Closes#18187
Covers PR #18256 fix for issue #18254 — when OPENROUTER_API_KEY is set in
BOTH os.environ (stale from parent shell) and ~/.hermes/.env (fresh),
_seed_from_env must prefer the .env value. Also guards the fallback case
where .env omits the key entirely (Docker/K8s/systemd deployments that
only inject via runtime env).
When _seed_from_env() reads API keys to populate the credential pool, it
should treat ~/.hermes/.env as the authoritative source — not os.environ.
Stale env vars inherited from parent shell processes (Codex CLI, test
scripts, etc.) can shadow deliberate changes to the .env file, causing
auth.json to cache an outdated key that leads to silent 401 errors.
This is especially visible with OpenRouter: if a parent process exported
OPENROUTER_API_KEY=test-key-fresh and the user later updates .env with a
valid key, restarting Hermes still picks up the stale os.environ value,
writes it back to auth.json, and all API calls fail with 401.
Fixes#18254
`_register_skill_group` captured the skill catalog in closure variables
(`entries` and `skill_lookup`) so the single `tree.add_command` call at
startup owned the only live copy. The closure is never re-entered after
startup, so `/reload-skills` — which rescans the on-disk skills dir and
refreshes the in-process `_skill_commands` registry — had no way to
propagate results into the `/skill` autocomplete on Discord. New skills
stayed invisible in the dropdown, and deleted skills returned
"Unknown skill" when the stale autocomplete entry was clicked.
The fix is purely a dataflow change: promote `entries` and `skill_lookup`
to instance attributes (`_skill_entries`, `_skill_lookup`), split the
collector-driven rebuild into a helper (`_refresh_skill_catalog_state`),
and add a public `refresh_skill_group()` method that re-runs the helper
and is safe to call at any point after the initial registration.
The gateway's `_handle_reload_skills_command` then iterates
`self.adapters` and calls `refresh_skill_group()` on any adapter that
exposes it (currently only Discord). Both sync and async implementations
are supported; adapters that don't override the method (Telegram's
BotCommand menu, Slack subcommand map, etc.) are silently skipped — the
in-process `reload_skills()` call covers them.
No `tree.sync()` is required because Discord fetches autocomplete
options dynamically on every keystroke — mutating the instance state the
callbacks already read from is sufficient. That sidesteps the per-app
command-bucket rate limit (~5 writes / 20 s) that made the previous
bulk-sync-on-reload approach unusable (#16713 context).
Tests: tests/gateway/test_reload_skills_discord_resync.py — five cases
covering (1) refresh replaces entries, (2) entries stay sorted after
refresh, (3) collector exception leaves cached state intact, (4)
`_refresh_skill_catalog_state` populates the instance attrs, (5)
orchestrator calls `refresh_skill_group()` on sync + async adapters and
skips adapters that don't expose it.
_check_unavailable_skill is meant to turn a typed "/foo" command that
doesn't resolve into a specific hint — "disabled, enable with hermes
skills config" or "available but not installed, install with hermes
skills install …" — instead of the generic "unknown command" reply.
It was doing the match with `skill_md.parent.name.lower().replace("_", "-")`,
comparing that to the typed command. For every skill whose directory name
drifted from its declared frontmatter `name:`, that comparison failed and
the user got the unhelpful generic path. On a standard install today 19
skills have this drift, e.g.:
dir: mlops/stable-diffusion
frontmatter: name: Stable Diffusion Image Generation
registered slug (what the user types): /stable-diffusion-image-generation
dir: mlops/qdrant
frontmatter: name: Qdrant Vector Search
registered slug: /qdrant-vector-search
dir: mlops/flash-attention
frontmatter: name: Optimizing Attention Flash
registered slug: /optimizing-attention-flash
In every case, _check_unavailable_skill would fall through because
"stable-diffusion" != "stable-diffusion-image-generation", even with the
skill sitting right there on disk.
Fix: extract a small `_skill_slug_from_frontmatter` helper that reads the
SKILL.md frontmatter and normalizes exactly like scan_skill_commands
(lower, spaces/underscores → hyphens, strip non-[a-z0-9-], collapse
runs of hyphens, strip edges). Use it in both the
disabled-skills branch and the optional-skills branch. The disabled-set
membership check now uses the declared frontmatter name (which is what
`hermes skills config` writes into skills.disabled / platform_disabled),
not the slug.
Tests: five cases in tests/gateway/test_unavailable_skill_hint.py —
the drift case for the disabled branch, unknown-command negative,
matched-but-not-disabled negative, non-alnum stripping, and the drift
case for the optional-skills branch. All five fail against main and
pass with the fix.
``discord_skill_commands_by_category`` was lagging the flat
``discord_skill_commands`` collector on two counts. Both were actively
dropping skills from Discord's ``/skill`` autocomplete dropdown.
1. External-dir skills were filtered out. #18741 widened the flat
collector to accept ``SKILLS_DIR + skills.external_dirs`` but left
this sibling collector — the one ``_register_skill_group`` actually
uses on Discord — still matching ``SKILLS_DIR`` only. External
skills were visible in ``hermes skills list`` and the agent's
``/skill-name`` dispatch but silently absent from Discord's
``/skill`` picker. Widen the accepted roots to match, and derive
categories from whichever root the skill lives under so
``<ext>/mlops/foo/SKILL.md`` still lands in the ``mlops`` group.
2. 25-group × 25-subcommand caps were still applied. PR #11580
refactored ``/skill`` to a flat autocomplete (whose options Discord
fetches dynamically — no per-command payload concern) and its
docstring promises "no hidden skills." The collector kept the old
nested-layout caps anyway, silently dropping anything past the 25th
alphabetical category. On installs with 29 category dirs today (real
example: tail categories ``social-media``, ``software-development``,
``yuanbao`` going missing) this was biting immediately. Remove the
caps; ``hidden`` now reports only 32-char name-clamp collisions
against reserved names.
Tests: guard both behaviors. ``test_no_legacy_25x25_cap`` builds 30
categories × 30 skills each and asserts all 900 are returned.
``test_external_dirs_skills_included`` monkeypatches
``get_external_skills_dirs`` and asserts an external-dir skill makes
it into the result grouped under its own top-level directory.
After a transient Telegram 502, _handle_polling_network_error's
stop()+start_polling() cycle can leave PTB's Updater with `running=True`
but a wedged consumer task that never makes progress. No error_callback
fires in that state, so the reconnect ladder never advances past attempt
1, the MAX_NETWORK_RETRIES fatal-error path is never reached, and the
gateway sits silent indefinitely.
Schedule a heartbeat probe (60s after a successful reconnect) that
verifies Updater.running is still True and bot.get_me() responds within
a tight asyncio.wait_for timeout. Either failure feeds back into the
reconnect ladder so the existing escalation path fires.
No PTB-internal coupling, no Application rebuild — minimal additive
defense inside the existing reconnect abstraction.
Tests cover healthy / Updater non-running / probe timeout / probe
network error / already-fatal cases, plus an integration check that the
probe is actually scheduled after a successful start_polling().
Closes the silent-wedge case observed in the wild after a transient
Telegram 502; existing reconnect tests updated to mock bot.get_me() now
that the success path schedules a heartbeat probe.
Providers like Google Vertex, Azure, and Amazon Bedrock reject API
requests with duplicate tool names (HTTP 400: 'Tool names must be
unique'). The upstream injection paths in run_agent.py already dedup
after PR #17335, but two API-boundary functions pass tools through
without checking:
- agent/auxiliary_client.py: _build_call_kwargs() (all non-Anthropic
providers in chat_completions mode)
- agent/anthropic_adapter.py: convert_tools_to_anthropic() (Anthropic
Messages API path)
Add defensive dedup guards at both sites. Duplicates are dropped with
a warning log, converting a hard 400 failure into a recoverable
condition. This is intentionally conservative — the root-cause dedup
in run_agent.py is the primary defense; these guards add resilience
against future injection-path regressions.
Includes 8 new tests covering unique passthrough, duplicate removal,
empty/None edge cases.
Closes#18478
When HERMES_HOME is unset but ~/.hermes/active_profile names a non-default
profile, any data this process writes lands in the default profile — not the
one the operator expects. Before this change the fallback was silent, so
cross-profile contamination (#18594) was invisible until a user noticed
their memory/state ended up in the wrong place.
Now we emit a one-shot warning to stderr the first time this happens in
a process. No raise — there are 30+ module-level callers of get_hermes_home()
and raising from any of them would brick import. Behavior is otherwise
unchanged; subprocess spawners (systemd template, kanban dispatcher, docker
entrypoint) already propagate HERMES_HOME correctly.
Bypasses logging.getLogger() because this runs before logging is configured
in a significant fraction of callers (module import time).
Refs #18594. Credit to @liuhao1024 for surfacing the silent-fallback case
in PR #18600; we kept the diagnostic signal without the import-time raise.
Path.read_text() uses the system locale by default. On Windows CN/JP/KR
locales (GBK/CP932/CP949), reading a UTF-8 .env raises UnicodeDecodeError
as soon as it contains any non-ASCII byte (e.g. an em dash).
Pin encoding="utf-8" on every .env read in hermes_cli to match how the
rest of the codebase (load_dotenv at doctor.py:26) already decodes it.
Adds a regression test that monkeypatches Path.read_text to simulate a
GBK locale and asserts 'hermes doctor' no longer raises.
Refs #18637
Skills configured through `skills.external_dirs` in config.yaml were
visible via `hermes skills list`, `get_skill_commands()`, and the
agent's `/skill-name` dispatch, but silently excluded from the
Telegram and Discord slash-command menus. The filter in
`_collect_gateway_skill_entries` only accepted skills whose
`skill_md_path` started with `SKILLS_DIR`, so anything under an
external directory fell through.
Widen the accepted-prefix set to include all configured external
dirs alongside the local skills dir. Every prefix is now
slash-terminated so `/my-skills` cannot also admit
`/my-skills-extra`. Also guard against empty `skill_md_path`
values so they can't accidentally match.
Fixes#8110
Salvages #8790 by luyao618.
Co-authored-by: Yao <34041715+luyao618@users.noreply.github.com>
The process-global `_skill_commands` dict in agent/skill_commands.py
was seeded by whichever platform scanned first, and
`get_skill_commands()` only rescanned when the cache was empty. In a
long-lived gateway process serving multiple platforms (Telegram +
Discord + Slack), the first platform's
`skills.platform_disabled` view was silently inherited by the
others — so a skill disabled for Telegram would also disappear from
Discord's slash menu, and vice versa.
Track the platform scope the cache was populated for
(`_skill_commands_platform`) and rescan in `get_skill_commands()`
when the currently-active platform no longer matches. Platform
resolution uses the same precedence as `_is_skill_disabled`:
`HERMES_PLATFORM` env var then `HERMES_SESSION_PLATFORM` from the
gateway session context.
Fixes#14536
Salvages #14570 by LeonSGP43.
Co-authored-by: LeonSGP <leon@sgp43.com>
* fix(curator): authoritative absorbed_into declarations on skill delete
Closes#18671. The classification pipeline that feeds cron-ref rewriting
used to infer consolidation vs pruning from two brittle signals: the
curator model's post-hoc YAML summary block, and a substring heuristic
scanning other tool calls for the removed skill's name. Both miss in
real consolidations — the model forgets the YAML under reasoning
pressure, and the heuristic misses when the umbrella's patch content
describes the absorbed behavior abstractly instead of naming the old
slug. When both miss, the skill falls through to 'no-evidence fallback'
pruned, and #18253's cron rewriter drops the cron ref entirely instead
of mapping it to the umbrella. Same observable symptom as pre-#18253:
'Skill(s) not found and skipped' at the next cron run.
The fix makes the model declare intent at the moment of deletion.
skill_manage(action='delete') now accepts absorbed_into:
- absorbed_into='<umbrella>' -> consolidated, target must exist on disk
- absorbed_into='' -> explicit prune, no forwarding target
- missing -> legacy path, falls through to heuristic/YAML
The curator reconciler reads these declarations off llm_meta.tool_calls
BEFORE either the YAML block or the substring heuristic. Declaration
wins. Fallback logic stays intact for backward compat with any caller
(human or older curator conversation) that doesn't populate the arg.
Changes
- tools/skill_manager_tool.py: add absorbed_into param to skill_manage
+ _delete_skill. Validate target exists when non-empty. Reject
absorbed_into=<self>. Wire through dispatcher + registry + schema.
- agent/curator.py: new _extract_absorbed_into_declarations() walks
tool calls for skill_manage(delete) with the arg. _reconcile_classification
accepts absorbed_declarations= and treats them as authoritative. Curator
prompt updated to require the arg on every delete.
- Tests: 7 new skill_manager tests covering the tool contract (valid
target, empty string, nonexistent target, self-reference, whitespace,
backward compat, dispatcher plumbing). 11 new curator tests covering
the extractor + authoritative reconciler path + mixed-legacy-and-
declared runs.
Validation
- 307/307 targeted tests pass (curator + cron + skill_manager suites).
- E2E #18671 repro: 3 narrow skills, 1 umbrella, cron job referencing
all 3. Model emits NO YAML block. Heuristic misses (patch prose
doesn't name old slugs). Delete calls carry absorbed_into. Result:
both PR skills correctly classified 'consolidated' + cron rewritten
['pr-review-format', 'pr-review-checklist', 'stale-junk'] ->
['hermes-agent-dev']; stale-junk pruned via absorbed_into=''.
- E2E backward-compat: delete without absorbed_into, model emits YAML
-> routed via existing 'model' source, cron still rewritten correctly.
* feat(curator): capture + restore cron skill links across snapshot/rollback
Before this, rolling back a curator run restored the skills tree but cron
jobs still pointed at the umbrella skills the curator had rewritten them
to. The user would see their old narrow skills back on disk but their
cron jobs still configured with the merged umbrella — not actually 'back
to how it was'.
Snapshot side: snapshot_skills() now captures ~/.hermes/cron/jobs.json
alongside the skills tarball, as cron-jobs.json. The manifest gets a new
'cron_jobs' block with {backed_up, jobs_count} so rollback (and the CLI
confirm dialog) can surface what's in the snapshot. If jobs.json is
missing/unreadable/malformed, snapshot proceeds without cron data — the
skills backup is the core guarantee; cron is additive.
Rollback side: after the skills extract succeeds, the new
_restore_cron_skill_links() reconciles the backed-up jobs into the live
jobs.json SURGICALLY. Only 'skills' and 'skill' fields are restored, and
only on jobs matched by id. Everything else about a cron job — schedule,
last_run_at, next_run_at, enabled, prompt, workdir, hooks — is live
state the user or scheduler has modified since the snapshot; overwriting
it would regress unrelated activity.
Reconciliation rules:
- Job in backup AND live, skills differ → skills restored.
- Job in backup AND live, skills match → no-op.
- Job in backup, NOT in live → skipped (user deleted it
after snapshot; their choice
is later than the snapshot).
- Job in live, NOT in backup → untouched (user created it
after snapshot).
- Snapshot missing cron-jobs.json at all → rollback still succeeds,
reports 'not captured'
(older pre-feature snapshots
keep working).
Writes go through cron.jobs.save_jobs under the same _jobs_file_lock the
scheduler uses, so rollback doesn't race tick().
Also:
- hermes_cli/curator.py: rollback confirm dialog now shows
'cron jobs: N (will be restored for skill-link fields only)' when the
snapshot has cron data, or 'not in snapshot (<reason>)' otherwise.
- rollback()'s message string includes a 'cron links: ...' clause
summarizing the reconciliation outcome.
Tests
- 9 new cases: snapshot-with-cron, snapshot-without-cron, malformed-json
captured-as-raw, full rollback-restores-skills-and-cron, rollback
touches only skill fields, rollback skips user-deleted jobs, rollback
leaves user-created jobs untouched, rollback still works with
pre-feature snapshot that has no cron-jobs.json, standalone unit test
on _restore_cron_skill_links exercising the full report shape.
Validation
- 484/484 targeted tests pass (curator + cron + skill_manager suites).
- E2E: real snapshot_skills, real cron rewrite, real rollback. Before:
['pr-review-format', 'pr-review-checklist', 'pr-triage-salvage'].
After curator: ['hermes-agent-dev']. After rollback: ['pr-review-format',
'pr-review-checklist', 'pr-triage-salvage']. Non-skill fields (id,
name, prompt) preserved across the round trip.
The old defaults (StartLimitIntervalSec=600, StartLimitBurst=5,
RestartSec=30) meant any network outage over ~5 minutes would
permanently kill the gateway until manual intervention.
Changes:
- StartLimitIntervalSec=0 (never give up)
- Restart=always (not just on-failure)
- RestartSec=60 with RestartMaxDelaySec=300, RestartSteps=5
(exponential backoff: 60 → 120 → 180 → 240 → 300s cap)
- After=network-online.target + Wants= (both units now wait for
actual connectivity, not just network.target)
Power outage → internet down → internet back = auto-recovery.
When the dashboard is bound to 0.0.0.0 with --insecure (e.g. behind
Tailscale Serve), WebSocket endpoints (/api/pty, /api/ws, /api/pub,
/api/events) rejected connections from non-loopback client IPs with
code 4403 — causing 'events feed disconnected' in the UI.
Extract the repeated loopback check into _ws_client_is_allowed() which
respects the public bind flag. Session token auth still guards all
endpoints regardless of bind mode.
FixesNousResearch/hermes-agent#11768
Root cause: target.strip().lower() was lowercasing the entire target string,
corrupting case-sensitive chat IDs like Slack C123ABC and Matrix !RoomABC.
Fix: Only lowercase the platform prefix for case-insensitive matching;
preserve the original case for chat_id and thread_id values.
YAML loads a bare numeric value such as
discord:
free_response_channels: 1491973769726791812
as an int. _discord_free_response_channels() / _slack_free_response_channels()
checked `isinstance(raw, list)` and `isinstance(raw, str)` in that order and
then fell through to `return set()`, so a single-channel config that happened
to be unquoted was silently dropped with no log line — the bot kept demanding
@mentions even though the channel was configured to free-response.
A multi-channel value like `1234567890,9876543210` does not trip this because
the comma forces YAML to parse it as a string. Single-channel configs are
the only case that breaks, which is exactly the footgun that's hardest to
diagnose (the config "looks right" and the feature just doesn't activate).
Note that the old-schema env-var bridge at gateway/config.py:614+ already
runs `str(frc)` when forwarding to SLACK_/DISCORD_FREE_RESPONSE_CHANNELS,
so the env-var fallback worked. The bug only surfaces on the
`config.extra["free_response_channels"]` path populated by the `platforms:`
bridge at gateway/config.py:576, which passes the raw YAML value through
unchanged.
Fix at the reader: treat any non-list value as a scalar, coerce with str(),
then apply the same CSV split semantics. This keeps the public contract
stable (list or str-like continues to work identically) while accepting
the ints that the YAML loader is free to hand us.
Added tests for both Discord and Slack covering:
- bare int value in config.extra
- list of ints in config.extra
Slack has built-in slash commands (e.g. /status, /me, /join) that apps
cannot register. When running `hermes slack manifest --write`, the
generated manifest included /status, causing Slack to reject the entire
manifest with a reserved-command error.
Add _SLACK_RESERVED_COMMANDS frozenset of all known Slack built-ins and
skip them in slack_native_slashes(). Affected commands remain reachable
via /hermes <command>.
Tests updated:
- New test_excludes_slack_reserved_commands validates no leaks
- test_includes_canonical_commands no longer asserts /status
- test_telegram_parity accounts for expected Slack-only exclusions
Self-review fixes for the slash ephemeral ack:
- Only stash response_url when text starts with '/' (gateway command).
Free-form questions via '/hermes <question>' must produce public agent
replies visible to the whole channel, not ephemeral.
- Use a ContextVar (_slash_user_id) to thread the invoking user's ID
from _handle_slash_command through to send(). _pop_slash_context now
matches the exact (channel_id, user_id) key when the ContextVar is
set, preventing concurrent users on the same channel from stealing
each other's ephemeral context. ContextVars propagate to child
asyncio.Tasks, so the value survives through handle_message →
_process_message_background → _send_with_retry → send().
- Add truncate_message() in _send_slash_ephemeral to prevent silent
failures on long responses (response_url has the same ~40k limit).
- Log send_private_notice failures at debug level instead of bare
except/pass — aids diagnostics without spamming.
- Document app_mention dedup dependency on shared event ts.
- Add tests: free-form question must NOT stash context, concurrent
users on the same channel get isolated contexts, non-slash send()
path fallback behavior.
Adds platform-level private notice delivery abstraction so operational
messages (e.g. sethome prompt) can be sent ephemerally on Slack when
configured with `slack.notice_delivery: private`.
Changes:
- gateway/config.py: _normalize_notice_delivery() + GatewayConfig.get_notice_delivery()
with per-platform config bridging
- gateway/platforms/base.py: send_private_notice() default implementation
(falls through to send())
- gateway/platforms/slack.py: send_private_notice() via chat_postEphemeral
- gateway/run.py: _deliver_platform_notice() helper replaces direct
adapter.send() for the sethome notice, with private→public fallback
- gateway/platforms/slack.py: app_mention handler now forwards to
_handle_slack_message (safe due to ts-based dedup) instead of no-op pass,
fixing edge-case Slack configs where mentions arrive only as app_mention
- gateway/platforms/slack.py format_message: negative lookbehind prevents
markdown images (![]()) from becoming broken Slack links; italic regex
now requires non-whitespace boundaries so 'a * b * c' stays literal
Based on PR #9340 by @probepark.
Slack slash commands (/q, /btw, /stop, /model, etc.) previously showed
no user-visible acknowledgement and posted command replies as public
channel messages. This diverged from Discord, which uses ephemeral
deferred responses for slash commands.
Changes:
- handle_hermes_command now passes response_type='ephemeral' and a
'Running /cmd…' text to ack(), giving the user immediate 'Only visible
to you' feedback when they invoke any native slash command.
- _handle_slash_command stashes the Slack response_url from the command
payload in a per-channel context dict before dispatching to
handle_message.
- send() checks for a pending slash context and, when found, POSTs to
the response_url with replace_original=true to swap the initial ack
with the real command reply (e.g. 'Queued for the next turn.'),
keeping it ephemeral.
- Stale slash contexts are garbage-collected on lookup (120s TTL).
- The response_url POST is non-fatal: if it fails, the user already saw
the initial ack, and send() returns success=True.
Fixes#18182
Long-running gateway processes that survive 'hermes update' keep
pre-update modules cached in sys.modules. When new tool files on
disk then try to 'from hermes_cli.config import cfg_get' (added in
PR #17304), the import resolves against the stale module object
and raises ImportError — hitting users on Matrix, Telegram, Feishu,
and other platforms.
Two defenses:
1. Gateway self-check (gateway/run.py). On __init__, snapshot the
newest mtime across sentinel source files (hermes_cli/config.py,
run_agent.py, gateway/run.py, etc.). On every inbound message,
re-read those mtimes; if any is newer than boot time + 2s slack,
request a graceful restart via the normal drain path and return
a one-line ack to the user. Idempotent, works regardless of how
the update happened (hermes update, manual git pull, installer).
2. Post-restart survivor sweep ('hermes update'). After the existing
restart loop, sleep 3s, rescan for gateway PIDs we already tried
to kill, and SIGKILL any survivors. The detached profile watchers
and systemd then relaunch with fresh code instead of waiting out
the 120s watcher timeout.
Closes#17648.
* fix(curator): defer first run and add --dry-run preview (#18373)
Curator was meant to run 7 days after install, not on the very first
gateway tick. On a fresh install (no .curator_state), should_run_now()
returned True immediately because last_run_at was None — so the gateway
cron ticker fired Curator against a fresh skill library moments after
'hermes update'. Combined with the binary 'agent-created' provenance
model (anything not bundled and not hub-installed), this consolidated
hand-authored user workflow skills without consent.
Changes:
- should_run_now(): first observation seeds last_run_at='now' and returns
False. The next real pass fires one full interval_hours later (7 days
by default), matching the original design intent.
- hermes curator run --dry-run: produces the same review report without
applying automatic transitions OR permitting the LLM to call
skill_manage / terminal mv. A DRY-RUN banner is prepended to the
prompt and the caller skips apply_automatic_transitions. State is
NOT advanced so a preview doesn't defer the next scheduled real pass.
- hermes update: prints a one-liner on fresh installs pointing at
--dry-run, pause, and the docs. Silent on steady state.
- Docs: curator.md and cli-commands.md explain the deferred first-run
behavior and warn that hand-written SKILL.md files share the
'agent-created' bucket, with guidance to pin or preview before the
first pass.
Tests:
- test_first_run_defers replaces the old 'first run always eligible'
assertion — same fixture, inverted expectation.
- test_maybe_run_curator_defers_on_fresh_install covers the gateway tick
path end-to-end.
- Three new dry-run tests cover state-advance suppression, prompt
banner injection, and apply_automatic_transitions skipping.
Fixes#18373.
* feat(curator): pre-run backup + rollback (#18373)
Every real curator pass now snapshots ~/.hermes/skills/ into
~/.hermes/skills/.curator_backups/<utc-iso>/skills.tar.gz before calling
apply_automatic_transitions or the LLM review. If a run consolidates or
archives something the user didn't want touched, 'hermes curator
rollback' restores the tree in one command. Dry-run is skipped — no
mutation means no snapshot needed.
Changes:
- agent/curator_backup.py (new): tar.gz snapshot + safe rollback. The
snapshot excludes .curator_backups/ (would recurse) and .hub/ (managed
by the skills hub). Extract refuses absolute paths and .. components,
and uses tarfile's filter='data' on Python 3.12+. Rollback takes a
pre-rollback safety snapshot FIRST, stages the current tree into
.rollback-staging-<ts>/ so the extract lands in an empty dir, and
cleans the staging dir on success. A failed extract restores the
staged contents.
- agent/curator.py: run_curator_review() calls curator_backup.
snapshot_skills(reason='pre-curator-run') before apply_automatic_
transitions. Best-effort — a failed snapshot logs at debug and the
run continues (a transient disk issue shouldn't silently disable
curator forever).
- hermes_cli/curator.py: new 'hermes curator backup' and 'hermes curator
rollback' subcommands. rollback supports --list, --id <ts>, -y.
- hermes_cli/config.py: curator.backup.{enabled, keep} config block
with sane defaults (enabled=true, keep=5).
- Docs: curator.md gets a 'Backups and rollback' section; cli-commands
.md table gets the new rows.
Tests (new file tests/agent/test_curator_backup.py, 16 cases):
- snapshot creates tarball + manifest with correct counts
- snapshot excludes .curator_backups/ (recursion guard) and .hub/
- snapshot disabled via config returns None without creating anything
- snapshot uniquifies ids within the same second (-01 suffix)
- prune honors keep count, newest-first
- list_backups + _resolve_backup cover newest-default and unknown-id
- rollback restores a deleted skill with content intact
- rollback is itself undoable — safety snapshot shows up in list_backups
- rollback with no snapshots returns an error
- rollback refuses tarballs with absolute paths or .. components
- real curator runs take a 'pre-curator-run' snapshot; dry-runs do not
All curator tests: 210 passing locally.
Prevents ghost sessions from accumulating in state.db when the TUI/web
dashboard is opened and closed without sending a message.
Changes:
- run_agent.py: Add _ensure_db_session() gate method, called at
run_conversation() entry. Remove eager create_session() from __init__.
Handle compression rotation flag correctly.
- tui_gateway/server.py: Remove eager db.create_session() in
_start_agent_build(). Add post-first-message pending_title re-apply.
- hermes_state.py: Extract _insert_session_row() shared helper (DRY).
Add prune_empty_ghost_sessions() for one-time migration.
- cli.py: One-time ghost session prune on startup. Fix _pending_title
to call _ensure_db_session() before set_session_title().
- hermes_cli/main.py: Guard TUI exit summary on message_count > 0.
- tests: Update test_860_dedup to call _ensure_db_session() before
direct _flush_messages_to_session_db() calls.
Closes: ghost session clutter in hermes sessions list and web dashboard.
Telegram's client does not display empty forum topics in the chat's
topic list. After createForumTopic succeeds, send a short pin message
into the new topic so it becomes immediately visible to the user.
Only fires for newly created topics (no thread_id in config yet).
Failure to send the seed is non-fatal (debug-logged, topic still works).
The bot-owner identity check inside OwnerCommandMiddleware was commented
out and replaced with a hardcoded `is_owner = True`, so any group member
could trigger allowlisted privileged commands (/approve, /deny, /stop,
/reset, /retry, /undo, /new, /background, /bg, /btw, /queue, /q) by
sending the slash command without @-mentioning the bot. The most severe
case is /approve: a non-owner could approve a dangerous tool call the
bot was waiting on the owner to confirm.
Re-enable the documented identity check (push.from_account ==
push.bot_owner_id) so only the configured owner can issue these
commands.
Adds a new top-of-sidebar docs page at /docs/user-stories that is a
masonry-style collage of 99 real user stories sourced from X/Twitter,
GitHub issues/PRs, Reddit, Hacker News, YouTube, blogs (Medium, Substack,
dev.to), podcasts, LinkedIn, GitHub Gists, and Product Hunt.
Every tile links to the original post/issue/video/gist where someone
described a specific use case: personal assistants, dev workflows,
trading bots, research briefs, family WhatsApp agents, Kubernetes
deployments, legal-domain self-hosted setups, and more.
- docs/user-stories.mdx: MDX entry mounting the collage component
- src/components/UserStoriesCollage: React component with category +
source filters, CSS-columns masonry layout, per-category accent colors
- src/data/userStories.json: source-of-truth dataset (force-added; the
root .gitignore's unanchored 'data/' rule would otherwise swallow it,
same reason skills.json is explicitly listed in website/.gitignore)
- sidebars.ts: link added at the top of the docs sidebar
Four callsites hardcoded Path.home() / '.hermes' with no HERMES_HOME
check, breaking Docker deployments and profile isolation (hermes -p):
- plugins/hermes-achievements/dashboard/plugin_api.py:
state_path(), snapshot_path(), checkpoint_path() bare-literal paths
- scripts/profile-tui.py:
DEFAULT_STATE_DB and DEFAULT_LOG defaults ignored HERMES_HOME
- hermes_cli/slack_cli.py:
except-Exception fallback for slack-manifest.json dump
- optional-skills/migration/openclaw-migration/scripts/openclaw_to_hermes.py:
--target argparse default
Use get_hermes_home() (with an ImportError shim for the standalone
scripts) or 'os.environ.get("HERMES_HOME") or str(Path.home()/".hermes")'
where importing hermes_constants is impractical.
E2E-verified: with HERMES_HOME=/tmp/x all three achievements paths and
both profile-tui defaults route under /tmp/x.
Salvaged from #18068 (original scope was broader mechanical cleanup
claiming 23 callsites were buggy; most were already respecting
HERMES_HOME via os.environ.get(key, default) — only these 4 had no env
check at all). Credit: @web-dev0521.
Two machine-readable entry points to the Hermes Agent docs:
/llms.txt curated index of every doc page, one link per page
with short descriptions. ~17 KB, safe to load into
an LLM context window.
/llms-full.txt every page under website/docs/ concatenated as markdown.
~1.8 MB. For one-shot ingestion by coding agents and
RAG pipelines.
Both files are also served from /docs/llms.txt and /docs/llms-full.txt
(Docusaurus serves website/static/ under baseUrl=/docs/). Some agents and
IDE plugins probe the classic site-root path; the deploy workflow now copies
both files to _site root so either URL works.
Conforms to the emerging llmstxt.org spec: H1 project name, blockquote
summary, short install command, GitHub link, then curated sections
mirroring the docs-site navigation (Getting Started, Using Hermes,
Features, Messaging, Integrations, Guides, Developer Guide, Reference).
Generated by website/scripts/generate-llms-txt.py. Wired into prebuild.mjs
so every 'npm run build' and 'npm run start' refreshes the files alongside
the existing skills.json extraction. Both outputs are gitignored (same
precedent as src/data/skills.json).
Descriptions in llms.txt are pulled from each page's frontmatter, so they
stay current automatically. All ~80 section slugs are validated against
the filesystem at generation time; an invalid slug would fail the prebuild.
Adds a proper feature page at user-guide/features/goals.md covering
the /goal slash command — Hermes' take on the Ralph loop shipped in
PR #18262. The slash-commands reference table had two table rows but
no narrative doc walking through the judge model, fail-open semantics,
turn budget, persistence, user-message preemption, or the aux-model
config override.
Adds a walkthrough example showing a multi-turn goal running to
completion, covers the two judge failure modes with how to recover,
and credits Codex CLI 0.128.0 / Eric Traut as prior art.
Also cross-links both slash-commands.md rows to the new page so
readers discovering /goal from the command reference can dive in.
The anyOf collapse in _repair_schema returned early, skipping the
nullable-strip and enum-cleanup steps. When a schema had anyOf
[{enum: [..., null, '']}, {type: null}] alongside a parent-level
'nullable: true', collapsing to the single non-null branch produced a
merged node that still had both 'nullable' and the bad enum values —
Moonshot would still 400 on it.
Fix: fall through to Rules 1/3 when the collapse produces a single
merged node; only return early for the multi-branch case (pure
anyOf preservation) or when there was no null branch to remove.
Adds a test that locks in the combined-case expectation.
When a schema node inside anyOf has enum values but no explicit 'type',
Rule 3 (enum cleanup) ran before _fill_missing_type, so node_type was
None and the enum was never cleaned. Moonshot then rejected the schema
with 'enum value (<nil>) does not match any type in [string]'.
Fix: reorder operations — fill missing type first, strip nullable,
then clean enum. This ensures enum cleanup always has a type to check.
Also fixes test expectation: empty string in enum is now correctly
stripped (Moonshot rejects it too).
Closes#16875
Add a standing-goal slash command that keeps Hermes working toward a
user-stated objective across turns until it is achieved, paused, or
the turn budget runs out. Our take on the Ralph loop — cf. Codex CLI
0.128.0's /goal.
After each turn, a lightweight auxiliary-model judge call asks 'is
this goal satisfied by the assistant's last response?'. If not, and
we're under the turn budget (default 20), Hermes feeds a continuation
prompt back into the same session as a normal user message. Any real
user message preempts the continuation loop automatically.
Judge failures fail OPEN (continue) so a flaky judge never wedges
progress — the turn budget is the real backstop.
### Commands
- `/goal <text>` — set a standing goal (kicks off the first turn)
- `/goal` or `/goal status` — show current state
- `/goal pause` — pause the continuation loop
- `/goal resume` — resume (resets turn counter)
- `/goal clear` — drop the goal
Works on both CLI and gateway platforms via the central CommandDef
registry.
### Design invariants preserved
- **Prompt cache**: continuation prompts are regular user-role
messages appended to history. No system-prompt mutation, no toolset
swap.
- **Role alternation**: continuation is a user turn, never injected
mid-tool-loop.
- **Session persistence**: goal state lives in SessionDB.state_meta
keyed by `goal:<session_id>`, so `/resume` picks it up.
- **Mid-run safety**: on the gateway, `/goal status|pause|clear` are
allowed mid-run (control-plane only); setting a new goal requires
`/stop` first so we don't race a second continuation prompt against
the current turn.
### Files
- `hermes_cli/goals.py` (new, 380 lines) — GoalManager + judge + state
- `hermes_cli/commands.py` — CommandDef entry
- `hermes_cli/config.py` — `goals.max_turns` default
- `hermes_cli/web_server.py` — dashboard category merge
- `cli.py` — /goal handler + post-turn continuation hook in
process_loop
- `gateway/run.py` — /goal handler + post-turn continuation hook
wrapping _handle_message_with_agent
- `tests/hermes_cli/test_goals.py` (new, 26 tests) — judge parsing,
fail-open semantics, lifecycle, persistence, budget exhaustion
- `website/docs/reference/slash-commands.md` — docs entry
* docs(sidebar): collapse exploding skills tree to a single Skills node
The Skills sub-tree in the left sidebar expanded to 200+ entries
(22 bundled categories + 15 optional categories, every skill a page).
That's most of the nav on a first visit — docs for the actual product
get drowned in it.
Collapse the sidebar to:
Skills
godmode (hand-written spotlight)
google-workspace (hand-written spotlight)
Bundled catalog (reference/skills-catalog — table of all bundled)
Optional catalog (reference/optional-skills-catalog — table of all optional)
Per-skill pages still generate and are still reachable at their URLs;
they're linked from the two catalog tables and from the Skills overview
page. They just don't appear in the left nav anymore.
sidebars.ts goes from 649 lines to 247. generate-skill-docs.py loses
the bundled/optional sidebar render helpers.
Also picks up incidental generator output drift on current main
(comfyui skill content refresh; 4 new skill pages for
devops-kanban-orchestrator, devops-kanban-worker,
productivity-here-now, productivity-shopify; two catalog refreshes).
These are what the generator produces on main today — keeping them
committed avoids the next docs build showing 'working tree dirty'.
* docs(sidebar): drop godmode and google-workspace spotlight pages
Keep the Skills sidebar node strictly principled: two catalog links,
nothing else. There was no rule for which skills got spotlight pages
and which got auto-generated pages — just that these two happened to
be hand-written first.
Both pages still build and are still reachable at
/docs/user-guide/skills/godmode and
/docs/user-guide/skills/google-workspace. They're linked from the
catalog tables and the Skills overview page.
Sidebar Skills node now:
Skills
├── Bundled catalog
└── Optional catalog
hermes update had two interactive [Y/n] prompts with no bypass:
1. Config migration (after new env/config options are added)
2. Autostash restore (when uncommitted work was stashed before pull)
hermes uninstall already has --yes/-y; mirrors that.
Under --yes:
- Config-migrate prompt → auto-yes, migrate_config(interactive=False)
so new config fields are applied but API-key prompts are skipped
(user runs 'hermes config migrate' later for those). Matches
gateway-mode semantics.
- Stash-restore prompt → auto-yes, git stash apply runs automatically.
Closes the 'can I hermes update -y, No ! Fix' gap reported by @murelux.
Adds opt-in auto-deletion for slash-command reply messages like
"New session started!", "Restarting gateway…", "Stopped.", and
YOLO toggles. After the TTL elapses the gateway calls the adapter's
delete_message; on platforms without a delete API (everything except
Telegram today) the TTL is silently ignored and the message stays.
Requested on Twitter by @charlesmcdowell — tool-call bubbles are useful
real-time, but system notices clutter the thread once the agent finishes.
Implementation:
- EphemeralReply(str) sentinel in gateway/platforms/base.py. Subclasses
str so existing 'X' in response / response.startswith(...) checks in
tests and call sites keep working unchanged; isinstance() still
distinguishes it for the send path.
- _process_message_background and both busy-session bypass paths
(in base.py) call _unwrap_ephemeral() on the handler return, send
the unwrapped text, and schedule a detached delete task when the
TTL > 0 AND the adapter class overrides delete_message.
- display.ephemeral_system_ttl (default 0 = disabled) in DEFAULT_CONFIG.
Handler can pass ttl_seconds explicitly to override.
- Wrapped the highest-noise return sites: /new, /reset, /stop,
/yolo on/off, /restart success + "already in progress". Draining
notices and /help output left as plain strings — those are
informational and users want to read them.
Backward-compat: default TTL 0 → no scheduling, no behavior change
for existing users. Platforms without delete_message silently no-op.
When the curator consolidates skill X into umbrella Y, any cron job
that listed X in its skills field would fail to load X at run time —
the scheduler logs a warning and skips it, so the scheduled job runs
without the instructions it was scheduled to follow.
cron.jobs.rewrite_skill_refs(consolidated, pruned) now updates jobs
in-place: consolidated names route to the umbrella target (dedup
when umbrella is already present), pruned names are dropped.
agent.curator._write_run_report calls it after classification,
best-effort so a cron-side failure never breaks the curator itself.
Results are recorded in run.json (counts.cron_jobs_rewritten + full
cron_rewrites payload), a separate cron_rewrites.json for convenience
when jobs were touched, and a section in REPORT.md.
Reported by @tombielecki.
DeepSeek V4 Pro tightened thinking-mode validation and rejects empty-string
reasoning_content with HTTP 400:
The reasoning content in the thinking mode must be passed back to the API.
run_agent.py injected "" at three fallback sites — the tool-call pad in
_build_assistant_message and both injection branches of
_copy_reasoning_content_for_api (cross-provider poison guard + unconditional
thinking pad). All three now emit " " (single space), which satisfies the
non-empty check on V4 Pro without leaking fabricated reasoning.
Also upgrades stale empty-string placeholders on replay: sessions persisted
before this change have reasoning_content="" pinned at creation time; when
the active provider enforces thinking-mode echo, the replay path now rewrites
"" -> " " so existing users don't 400 on their first V4 Pro turn after
updating. Non-thinking providers still round-trip "" verbatim.
Updates 9 existing assertions + adds 2 regression tests (stale-placeholder
upgrade, non-thinking verbatim preservation).
Refs #15250, #17400.
Closes#17341.
The user-visible /compress banner and the post-compression last_prompt_tokens
writeback both counted only the raw message transcript (chars/4). With a 15KB
system prompt and 30 tool schemas (~26KB), a 4-message transcript that looks
like ~45 tokens to the transcript-only estimator is really ~10.5K tokens of
request pressure — a 234x gap.
Two user-facing consequences:
- Banner shows 'Compressing … (~45 tokens)…' while compression is actually
firing on 10K+ tokens of real pressure, confusing users about why
compression triggered (reported by @codecovenant on X; #6217).
- Post-compression last_prompt_tokens writeback omits tool schemas, so the
next should_compress() check compares real usage against a stale
underestimate — compression triggers late, potentially past the model's
context limit on small-context models (#14695).
Swap estimate_messages_tokens_rough() for estimate_request_tokens_rough()
at every user-visible banner and at the post-compression writeback.
estimate_request_tokens_rough() already existed for exactly this purpose
and includes system prompt + tool schemas.
Touched call sites:
- run_agent.py: post-compression last_prompt_tokens writeback, post-tool
call should_compress() fallback when provider usage is missing
- cli.py: /compress banner + summary
- gateway/run.py: gateway /compress banner + summary
- tui_gateway/server.py: TUI /compress status + summary
- acp_adapter/server.py: ACP /compact before/after
Left intentionally alone:
- Session-hygiene fallback and the 'no agent' /status path in gateway/run.py
— no agent instance is in scope to query for system prompt/tools, and the
existing 30-50% overestimate wobble on hygiene is safety-accepted.
- Verbose-mode 'Request size' logging — informational only, already counts
system prompt via api_messages[0].
Also relabels the feedback line from 'Rough transcript estimate' to
'Approx request size' so the metric label matches what it actually measures.
Credits: diagnoses from @devilardis (#14695) and @Jackten (#6217);
user report @codecovenant on X (2026-04-30).
Closes#14695Closes#6217
When a user types /steer <text> on an ACP session that isn't actively
running a turn (and there's no interrupted-prompt salvage available),
_cmd_steer silently appended to state.queued_prompts and replied
"No active turn — queued for the next turn". That looks identical to
/queue output even though the user never typed /queue — @EddyLeeKhane
reported this as "/steer never works, gets queued instead".
Rewrite the payload to a plain user prompt before the slash-intercept
fires, matching the gateway's idle-/steer fallthrough in
gateway/run.py ~L4898.
`hermes update` ran the config migration (11 → 17) successfully then
crashed at `agent/skill_utils.py:340` during the post-migration
skill-config prompt. User @FlockonUS reported this on Twitter.
Root cause: `get_missing_skill_config_vars` in hermes_cli/config.py
only guarded the import of `discover_all_skill_config_vars`, not the
call. Any runtime exception inside the skill scan (malformed SKILL.md,
unreadable external skill dir, etc.) propagated up through
`migrate_config` and aborted `hermes update` after the version bump.
Wrap the call in try/except so skill-config prompting — which is a
post-migration nicety — can never block the migration itself.
The initial guardrail PR consolidated failure classification by pointing
display._detect_tool_failure at the new classify_tool_failure helper,
which was strictly broader: it flagged any JSON result with
"success": false / "failed": true / non-empty "error", plus plain-text
"traceback" and "error:" prefixes. That would uptick the user-visible
[error] tag on tools that return {"success": false} as a benign signal
(memory fullness, todo state, etc.) and feed the failure-streak counter
at the same time.
Restore display._detect_tool_failure to its pre-PR semantics verbatim.
Tighten classify_tool_failure (the guardrail's internal safety-fallback
used only when callers don't pass failed=) to match _detect_tool_failure
exactly, so the two never disagree. Production callers in run_agent.py
already pass an explicit failed= derived from _detect_tool_failure, so
the guardrail counter is driven by the same signal the CLI shows.
- Emit providers in CANONICAL_PROVIDERS order (matching hermes model)
with user-defined/custom providers appended after
- Remove digit quick-select (1-9,0) handler — inconsistent with
absolute row numbering and already removed from hint text
- Remove unused windowOffset import
_process_message_background snapshotted callback_generation from the
interrupt event at the TOP of the task — before the handler ran.
_hermes_run_generation is only set on the event by
GatewayRunner._bind_adapter_run_generation during
_handle_message_with_agent, which runs DURING the handler await. The
early snapshot always captured None, which then flowed into
pop_post_delivery_callback(..., generation=None) in the finally block.
In pop_post_delivery_callback, generation=None with a tuple-registered
entry (generation, callback) bypasses the ownership check — it pops and
fires the callback regardless of which run owns it. Result: a stale run
could fire a fresher run's post-delivery callback (e.g. a
background-review notification attributed to the wrong turn).
Fix: move the snapshot into the finally block, after the handler has
run and _hermes_run_generation has been bound to the current run.
Regression test added: simulates a stale handler at generation=1 and a
fresher callback registered at generation=2. Pre-fix: snapshot=None →
pop fires the generation=2 callback under generation=1's ownership
("newer" fires). Post-fix: snapshot=1 → pop skips the mismatched
entry, callback stays in the dict for the correct run to claim.
Verified: test FAILS on current main (captures "newer" in fired list),
PASSES with this fix.
Salvaged from PR #12565 (the callback-ownership portion only; the
/status totals portion was already fixed on main in 7abc9ce4d via #17158).
Co-authored-by: Oxidane-bot <1317078257maroon@gmail.com>
Widens #16528 to two sibling sites that had the same quoted-boolean
bug: a YAML string "false" (or "0", "no", "off") silently evaluated
truthy under bool() / if-check.
- gateway/run.py _load_show_reasoning: is_truthy_value wrap
- tools/skill_manager_tool.py _guard_agent_created_enabled: is_truthy_value wrap
- regression tests for both
SELECT in get_messages_as_conversation() was missing finish_reason, so
assistant messages round-tripped through replay (including /branch copies)
silently dropped the provider's stop signal. Adds it to the SELECT, restores
it on assistant rows, and locks it in with a round-trip test.
When running on a host with sudoers NOPASSWD configured for the current
user, interactive Hermes sessions were unnecessarily entering the
password prompt path before executing sudo commands. Outside Hermes,
`sudo -n true` exits 0 for that user.
Add `_sudo_nopasswd_works()` that probes `sudo -n true` and, when it
succeeds, lets `_transform_sudo_command()` return the command unchanged
with no stdin password. The probe:
- Is scoped to the `local` terminal backend only, so Docker/SSH/Modal
and other remote backends do not inherit host sudo state.
- Re-probes every call (no process-lifetime cache) so an expired sudo
timestamp cannot silently make a later command block waiting for a
password that Hermes never prompts for.
- Is bypassed entirely when `SUDO_PASSWORD` is configured or a cached
password already exists, preserving existing explicit-password flows.
Co-authored-by: Junting Wu <juntingpublic@gmail.com>
The fix for this bug (isinstance guard) was merged via commit 3ff9e010,
but test coverage was not included. Adding 4 tests:
- dict metadata with hermes keys (normal case)
- string metadata (bug case — previously caused AttributeError)
- None metadata
- missing metadata key
Proves token A's detected capabilities do not leak to token B after the
fix in the preceding commit. Before the fix this test would have seen
both tokens return token A's cached value.
_capability_cache was a single module-level dict shared across all
tokens. If the bot token rotates or multiple tokens are used in one
process, capabilities detected for token A would be returned for
token B, causing wrong schema gating and incorrect runtime behavior.
Replace the single Optional cache with a Dict keyed by token so each
token gets its own isolated capability entry.
_SupervisorRegistry.get_or_start() returned an existing supervisor
whenever the cdp_url matched, without checking if the supervisor's
thread or event loop was still alive. A crashed supervisor would be
silently reused, causing missed dialog/frame updates.
Now checks both _thread.is_alive() and _loop.is_running() before
returning the cached instance. An unhealthy supervisor is torn down
and recreated, matching the existing URL-changed code path.
_get_peer() and _get_or_create_honcho_session() accessed _peers_cache
and _sessions_cache without holding _cache_lock, while other paths
in the same class use the lock consistently. Under concurrent tool
calls or prefetch threads, this can produce stale reads or lost
cache updates.
Wrap both unguarded cache read sites in _cache_lock. Network calls
(honcho.peer() and honcho.session()) remain outside the lock to
avoid holding it during I/O.
Three int() calls in HonchoClient.from_global_config() parsed
dialecticMaxChars, messageMaxChars, and dialecticMaxInputChars
directly without guards. A malformed value in honcho.json would
raise ValueError and abort provider initialization entirely.
Add _parse_int_config() helper following the existing
_parse_context_tokens() pattern, and replace all three raw
int() calls with it.
Add two operator-facing toggles for inbound Feishu admission, enabling
bot-to-bot scenarios such as A2A orchestration and inter-bot
notifications:
FEISHU_ALLOW_BOTS=none|mentions|all (default: none)
Accept messages from other bots. `mentions` requires the peer
bot to @-mention Hermes; `all` admits every peer-bot message.
FEISHU_REQUIRE_MENTION=true|false (default: true)
Whether group messages must @-mention the bot. Override per-chat
via `group_rules.<chat_id>.require_mention` in config.yaml.
Defaults preserve prior behavior. Self-echo protection is always on:
when the bot's identity is unresolved (auto-detection failed and
FEISHU_BOT_OPEN_ID unset), peer-bot messages are rejected fail-closed
to avoid feedback loops.
Admitted peer bots bypass the human-user allowlist
(FEISHU_ALLOWED_USERS) to match existing Discord behavior; humans
still need an explicit allowlist entry. yaml feishu.allow_bots is
bridged to the env var so the adapter and gateway auth layer share
one source of truth.
Resolving peer-bot display names requires the
application:bot.basic_info:read scope; without it, peers still route
but appear as their open_id.
Test: tests/gateway/test_feishu_bot_admission.py covers the admission
pipeline, group-policy bot-bypass, hydration, and event-dispatch
plumbing as a parametrized matrix.
Change-Id: I363cccb578c2a5c8b8bf0f0a890c01c89909e256
reset_session() creates a fresh SessionEntry with created_at == updated_at,
but get_or_create_session() bumps updated_at on the next inbound message,
causing _is_new_session in _handle_message_with_agent to evaluate False.
The topic/channel skill auto-load gate (group_topics, channel_skill_bindings)
silently skips the first message after a manual reset.
Add an is_fresh_reset flag on SessionEntry, set by reset_session() and
consumed once by the message handler. Kept distinct from was_auto_reset
because that flag also drives a 'session expired due to inactivity'
user-facing notice and a context-note prepend — both wrong for an
explicit /new or /reset.
Persisted through to_dict/from_dict so the flag survives gateway
restart between /reset and the next message.
Fixes#6508
Co-authored-by: warabe1122 <45554392+warabe1122@users.noreply.github.com>
Co-authored-by: willy-scr <187001140+willy-scr@users.noreply.github.com>
/status was reading session_entry.total_tokens from the in-memory
SessionStore (gateway/session.py), which the agent never writes to —
so the token count was always 0.
The agent already persists token deltas to the SQLite SessionDB
(run_agent.py:11497) for every platform with a session_id. Route
/status through that single source of truth instead of duplicating
token writes into a second store.
Fix:
- gateway/run.py: _handle_status_command now calls
self._session_db.get_session(session_id) and sums the five token
component columns (input/output/cache_read/cache_write/reasoning).
Falls back to 0 when no SessionDB is configured or no row exists.
- Two new regression tests covering the populated-row and
missing-row paths.
Co-authored-by: Hermes <127238744+teknium1@users.noreply.github.com>
Minor follow-up to the native-image-buffer isolation fix. The write site
in _prepare_inbound_message_text was calling build_session_key directly,
while every other call site in gateway/run.py uses the _session_key_for_source
helper — which consults session_store._generate_session_key first and falls
back to build_session_key. Keeping the write key and consume key on the
same helper prevents key drift if the session store ever overrides the
default keying behavior.
_SLASH_WORKER_TIMEOUT_S and _pool used raw float()/int() on env vars
at module level. A non-numeric value (e.g. HERMES_TUI_SLASH_TIMEOUT_S=abc)
raises ValueError during import, preventing TUI gateway from starting
with no useful error message.
Wrap both parses in try/except with safe fallbacks:
- HERMES_TUI_SLASH_TIMEOUT_S: fallback to 45.0s
- HERMES_TUI_RPC_POOL_WORKERS: fallback to 4 workers
sqlite3 can only bind str/bytes/int/float/None to query parameters.
Multimodal message content is a list of parts (text + image_url), which
raised 'Error binding parameter 3: type list is not supported' in
append_message and replace_messages.
In the CLI/TUI this surfaced as a visible crash when users pasted
screenshots. In the gateway it was silently swallowed by a bare except
in append_to_transcript, causing multimodal turns to be lost from the
session transcript.
Fix at the DB layer: _encode_content wraps lists/dicts as
'\\x00json:' + json.dumps(...) on write, _decode_content unwraps on
read. Plain strings are untouched, so existing FTS search, previews,
and JSONL compat are unaffected. Paired decode in get_messages,
get_messages_as_conversation, and search_messages context previews.
Regression test covers: list content round-trip, dict content
round-trip, string content stored unchanged, replace_messages with
multimodal content.
Also included: aligned fix#17522 for TUI image attachment with
paths containing spaces (see previous commit).
Remove frontend regex pre-check that truncated paths containing spaces,
quotes, or Windows drive letters. Backend _detect_file_drop correctly
handles these patterns. This fixes image attachment for common filenames
like "Screenshot 2026-04-29.png".
Add tests:
- test_input_detect_drop_path_with_spaces: attaches image with spaces in name
- test_input_detect_drop_path_with_spaces_and_remainder: remainder handling
Also restored missing in test_rollback_restore_resolves_number_and_file_path.
Scope: tui, vision, tests
Widens the cherry-picked fix from @jatingodnani (#17343) to the
gateway path. On main, user_config.agent.disabled_toolsets was only
honored by _get_platform_tools' name-level subtraction — it did not
catch tools pulled in implicitly by a composite toolset (browser
includes web_search, hermes-* platforms include most tools).
Changes:
- gateway/run.py: resolve disabled_toolsets alongside enabled_toolsets
and pass to AIAgent at both user-facing construction sites (normal
message loop + single-turn cron-like path). Hygiene/compression
agents (fixed enabled_toolsets=[memory]) are intentionally untouched.
- gateway/run.py: add (agent, disabled_toolsets) to
_CACHE_BUSTING_CONFIG_KEYS so editing the list in config.yaml
invalidates the cached AIAgent on the next message.
- cli.py: drop unused 'import platform' left over from PR #17343's
import churn; restore 'import sys' used throughout the file.
- model_tools.py: drop unused 'import os, sys' added by PR #17343;
fix comment reference from #15291 (unrelated OAuth issue) to #17309.
Co-authored-by: jatin godnani <godnanijatin@gmail.com>
Refactor tool resolution logic in model_tools.py to ensure that
disabled_toolsets are always subtracted at the end, preventing
composite toolsets (e.g. 'browser') from implicitly enabling tools
that should be hidden.
- Added 'disabled_toolsets' to DEFAULT_CONFIG in hermes_cli/config.py
- Updated HermesCLI in cli.py to load and propagate disabled toolsets to AIAgent
- Implemented robust two-phase resolution (additive then subtractive) in model_tools.py
Themes previously embedded layout-affecting values (baseSize, lineHeight,
density, letterSpacing) alongside visual identity properties, coupling
user ergonomic preferences to color theme selection.
This change establishes a clear separation of concerns:
- Themes own: palette, font family, border-radius, and font-coupled
letterSpacing (e.g. Inter's -0.005em tracking)
- Layout scale (baseSize, lineHeight, density) is standardized via
DEFAULT_TYPOGRAPHY and DEFAULT_LAYOUT — not overridden per theme
All themes now spread DEFAULT_TYPOGRAPHY and DEFAULT_LAYOUT as their
base, removing silent divergence and making future layout settings
(e.g. user-configurable density) trivially applicable across all themes
without per-theme special-casing.
All built-in themes now spread DEFAULT_TYPOGRAPHY, removing independent
baseSize overrides and converging on 15px. All themes also use
density: comfortable, removing the compact/spacious divergence that
caused item-count shifts on fixed-height pages (e.g. Skills).
Two additional per-theme overrides are also normalized:
- rose: lineHeight: "1.7" removed — was paired with density: spacious
for an airy feel; once density was normalised the elevated line-height
became an orphaned artefact causing nav item height drift.
- cyberpunk: letterSpacing changed from "0.02em" to "0" — extra tracking
on top of an already-wide monospace font caused text to wrap earlier
than in other themes.
Switching themes is now a purely cosmetic change — color palette,
font family, border-radius, and typographic style differ; font size,
spacing, line-height, and letter-spacing do not.
- Move the disabled-ack guard above the debounce so we don't stamp
_busy_ack_ts[session_key] when no ack was actually sent. Harmless
(never read when disabled) but cosmetically off.
- Document display.busy_ack_enabled in user-guide/messaging/index.md
and HERMES_GATEWAY_BUSY_ACK_ENABLED in reference/environment-variables.md.
- Add JezzaHehn to scripts/release.py AUTHOR_MAP for contributor credit.
Follow-up to #17491 (Jezza Hehn).
When a user sends a message while the gateway is busy processing,
an acknowledgment message is sent. This can be spammy for users
who send rapid messages.
Add display.busy_ack_enabled config option (default: true) to allow
users to suppress these busy-input acknowledgment messages.
Fixes#17457
When a user defines `custom_providers: [{name: kimi, ...}]` and references
`provider: kimi` from fallback_model or the main config, the built-in alias
rewriting (`kimi` → `kimi-coding`) was hijacking the request before the
named-custom lookup ran. `_get_named_custom_provider` also refused to
return a match when the raw name resolved to any built-in (including aliases),
so the custom endpoint was unreachable.
Fix at both layers of the resolution chain so every caller benefits, not
just `_try_activate_fallback`:
- hermes_cli/runtime_provider.py: narrow `_get_named_custom_provider`'s
built-in-wins guard to canonical provider names only. An alias like
`kimi` that resolves to a different canonical (`kimi-coding`) no longer
blocks the custom lookup; a canonical name like `nous` still does.
- agent/auxiliary_client.py: in `resolve_provider_client`, try the named-
custom lookup with the original (pre-alias-normalization) name before the
alias-normalized one, so aliased requests reach the user's custom entry.
Also honour `explicit_base_url` and `explicit_api_key` in the API-key
provider branch so callers that pass explicit hints (e.g. fallback
activation) can override the registered defaults.
Tests added for:
- custom `kimi` shadowing built-in alias (regression for #15743)
- custom `nous` NOT shadowing canonical built-in (behaviour preserved)
- bare `kimi` without any custom entry still routing to built-in
- explicit base_url/api_key override on the API-key provider branch
Original PR #17827 by @Feranmi10 identified the same bug class and
implemented a narrower fix in `_try_activate_fallback`; this reshapes the
fix to live in the shared resolution layer so all callers benefit.
Fixes#15743
Co-authored-by: Feranmi10 <89228157+Feranmi10@users.noreply.github.com>
Follow-up to the previous commit. Replace the post-fetch Python re-sort (which
required dropping LIMIT/OFFSET from SQL and scanning every session row) with a
recursive CTE that walks compression-continuation chains and computes
effective_last_active per root at SQL level. The outer query can then ORDER BY
+ LIMIT efficiently, and the Python projection loop no longer has to handle
ordering.
This preserves the correctness win (old compression roots whose live tip was
touched recently surface correctly) without the O(N) scan, which matters for
users with thousands of sessions.
Adds a regression test pinning the compression-tip case at limit=1 — the
stress case that any bounded-oversample shortcut would get wrong.
Co-authored-by: simbam99 <simbamax99@gmail.com>
- order session_search recent-mode results by last activity instead of session start time
- add an opt-in `order_by_last_active` path to `SessionDB.list_sessions_rich`
- add regression coverage for both the database ordering and recent-mode call path
- Reset keySaving on back() to prevent blocked key entry after Esc
- Show '(needs setup)' for non-API-key auth providers instead of
generic '(no key)'
- Set is_current correctly for unauthenticated providers that happen
to be the active session provider
- Guard model.save_key with is_managed() check — return error on
managed installs where .env is read-only
- New model.disconnect RPC method: clears API key env vars from .env
and OAuth/credential pool state via clear_provider_auth()
- Press 'd' on an authenticated provider opens confirmation prompt
- y/Enter confirms disconnect, n/Esc cancels
- Provider flips to unauthenticated state in-place (re-selectable
to re-auth by pressing Enter again)
- model.options now returns all canonical providers (not just
authenticated), each with authenticated/auth_type/key_env fields
- New model.save_key RPC method: saves API key to .env, sets in
process, returns refreshed provider with models
- Picker shows ● (authed) / ○ (no key) markers with dimmed styling
- Selecting an unauthenticated api_key provider opens inline masked
key input — after save, transitions directly to model selection
- Non-api_key auth providers show guidance to run hermes model
- Row numbers now show absolute position in list
The model picker displayed row numbers 1-12 regardless of scroll
position, making it impossible to tell where you were in the list.
Now shows the actual item index (e.g. 5, 6, 7... when scrolled down).
Also removed '1-9,0 quick' from the hint text since digit shortcuts
still work relative to the visible window, which would be confusing
with absolute numbering.
The TUI's _apply_model_switch() was converting the config.yaml
`providers:` dict into a list of dicts before passing it to
switch_model(). This caused resolve_provider_full() →
resolve_user_provider() to fail, since that function expects a dict
and does `user_config.get(name)` to look up provider entries.
The result: user-defined providers (e.g. ollama) appeared in CLI's
/model picker but were invisible in the TUI.
Fix:
- tui_gateway/server.py: pass cfg.get('providers') directly (dict),
matching what cli.py already does at line 5598.
- hermes_cli/model_switch.py: fix the validation-override block
(line ~893) which iterated user_providers as a list — now correctly
handles the dict format with support for both dict-keyed and
list-format models arrays.
The PR wired in a detached watcher that respawns manual profile gateways
after they exit. Pair that with a SIGUSR1 graceful drain (same path
systemd/launchd use) so in-flight agent runs finish instead of getting
SIGTERM'd. Fall back to SIGTERM if SIGUSR1 isn't wired or the gateway
doesn't exit within the drain budget — the watcher sees the exit and
relaunches either way.
Tested end-to-end against an orphaned gateway: graceful drain exits in
0.5s and the watcher fires the relaunch command.
When len(messages) <= protect_tail_count and a token budget is set, the
previous formula min(protect_tail_count, len(result) - 1) under-protected
the tail by one, allowing the oldest message to be summarized.
The test fails on the buggy formula (pruned == 1) and passes on the fix
(pruned == 0, tool content preserved verbatim).
Widen PR #17842's atomic-write fix to two sibling sites that exhibit the
same 'partial JSON on interrupted write' class of bug:
- gateway/platforms/feishu.py: dedup state (_dedup_state_path)
- gateway/platforms/helpers.py: ParticipatedThreadTracker save
Both are small recovery/coordination files that get rewritten frequently and
break cross-restart dedup if left partial.
Follow-up to #17963. The threaded branch of resolve_plugin_command_result
previously called Event.wait() with no timeout — a hung async plugin
handler would wedge the terminal indefinitely. Cap the wait at 30s and
raise TimeoutError instead. Added a regression test covering the hung
handler path.
Moves the here-now skill under optional-skills/productivity/here-now/ so
it's discoverable via the Skills Hub but not installed by default, and
tightens the SKILL.md description to a single line to match sibling
optional-skill descriptions.
Install with:
hermes skills install official/productivity/here-now
Closes#378
Add the here.now productivity skill with a bundled publish runtime so Hermes can publish files and folders to live URLs. Keep the skill thin and docs-first while fixing script path resolution and upload failure handling.
Made-with: Cursor
Closes#16082
The `hermes status` command listed provider API keys under the
◆ API Keys section but NVIDIA_API_KEY was absent. Users configured
with NVIDIA NIM had no way to verify their key was set from status
output. Add it alongside the other inference provider keys.
The switch_model override logic incorrectly iterated over user_providers
as if it were a list of dicts, but it's actually a dict mapping
provider_slug -> config. This meant private models defined in a provider's
`models:` section (e.g. nahcrof-dedicated with discover_models: false)
were never accepted when the API /models list didn't include them.
Fix: iterate over user_providers.items(), match by slug, and handle both
dict and list forms of the models config.
- Keep @preview tag for teams CLI
- Step 3: note client secret won't be shown again
- Step 6: use the Install in Teams link from teams app create output
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- Keep @preview tag for teams CLI
- Step 3: note client secret won't be shown again
- Step 6: just open the Install in Teams link from teams app create output
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- Keep @preview tag for teams CLI
- Step 3: note client secret won't be shown again
- Step 6: use the install link printed by teams app create
instead of a separate CLI command
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Was hardcoded to 3978; use ${TEAMS_PORT:-3978} so a custom port
set in .env is actually passed into the container.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
microsoft-teams-apps 2.0.0 added the `client` option to AppOptions,
accepting a ClientOptions instance. Use it to set the User-Agent
header to "Hermes" on all outgoing HTTP requests.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
The kanban PR (#17805, c86842546) added the `kanban` toolset and
`tools/kanban_tools.py`, but didn't update three pre-existing test
assertions that bake the full toolset/tool inventory:
* `tests/tools/test_registry.py::test_matches_previous_manual_builtin_tool_set`
hard-codes the manual list of builtin tool modules. `tools.kanban_tools`
was missing.
* `tests/test_tui_gateway_server.py::test_load_enabled_toolsets_rejects_disabled_mcp_env`
and `test_load_enabled_toolsets_falls_back_when_tui_env_invalid` both
expect `["memory"]` from `_load_enabled_toolsets()`. With kanban now
auto-recovered by `_get_platform_tools` (its tools live in hermes-cli's
universe but are not in CONFIGURABLE_TOOLSETS), the resolver returns
`["kanban", "memory"]`.
* `tests/hermes_cli/test_tools_config.py::test_get_platform_tools_preserves_explicit_empty_selection`
asserts `set()` for an explicit empty list. The recovery loop now also
surfaces `kanban`. Reframed to assert the contract the test name
describes — no CONFIGURABLE toolset gets re-enabled when the user
explicitly saved an empty list — which stays correct as more
non-configurable platform toolsets are added.
Verified the failures reproduce on clean origin/main (180a7036b) with
`.[all,dev]`-equivalent extras (fastapi, starlette, httpx, pytest-asyncio)
and that all four pass with this commit applied. CI on main itself is
currently red on these tests; this restores green for everyone's PRs.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Signal-cli sends dataMessage wrappers for profile key updates and other
metadata events that have no actual text content. These were reaching the
gateway as msg='' and triggering full agent turns for nothing.
Add early return in _handle_envelope() when both message field is empty/
missing/whitespace AND there are no attachments. Messages with media
attachments but no text still flow through.
- 12 lines added to gateway/platforms/signal.py
- 5 new tests in TestSignalContentlessEnvelope class
It was sitting at position 4 of the `hermes model` list, ahead of Anthropic,
OpenAI, Xiaomi, and other first-class API providers. Move it to the end of
CANONICAL_PROVIDERS and drop the "(200+ models, $5 free credit, no markup)"
parenthetical so the entry just reads "Vercel AI Gateway".
- New config key: dashboard.hidden_plugins (list of plugin names)
- GET /api/dashboard/plugins now filters out hidden plugins from sidebar
- POST /api/dashboard/plugins/{name}/visibility toggles visibility
- Hub response includes user_hidden boolean per plugin row
- Eye/EyeOff toggle on plugin cards with dashboard manifests
- i18n: 'Show in sidebar' / 'Hide from sidebar' (en/zh)
Use usePageHeader().setEnd to place the rescan button in the shared
header bar. Remove the inline H2 title (already shown by the header)
and the wrapper div.
- Add _validate_plugin_name() guard on all {name} path param endpoints
(rejects /, \, .. before reaching plugin logic)
- Strip after_install_path from install response (no internal paths to client)
- Update nix/tui.nix lockfile hash to match committed package-lock.json
- New PluginsPage.tsx: full plugin management UI (list, enable/disable,
install from git, remove, git pull updates, provider picker)
- Backend: dashboard_set_agent_plugin_enabled now also toggles the
plugin's toolset in platform_toolsets so enabling actually makes
tools visible in agent sessions
- Backend: /api/dashboard/plugins/hub returns auth_required + auth_command
per plugin (checks tool registry check_fn)
- Frontend: auth_required shown as Badge + CommandBlock with copy-able
auth command
- Fix: Select overflow in providers card (min-w-0 grid cells, removed
truncate/overflow-hidden that clipped dropdown)
- Refactor: _install_plugin_core extracted for non-interactive reuse,
PluginOperationError for structured error handling
- i18n: en/zh/types updated with all new plugin page strings
Adds optional-skills/productivity/shopify — curl-based guide for the
Shopify Admin GraphQL API (products, orders, customers, inventory,
metafields, bulk operations, webhooks) and the Storefront GraphQL API.
- API version 2026-01 (current stable)
- Custom-app access tokens (shpat_...) with X-Shopify-Access-Token header
- Notes the 2026-01-01 deprecation of admin-created custom apps, points
users at Dev Dashboard for new setups after that date
- Includes a reusable shop_gql() bash helper, cursor pagination,
rate-limit cost inspection, GID conventions, userErrors check
- Safety section warns on destructive mutations (delete/refund/cancel)
Installs cleanly via: hermes skills install official/productivity/shopify
The Ink TUI (\`hermes --tui\` + dashboard \`/chat\`) had no wiring for the
background self-improvement review. When the review fired and patched
a skill or saved a memory entry, the change landed but the user had
no visual indication it happened — only the CLI had a print surface
for the '💾 Self-improvement review: …' line.
Changes:
- tui_gateway/server.py: in _init_session, attach
agent.background_review_callback to an _emit('review.summary',
sid, {text}) closure. Wrapped in try/except so agents with locked
attribute slots don't break session startup.
- ui-tui/src/app/createGatewayEventHandler.ts: handle 'review.summary'
by routing ev.payload.text through sys(…), matching the existing
'background.complete' pattern. Empty / whitespace payloads are
ignored so the transcript never gets a blank system line.
- ui-tui/src/gatewayTypes.ts: extend the GatewayEvent discriminated
union with { type: 'review.summary', payload?: { text?: string } }.
Gateway platforms (Telegram, Discord, Slack, …) already route the
review summary via background_review_callback → post-delivery queue
in gateway/run.py, so they pick up the new 'Self-improvement review:'
prefix from the companion run_agent change with no platform edits.
Tests:
- tests/tui_gateway/test_review_summary_callback.py (Python, 2 tests):
_init_session attaches a callback that emits the right event; the
callback path survives agents that can't accept the attribute.
- ui-tui/src/__tests__/createGatewayEventHandler.test.ts (vitest, 2
new cases): review.summary events feed sys(...) with the full text;
empty / missing payloads are no-ops.
- TypeScript type-check passes.
- tui_gateway suite: 64/64 pass.
When the self-improvement background review fires after a turn, it runs
in a bg thread and emits a ' 💾 <summary>' line to announce what it
saved to memory or skills. Two problems made this invisible to users
even when the review successfully modified a skill:
1. The print went through `_cprint` (prompt_toolkit's print_formatted_text)
on a bg thread while the CLI's PromptSession was live. Direct
print_formatted_text races with the input-area redraw and the line
can land behind/above the prompt, scrolled off without the user
seeing it.
2. The message said only '💾 Skill created.' / '💾 Memory updated'
with no indication that the self-improvement loop was the one doing
this. Users who did catch the line couldn't tell the background
review from some other agent action.
Fixes:
- `_cprint` now detects when it's called from a non-app thread with a
running prompt_toolkit Application, and routes through
`run_in_terminal` via `loop.call_soon_threadsafe`. That pauses the
input, prints the line above the prompt, and redraws — the normal
prompt_toolkit contract for bg-thread output. Direct-print fallback
preserved for the no-app / same-thread / import-error paths. Affects
every bg-thread emission, not just the review summary (curator
summaries and auxiliary failure prints benefit too).
- The summary now reads ' 💾 Self-improvement review: <summary>' in
both the CLI and the gateway `background_review_callback` path, so
the origin is unambiguous.
Tests:
- New `tests/cli/test_cprint_bg_thread.py` covers all five routing
branches (no app, app-not-running, cross-thread schedule, same-thread
direct, app-loop-attribute-error, import-error).
- New case in `tests/run_agent/test_background_review.py` asserts the
attributed prefix shows up in both `_safe_print` and
`background_review_callback`.
Live E2E: exercised _cprint from a bg thread inside a real Application
event loop; confirmed get_app_or_none() sees the app, call_soon_threadsafe
schedules run_in_terminal, and the inner _pt_print runs.
Builds on #16855 (@lsdsjy) which fixed DeepSeek v4 reasoning_content
replay via model_extra fallback + capturing tool_calls at method entry.
Kimi / Moonshot thinking mode enforces the same echo-back contract and
hits the same 400 when a tool-call turn is persisted without
reasoning_content.
- _build_assistant_message: pad branch now uses _needs_thinking_reasoning_pad()
(DeepSeek OR Kimi) instead of _needs_deepseek_tool_reasoning() alone.
- Extract _needs_thinking_reasoning_pad() and reuse it in
_copy_reasoning_content_for_api so both sites share one predicate.
- tests/run_agent/test_deepseek_reasoning_content_echo.py: add
TestBuildAssistantMessagePadsStrictProviders parametrized over DeepSeek
(attr=None, attr-absent), Kimi (attr=None), Moonshot (via base_url),
and an OpenRouter negative control that must NOT pad. Proven to fail
2/5 cases on Kimi/Moonshot without this change.
- scripts/release.py: add AUTHOR_MAP entries for lsdsjy and season179.
Refs #17400.
Co-authored-by: season179 <season.saw@gmail.com>
Alongside the existing 'least recently used' section, surface two more
rankings so users can see which of their agent-created skills actually
get exercised:
- 'most used (top 5)' — sorted by use_count descending. Hidden when every
skill has use_count=0 (noise suppression on fresh installs).
- 'least used (top 5)' — sorted by use_count ascending. Always shown
when the catalog is non-empty.
use_count started tracking real agent skill activation in PR #17932
(bump_use wired into skill_view tool + slash invocation + --skill
preload), so these rankings are now meaningful.
Tests: 3 new in tests/hermes_cli/test_curator_status.py — happy path
with mixed use_counts, zero-use suppression of the most-used section,
and the no-skills clean-empty case.
Treat skill views and edits as activity when curator reports and applies lifecycle transitions, so recently loaded or patched skills are not displayed or transitioned as never used.\n\nAdds regression tests for activity derivation, automatic transitions, and CLI status output.
restore_skill() in tools/skill_usage.py used archive_root.iterdir(), which
only walked the top level of .archive/. Skills archived under nested layouts
(e.g. .archive/openclaw-imports/<skill>/ from older archive paths or
external imports) were invisible to both the exact-match and prefix-match
candidate scans, surfacing as a misleading "skill '<name>' not found in
archive" error even though the directory existed on disk.
Switch both candidate scans to archive_root.rglob('*') so the lookup
descends into category subdirectories.
Fixes#17942
* fix(curator): split 'archived' into consolidated vs pruned in run reports
Users who watched a curator run saw skills like 'anthropic-api' listed
under 'Skills archived' and interpreted that as pruning — but the curator
had actually absorbed those skills into a new umbrella (e.g. 'llm-providers')
during the same run. The directory gets archived for safety (all removals
are recoverable), but the content still lives under a different name.
Users then 'restored' what they thought were deleted skills and ended up
with confusingly duplicated skillsets (old-name + absorbed-inside-umbrella).
Classify removed skills using this run's skill_manage tool calls:
- consolidated: content absorbed into a surviving/newly-created skill
(evidenced by a skill_manage write_file/patch/create/edit whose target
is a different skill AND whose file_path/content references the
removed skill's name)
- pruned: archived without consolidation evidence (truly stale)
REPORT.md now shows two distinct sections:
- 'Consolidated into umbrella skills' — with `removed → merged into umbrella`
- 'Pruned — archived for staleness' — pure staleness archives
run.json schema additions (backward compatible):
- counts.consolidated_this_run, counts.pruned_this_run
- consolidated: [{name, into, evidence}, ...]
- pruned: [names]
- archived: retained as the union for backward compat
Also: relabel the auto-transitions 'archived' counter to 'archived (no
LLM, pure time-based staleness)' so it's clearly distinct from LLM-pass
archives.
Tests: 9 new tests in test_curator_classification.py covering consolidation
evidence parsing (write_file/patch/create), hyphen/underscore name variants,
self-reference rejection, destination-must-exist, mixed runs, and
malformed-JSON fallback safety. Existing test_report_md_is_human_readable
updated to cover the new section names.
E2E: isolated HERMES_HOME, realistic 3-skill run, REPORT.md verified
end-to-end.
* feat(curator): hybrid model-declared + heuristic classification
Extend the consolidated-vs-pruned split with LLM-authored intent:
1. Curator prompt now requires a structured YAML block at the end of the
final response (consolidations / prunings with short rationale).
2. _parse_structured_summary() extracts it tolerantly — missing block,
malformed YAML, partial lists all fall back to heuristic cleanly.
3. _reconcile_classification() merges model intent with the tool-call
heuristic:
- Model wins on rationale when its umbrella exists post-run
- Model hallucination (umbrella doesn't exist) is downgraded to the
heuristic's finding, or pruned if there's no evidence either
- Heuristic catches model omission — consolidations the model
enumerated tools for but forgot to list get surfaced with a
'(detected via tool-call audit)' tag
4. REPORT.md now shows per-row rationale alongside 'removed → umbrella'
and flags audit-only rows so the user knows why no reason is shown.
Backward compat: run.json's 'archived' field (union) is preserved.
'pruned' is now a list of dicts with {name, source, reason};
'pruned_names' is the flat-name list for legacy consumers.
Tests: 15 new covering YAML parse edge cases (malformed, empty lists,
bare-string entries, missing fields), reconciler rules (model wins,
hallucination fallback, heuristic catches omission, prune with reason),
and an end-to-end report-render test with all four paths exercised.
* change(nix): dedupe nix lockfile checking scripts in ci
* feat(nix): make .#fix-lockfiles run --apply if no args passed
* fix(nix): use same nodejs version everywhere & small lints
- prevent lockfile thrashing while using nix :3
- use lib.getExe instead of raw /bin/ paths
- use inputs'.self instead of passing system in manually
* fix(nix): update lock files yet again (hopefully for the last time)
* fix(nix): align indentation of collision check echo
---------
Co-authored-by: Hermes Agent <hermes@nousresearch.com>
Fixes HTTP 404 errors when using Anthropic-compatible providers (Kimi Coding, MiniMax, MiniMax-CN) for auxiliary tasks.
Root cause: `_to_openai_base_url()` rewrites `/anthropic` → `/v1` so the OpenAI SDK hits the right endpoint. But the rewritten URL was then passed to `_maybe_wrap_anthropic`, whose `_endpoint_speaks_anthropic_messages` detector only fires on `/anthropic` or `api.kimi.com/coding`. Detector saw `/v1` → returned False → no Anthropic wrap → 404 on every aux call.
Fix: preserve the raw base_url before rewriting and pass it to `_maybe_wrap_anthropic` for transport detection, while still giving the rewritten URL to the OpenAI client constructor.
Closes#17705, #17413, #17086, #10469.
Co-authored-by: oak <chengoak@users.noreply.github.com>
* fix(nix): replace magic-nix-cache with Cachix
magic-nix-cache caused recurring CI failures (TwirpErrorResponse
ResourceExhausted) by hitting GitHub Actions Cache's 10 GB limit and
200 req/min rate limit. This was flagged as 'unfixable infra flake' in
#17836 but is actually a fixable architecture choice.
Switch to Cachix (dedicated binary cache, no GHA quota dependency):
- Replace DeterminateSystems/magic-nix-cache-action with cachix/cachix-action
- Add cachix-auth-token input to nix-setup composite action
- Pass CACHIX_AUTH_TOKEN secret through all three nix workflows
- continue-on-error: true so cache failures never block CI
Cache 'hermes-agent' is public at hermes-agent.cachix.org.
Devs can pull locally with: cachix use hermes-agent
* fix: correct cachix-action commit SHA pin
---------
Co-authored-by: Hermes Agent <hermes@nousresearch.com>
Widen #17818 to cover the dominant 'agent actively used this skill' path:
when the model calls the skill_view tool, bump use_count alongside view_count.
The slash-command and --skill preload paths (covered by the cherry-picked
commit) only catch user-initiated invocation; most skill activation happens
via the agent calling skill_view to consume an indexed skill.
Curator's stale-timer keys off last_used_at (agent/curator.py:233), so
without this wire-up agent-created skills would transition to stale
simultaneously regardless of actual use.
bump_use() existed and was tested but had zero production call sites —
use_count stayed 0 for all skills, breaking Curator's stale-detection
logic which relies on last_used_at.
Wire bump_use() into:
1. build_skill_invocation_message() — when a user invokes /skill-name
2. build_preloaded_skills_prompt() — when a skill is preloaded at session start
Both are the canonical 'a skill is actively being used' moments, distinct
from 'browsing' (bump_view in skill_view tool call).
Closes#17782
Belt-and-suspenders on top of @briandevans' #17758 fix. The in-band
drain hand-off (await->create_task + session-guard preservation)
changed cleanup semantics in three places that the original PR
reasoned about but didn't test directly. Pin each invariant so a
future refactor can't silently regress them:
1. Normal single-message path still releases _active_sessions[sk] and
_session_tasks[sk] through end-of-finally. The #17758 follow-up
moved _release_session_guard under
if current_task is self._session_tasks.get(session_key)
For the 99%-common case current_task IS the stored task, so the
guard must still fire. Test would fail if the conditional were
ever tightened in a way that dropped the normal path.
2. Drain-task cancellation releases the session. If the drain task
spawned by the in-band hand-off is cancelled mid-handler (e.g.
/stop fired while draining a follow-up), its own finally must
fire _release_session_guard. Without this a cancel would leave
the session permanently pinned busy.
3. Late-arrival drain still spawns when no in-band drain preceded
it. Pre-existing path, but the #17758 follow-up added a
re-queue branch that only fires when ownership was already
handed off. When no handoff happened the else branch must still
spawn a fresh drain task — otherwise a message arriving during
stop_typing gets silently dropped.
All three tests pass against current main. Zero production code
changes.
Widen #17639 to the fourth sibling site (tools/skills_tool.py _EXCLUDED_SKILL_DIRS)
and register leoneparise in scripts/release.py AUTHOR_MAP so CI release script
resolves the contributor.
Archived skills (moved to ~/.hermes/skills/.archive/ by the curator)
were still surfaced in the <available_skills> system prompt under a
fake '.archive' category, causing the agent to load and try to use
deprecated skills. The os.walk in iter_skill_index_files() only
excluded .git/.github/.hub.
Add '.archive' to EXCLUDED_SKILL_DIRS, and to the two other places
that hardcode the same exclusion tuple (gateway/run.py and
agent/skill_commands.py).
Three fixes bundled for curator reliability on existing installs and
broken/partial installs:
1. run_agent.py: defer `import fire` into the __main__ block. `fire` is
only used by `fire.Fire(main)` when running run_agent.py directly as
a CLI — it is NOT needed for library usage. Importing it at module
top made `from run_agent import AIAgent` from a daemon thread (e.g.
the curator's forked review agent) crash with ModuleNotFoundError
on broken/partial installs where `fire` isn't present.
2. hermes_cli/config.py: add version 22 → 23 migration that writes the
`curator` + `auxiliary.curator` sections to config.yaml with their
defaults, only filling keys the user hasn't overridden. Existing
configs from before PR #16049 / the April 2026 `auxiliary.curator`
unification had neither section on disk, so users couldn't see or
edit the settings in their config.yaml (runtime deep-merge papered
over it at read time, but the file never reflected reality).
3. hermes_cli/config.py: `ensure_hermes_home()` now pre-creates
`~/.hermes/logs/curator/` alongside cron/sessions/logs/memories on
every CLI launch. Managed-mode (NixOS) variant mkdir's it
defensively after the activation-script existence checks, since the
activation script may not know about this subpath.
4. agent/curator.py: `_reports_root()` mkdir's the dir at call time as
belt-and-suspenders for entry paths that bypass both
ensure_hermes_home() and the v23 migration (gateway-only installs,
bare library use).
E2E validated in isolated HERMES_HOME: fresh install gets full defaults
seeded; partial-override config keeps user's `enabled: false` and
custom `interval_hours` while filling the missing keys; re-running the
migration is a no-op.
The #1630 fix introduced a blanket ``agent_failed_early`` transcript skip
to prevent context-overflow sessions from looping. That guard also
triggers for unrelated transient failures (429 rate limits, read
timeouts, connection resets, provider 5xx) which have nothing to do with
session size — and it silently drops the user's message, so the agent
has no memory of the last turn on retry.
Split the failure classification in ``GatewayRunner._run_agent``:
* Context-overflow (``compression_exhausted`` flag, explicit
context-length phrases, or generic 400 with a long history) → keep
the existing skip, preserving the #1630/#9893 fix.
* Anything else that failed → persist just the user message so the
conversation survives a retry.
Use specific multi-word phrases (``context length``, ``token limit``,
``prompt is too long``, etc.) to match ``run_agent.py``'s own
classifier; bare ``exceed`` false-positively flagged "rate limit
exceeded" as context overflow.
Covered by new tests in ``tests/gateway/test_7100_transient_failure_transcript.py``
and the existing #1630 suite still passes.
Existing test_tar_pipe_commands asserted the literal substring
'tar xf - -C /' in ssh_str, which is no longer present after the
#17767 fix adds --no-overwrite-dir between 'tar xf -' and '-C /'.
Split the one substring check into three independent assertions for
the tar stdin mode, the new --no-overwrite-dir flag (regression guard
for #17767), and the extract target.
_set_nested unconditionally replaced any non-dict value with an empty
dict when walking the dotted path, which silently destroyed list-typed
config nodes the moment someone set a value with a numeric index
(e.g. 'hermes config set custom_providers.0.api_key NEW'). Any sibling
entries and any fields inside the targeted entry that the user didn't
write were lost.
Fix:
- _set_nested now detects list nodes and navigates by numeric index,
and preserves both dicts AND lists at intermediate positions (scalars
are still replaced so bare-scalar -> nested overrides keep working).
- set_config_value drops its duplicated navigation logic and calls
_set_nested instead -- single source of truth for the rules.
Regression tests (tests/hermes_cli/test_set_config_value.py):
- test_indexed_set_preserves_sibling_list_entries -- exact #17876 repro
- test_indexed_set_preserves_non_targeted_fields -- inner-dict fields survive
- test_deeper_nesting_through_list -- dict -> list -> dict -> scalar path
35/35 existing + new tests pass.
E2E-verified with the issue's repro against a real on-disk config.yaml --
list stays a list, entry 0 updated, entry 1 intact.
Closes#17876
When hermes model picker switches to a custom_providers entry, the slug
assignment can write the literal string 'custom' to model.provider if a
prior failed switch already left that value in config.yaml.
Two fixes:
1. model_switch.py: filter out bare 'custom' in slug assignment, always
resolve to canonical custom:<name> form
2. providers.py: resolve_custom_provider() self-heals bare 'custom' by
falling back to the first valid custom_providers entry
Closes#17478
Long-lived Gateway processes were sending duplicate tool names to
providers that enforce uniqueness:
- DeepSeek: 'Tool names must be unique.'
- Xiaomi MiMo: 'tools contains duplicate names: lcm_expand'
- Moonshot/Kimi: 'function name lcm_grep is duplicated'
TUI was unaffected because TUI runs with quiet_mode=False and skips the
cache entirely.
Root cause (two layered bugs)
- model_tools.get_tool_definitions(quiet_mode=True) memoizes its result
in _tool_defs_cache. The cache-hit path returned list(cached) (safe),
but the FIRST uncached call stored and returned the SAME object.
run_agent.py mutates self.tools (memory + LCM context-engine schemas)
in-place, so the very first agent init in a Gateway process
poisoned the cache, and every subsequent init appended LCM schemas
again on top of the already-polluted list.
- run_agent.py's context-engine injection (lcm_grep / lcm_describe /
lcm_expand) had no dedup, unlike the memory-tools injection right
above it which already skips already-present names.
Fix (defense in depth, per the issue's suggested fix)
- model_tools.get_tool_definitions: on the uncached branch, cache the
computed list but return list(result) to the caller. Same pattern as
the cache-hit path.
- run_agent.py: build _existing_tool_names from self.tools and skip
schemas whose names are already present, mirroring the memory-tools
block. This also defends against plugin paths that may register the
same schemas via ctx.register_tool().
Tests (tests/test_get_tool_definitions_cache_isolation.py)
- test_first_uncached_call_returns_fresh_list \u2014 pins the fix; without
it, first-call alias caused all the symptoms.
- test_cache_hit_returns_fresh_list \u2014 pre-existing behavior stays.
- test_caller_mutation_does_not_poison_cache \u2014 simulates run_agent
appending lcm_grep / lcm_expand to the returned list and asserts the
next call doesn't see them.
- test_repeated_caller_mutation_does_not_accumulate \u2014 reproduces the
long-lived Gateway accumulation pattern across 5 agent inits.
- test_non_quiet_mode_does_not_use_cache \u2014 sanity, explains why TUI
was fine.
5/5 pass on the new file; 23/23 still pass on tests/test_model_tools.py.
When a user sets model.context_length in config.yaml, the value was only
used for Hermes' internal compression decisions (context_compressor) but
NOT for Ollama's num_ctx parameter. Ollama auto-detects context from GGUF
metadata (often 256K+) and allocates that much VRAM regardless of the
user's config — causing OOM on smaller GPUs like the P100 (16GB).
Root cause: two separate context values existed independently:
- context_compressor.context_length = config value (e.g. 65536) ✓
- _ollama_num_ctx = GGUF metadata value (e.g. 256000) ✗ ignored config
Changes:
1. Cap Ollama num_ctx to config context_length (run_agent.py)
When model.context_length is explicitly set and no explicit
ollama_num_ctx override exists, cap the auto-detected GGUF value
to the user's context_length. This is the core fix — it prevents
Ollama from allocating more VRAM than the user budgeted.
2. Pass config_context_length through all secondary call sites
Several paths called get_model_context_length() without the config
override, falling through to the 256K default fallback:
- cli.py: @-reference expansion and /model switch display
- gateway/run.py: @-reference expansion and /model switch display
- tui_gateway/server.py: @-reference expansion
- hermes_cli/model_switch.py: resolve_display_context_length()
3. Normalize root-level context_length in config (hermes_cli/config.py)
_normalize_root_model_keys() now migrates root-level context_length
into the model section, matching existing behavior for provider and
base_url. Users who wrote `context_length: 65536` at the YAML root
instead of under `model:` had it silently ignored.
4. Fix misleading comments (agent/model_metadata.py)
DEFAULT_FALLBACK_CONTEXT is 256K (CONTEXT_PROBE_TIERS[0]), not 128K
as two comments stated.
Tests: 3 new tests for root-level context_length normalization.
All existing context_length tests pass (96 tests).
The busy-session handler (_handle_active_session_busy_message) bypassed the
authorization gate that the cold path enforces via _is_user_authorized(). In
shared-thread contexts (Slack threads, Telegram forum topics, Discord threads)
where thread_sessions_per_user=False (the default), all participants share one
session_key. An unauthorized user posting in the same thread as an authorized
user would hit the active-session branch, skip the auth check, and have their
text merged into _pending_messages or injected via agent.interrupt().
This commit adds the same _is_user_authorized() check at the top of the busy
handler, before any message queuing, steering, or interrupt logic. Unauthorized
messages are silently dropped (return True) with a warning log — matching the
cold-path behavior.
Affected platforms: Slack, Telegram, Discord, any adapter with shared-session
thread contexts.
Closes#17775
The `gemini` provider also serves Gemma (e.g. `gemma-4-31b-it`) and
historically other Google models like PaLM. Those reject
`extra_body.thinking_config` with HTTP 400:
Unknown name "thinking_config": Cannot find field
`_build_gemini_thinking_config()` was unconditionally producing a
config dict for any model on the `gemini` / `google-gemini-cli`
provider, which `ChatCompletionsTransport.build_kwargs` then dropped
into `extra_body["thinking_config"]`. The result: every chat turn for
Gemma users on the gemini provider blew up at the API edge.
The fix is the same shape Hermes already uses for the Gemini-2.5 vs
Gemini-3 family clamping: normalise the model id, strip an
`OpenRouter`-style `google/` prefix, and short-circuit early when the
result doesn't start with `gemini`. We return `None` rather than
`{"includeThoughts": False}`, because the API rejects the field name
itself — even the polite "off" form trips the same 400.
Three regression tests cover Gemma with reasoning enabled, Gemma with
reasoning disabled, and the `google/gemma-…` OpenRouter-style id; the
existing Gemini-2.5 / Gemini-3 / `google/gemini-…` cases keep passing
because the Gemini guard fires after the prefix strip.
Fixes#17426
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Ports PR #17888's send_multiple_images ABC to every gateway platform that
has a native multi-attachment API, so images arrive as a single bundled
message instead of N separate ones.
Native overrides:
- Telegram: send_media_group (10 photos per album, chunks over); animated
GIFs peeled off and routed through send_animation (albums don't support
animations)
- Discord: channel.send(files=[...]) (10 attachments per message, chunks
over); URL images downloaded into BytesIO so they render inline; forum
channels use create_thread with files=[...]
- Slack: files_upload_v2(file_uploads=[...]) (10 per call, chunks over);
respects thread_ts; records thread participation
- Mattermost: single post with file_ids list (5 per post — Mattermost cap,
chunks over)
- Email: single SMTP message with multiple MIME attachments (no chunk cap,
SMTP size governs); remote URLs remain linked in body (parity with
existing send_image)
All platforms fall back to the base per-image loop on any failure, so a
single bad image in a batch never loses the rest.
Matrix, WhatsApp, and single-attachment platforms (BlueBubbles, Feishu,
WeCom, WeChat, DingTalk) continue to use the base default loop — their
server APIs only accept one attachment per message anyway.
Tests: adds tests/gateway/test_send_multiple_images.py with 19 targeted
tests covering base default loop, chunking, animation peel-off, fallback
paths, and empty-batch no-ops across all five new overrides.
Co-authored-by: Maxence Groine <maxence@groine.fr>
Adds a new `send_multiple_images` method to the ``BasePlatformAdapter``
that implements the default "One image per message" loop and allows for
platform-specific overriding.
Implements such an override for the Signal adapter, batching images
and trying (best-effort) to work around rate-limits for voluminous
batches using a specific scheduler.
Also implements batching + rate-limit handling in the `send_message`
tool.
New tests added for the Signal adapter, its rate-limit scheduler and the
`send_message` tool
Merge resolved conflicts in web/src/{i18n/{en,zh,types}.ts,lib/api.ts}
by keeping both this branch's `profiles` additions and upstream's new
`models` page additions.
Copilot review feedback:
- Implement POST /api/profiles/{name}/open-terminal endpoint (already
present); align Windows branch to `cmd.exe /c start "" <cmd>` so it
matches the new test and spawns a fresh window instead of /k reusing
the parent console.
- Move backslash escaping out of the macOS AppleScript f-string
expression (Python <3.12 disallows backslashes inside f-string
expression parts).
- Patch `_get_wrapper_dir` via monkeypatch in
test_profiles_create_creates_wrapper_alias_when_safe so the test no
longer writes to the real `~/.local/bin`.
- Extend test_dashboard_browser_safe_imports to scan `.ts` files in
addition to `.tsx`.
- Switch upstream's new ModelsPage.tsx away from the `@nous-research/ui`
root barrel onto per-component subpaths to satisfy the stricter scan.
- Fix NouiTypography `leading-1.4` -> `leading-[1.4]` so Tailwind
actually emits the line-height for the `sm` variant.
- Guard ProfilesPage.openSoulEditor against out-of-order responses by
tracking the latest requested profile via a ref.
- Replace ProfilesPage's hand-rolled setup command with a fetch to
`/api/profiles/{name}/setup-command` so the copied command always
matches what the backend would actually run (handles wrapper-alias
collisions and reserved names correctly).
- Wire SOUL.md textarea label `htmlFor` -> textarea `id` so screen
readers and clicking the label work as expected.
Follow-up to the try/except guards added in the previous commit.
Four sibling call sites all read HERMES_AGENT_TIMEOUT /
HERMES_AGENT_TIMEOUT_WARNING / HERMES_AGENT_NOTIFY_INTERVAL via the
same read-env-or-fallback pattern, so factor it into _float_env(name,
default) alongside the existing _auto_continue_freshness_window()
helper.
Two defensive fixes in gateway/run.py:
1. yaml.safe_load returning None on empty config files (line 12706):
GatewayConfig.from_dict(data) crashes with AttributeError when the YAML
file is empty because safe_load returns None. All 6 other yaml.safe_load
call sites already use `or {}` — this one was missed.
Impact: gateway fails to start with empty --config file.
2. float() on env vars without ValueError guard (lines 3951, 11757, 11805,
11807): HERMES_AGENT_TIMEOUT, HERMES_AGENT_TIMEOUT_WARNING, and
HERMES_AGENT_NOTIFY_INTERVAL are cast via float() directly from
os.getenv(). A typo (e.g. "abc") raises ValueError and crashes the
agent turn or gateway startup.
Impact: single misconfigured env var crashes the entire gateway.
The sandbox-side `_call()` in both the UDS and file-based transports was
not thread-safe, so scripts that call tools from multiple threads (e.g.
`ThreadPoolExecutor` over `terminal()`) inside a single `execute_code`
run could silently receive each other's responses.
Root cause:
* UDS transport — a single module-level `_sock` was shared across all
threads; the newline-framed protocol has no request-id; and the
server-side RPC loop handles one connection serially. With concurrent
callers, each thread would `sendall()` then race to `recv()` the next
newline-terminated response from the shared buffer, so responses got
delivered to the wrong caller.
* File transport — `_seq += 1` is a non-atomic read-modify-write, so
two threads could allocate the same sequence number and clobber each
other's request/response files.
Fix: guard `_call()` with a `threading.Lock` in the UDS case (covering
send+recv), and guard `_seq` allocation with a lock in the file case.
No protocol change.
Regression tests cover both the generated-source level (lock is present
and used) and an end-to-end concurrency test: running a sandboxed
ThreadPoolExecutor of 10 `terminal()` calls against a slow mock
dispatcher, asserting every caller sees its own tagged response. The
test fails without the fix (10/10 mismatched, matching real-world
repro) and passes with it.
The v11→v12 migrate_config step writes the API mode for every entry
under the new transport: field (per the v12+ schema in
_normalize_custom_provider_entry). _get_named_custom_provider
read the legacy api_mode: spelling only, so for every migrated
config the lookup returned None for the api mode.
Downstream, _resolve_named_custom_runtime then falls back through
custom_provider.get("api_mode") or _detect_api_mode_for_url(base_url)
or "chat_completions". For loopback URLs (proxies, local servers)
or unknown hostnames, the URL detector returns None and the resolver
silently downgrades the configured codex_responses /
anthropic_messages transport to chat_completions. Requests
get sent to /v1/chat/completions instead of /v1/responses or
/v1/messages and the provider 404s — or worse, returns a usable
chat_completions response while skipping the model's reasoning /
caching surface.
Fix: read both field names — entry.get("api_mode") or
entry.get("transport") — at the two match-by-key + match-by-name
branches in _get_named_custom_provider. The runtime normaliser
_normalize_custom_provider_entry already accepts both spellings;
this lifts the same compat into the direct-dict reader so v12+
configs work without going through the shim.
Adds three regression tests under
tests/hermes_cli/test_user_providers_model_switch.py:
- transport field is read on the match-by-key branch
- legacy api_mode spelling still works for hand-edited configs
- transport is read on the match-by-display-name branch
run_job() ignored the result's `failed=True` / `completed=False` flags
that agent.run_conversation populates on API exhaustion, mid-run
interrupts, and model aborts. Because final_response on those paths is
often a non-empty error string ("API call failed after 3 retries:
Request timed out."), the existing empty-response soft-fail in
_process_job did not trip either: the error text was delivered as if it
were the agent's reply and last_status was set to "ok" with no error
notification. Detect those flags right after the dict-shape guard and
raise so the existing except handler builds the proper failure tuple,
preserving the agent's error message via result["error"].
Adds a parametrized regression covering: API-retry-exhausted with error
text in final_response, completed=False with no final_response,
completed=False without an explicit failed flag, and the partial-reply
plus failed=True case. Plus a guard that a normal completed=True success
result is still treated as success.
Fixes#17855
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
When the in-band pending-message drain spawns a fresh task and
transfers ownership via _session_tasks[session_key] = drain_task,
the original task still unwinds through the finally block. The
drain task picks up the same interrupt_event in its own
_process_message_background entry, so an unconditional
_release_session_guard(session_key, guard=interrupt_event) at the
end of the finally matches and deletes _active_sessions[session_key]
while the drain task is still pending its first await.
A concurrent inbound message arriving in that handoff window passes
the Level-1 guard (no entry exists) and spawns a second
_process_message_background for the same session — two agents on
one session_key, duplicate responses, duplicate tool calls.
Fix: only call _release_session_guard when the current task still
owns _session_tasks[session_key]. When ownership has been
transferred to a drain task, leave _active_sessions populated; the
drain task's own lifecycle releases it. This mirrors the
late-arrival drain path in the same finally block, which already
leaves both entries alone after handing off.
Also reorder stdlib imports in the new regression test file to
match the gateway test convention (stdlib before third-party).
Regression test: capture _active_sessions[sk] identity at every
handler entry across a 2-step in-band drain chain and assert the
guard Event identity stays the same. Pre-fix, the original task's
finally deletes the entry, the drain task falls through to the
`or asyncio.Event()` branch, and a fresh Event is installed —
identity diverges. Post-fix, the entry is preserved and the drain
task reuses the original Event.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
`_process_message_background` finished a turn, found a queued
follow-up, and drained it via `await
self._process_message_background(pending_event, session_key)`. Each
chained follow-up added a frame to the call stack instead of starting
fresh. Under sustained pending-queue activity (e.g. a user sending
follow-ups faster than the agent finishes turns) the C stack would
exhaust at ~2000 nested frames and SIGSEGV the process.
Mirror the late-arrival drain pattern that already exists in the same
function: spawn a new `asyncio.create_task(...)` for the pending event
and return so the current frame can unwind. The new task takes
ownership via `_session_tasks[session_key]`.
The late-arrival drain in `finally` could now race with the in-band
drain across the `await typing_task` / `await stop_typing` window, so
add a guard: if `_session_tasks[session_key]` is no longer the current
task, an in-band drain already spawned a follow-up task — re-queue the
late-arrival event so that task picks it up after its current event,
instead of spawning a second concurrent task for the same session_key.
Regression test (`test_pending_drain_no_recursion.py`) chains 12
follow-ups and asserts the recorded
`_process_message_background` stack depth stays bounded at handler
entry. Pre-fix: depths grow linearly `[1,2,3,…,12]`. Post-fix: all
depths are `1`.
`test_duplicate_reply_suppression::test_stale_response_suppressed_when_interrupted`
called `_process_message_background` directly and implicitly relied on
the old recursive `await` semantic — updated to wait for the spawned
drain task before checking the sent list.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
tar xf - -C / extracts the staging directory tree to the remote root.
GNU tar default behavior overwrites metadata (including mode) of existing
directories. When the local umask is 002 (Ubuntu default), the staging
dirs are 0775, and tar chmod's /home/<user> to 0775 — breaking sshd
StrictModes which requires 0755 or stricter for home dirs.
Add --no-overwrite-dir to the remote tar command so existing directory
metadata is preserved.
Fixes#17767
Piper (OHF-Voice/piper1-gpl) is a fast, local neural TTS engine from the
Home Assistant project that supports 44 languages with zero API keys.
Adds it as a native built-in provider alongside edge/neutts/kittentts,
installable via 'hermes tools' with one keystroke.
What ships:
- New 'piper' built-in provider in tools/tts_tool.py
- Lazy import via _import_piper()
- Module-level voice cache keyed on (model_path, use_cuda) so switching
voices doesn't invalidate older cached voices
- _resolve_piper_voice_path() accepts either an absolute .onnx path or a
voice name (auto-downloaded on first use via 'python -m
piper.download_voices --download-dir <cache>')
- Voice cache at ~/.hermes/cache/piper-voices/ (profile-aware via
get_hermes_dir)
- Optional SynthesisConfig knobs: length_scale, noise_scale,
noise_w_scale, volume, normalize_audio, use_cuda — passed through
only when configured, so older piper-tts versions aren't broken
- WAV output then ffmpeg conversion path (same as neutts/kittentts) so
Telegram voice bubbles work when ffmpeg is present
- Piper added to BUILTIN_TTS_PROVIDERS so a user's
tts.providers.piper.command cannot shadow the native provider
(regression test included)
- 'hermes tools' wizard entry
- Piper appears under Voice and TTS as local free, with
'pip install piper-tts' auto-install via post_setup handler
- Prints voice-catalog URL and default-voice info after install
- config.yaml defaults
- tts.piper.voice defaults to en_US-lessac-medium
- Commented advanced knobs for discoverability
- Docs
- New 'Piper (local, 44 languages)' section in features/tts.md
explaining install path, voice switching, pre-downloaded voices,
and advanced knobs
- Piper listed in the ten-provider table and ffmpeg table
- Custom-command-providers section updated to drop the Piper example
(now native) and add a piper-custom example for users with their own
trained .onnx models
- overview.md bumps provider count to ten
- Tests (tests/tools/test_tts_piper.py, 16 tests)
- Registration (BUILTIN_TTS_PROVIDERS, PROVIDER_MAX_TEXT_LENGTH)
- _resolve_piper_voice_path across every branch: direct .onnx path,
cached voice name, fresh download with correct CLI args, download
failure, successful-exit-but-missing-files, empty voice to default
- _generate_piper_tts: loads voice once, reuses cache, voice-name
download wiring, advanced knobs flow through SynthesisConfig
- text_to_speech_tool end-to-end dispatch and missing-package error
- check_tts_requirements: piper availability toggles the return value
- Regression guard: piper cannot be shadowed by a command provider
with the same name
- Pre-existing test_tts_mistral test broadened to mock the new
piper/kittentts/command-provider checks (otherwise it false-passes
when piper is installed in the test venv)
E2E verification (live):
Actual pip install piper-tts, config piper + en_US-lessac-low,
text_to_speech_tool call, voice auto-downloaded from HuggingFace,
WAV synthesized, ffmpeg-converted to Ogg/Opus. Second call hits the
cache (~60ms). Cache dir populated with .onnx and .onnx.json.
This caught a real bug during development: the first pass used '-d' as
the download-dir flag; the actual piper.download_voices CLI wants
'--download-dir'. Fixed before PR opened.
Six tests in this file failed in CI (-n auto) after #17832 landed because
other tests on the same xdist worker reload hermes_cli.main:
tests/hermes_cli/test_env_loader.py:85-86
sys.modules.pop('hermes_cli.main', None)
importlib.import_module('hermes_cli.main')
tests/hermes_cli/test_skills_subparser.py:24-25
del sys.modules['hermes_cli.main']
When either ran first on a worker, our top-of-file
'from hermes_cli.main import _kill_stale_dashboard_processes' captured a
stale function object whose __globals__ points at the old module dict.
patch('hermes_cli.main._find_stale_dashboard_pids', ...) then patched the
new module, but the stale function resolved the dependency via its stale
__globals__, so every patch became a no-op: pids=[] → early return → no
signals, no output, assertions failed.
Fix: add an autouse fixture that rebinds the three module-level names to
whatever is currently live in sys.modules['hermes_cli.main'] before each
test runs. The pollutants in the other two files are load-bearing for
their own tests, so fixing it on the consumer side is correct.
Repro: pytest tests/hermes_cli/test_env_loader.py tests/hermes_cli/test_update_stale_dashboard.py
Voscko reported curator.auxiliary.provider/model was advertised in the
docs but ignored — the review fork read only model.provider/default. The
narrow fix would wire the one-off key through, but that leaves curator
as a parallel system: not in `hermes model` → auxiliary picker, not in
the dashboard Models tab, missing per-task base_url/api_key/timeout/
extra_body.
Unify curator with the rest of the aux task system so `hermes model`
and the dashboard configure it like every other aux task.
Four sources of truth updated:
- hermes_cli/config.py — add 'curator' slot to DEFAULT_CONFIG.auxiliary
(timeout=600 since reviews run long), drop the one-off curator.auxiliary
block from DEFAULT_CONFIG.curator.
- hermes_cli/main.py — add ('curator', 'Curator', 'skill-usage review pass')
to _AUX_TASKS so the CLI picker offers it.
- hermes_cli/web_server.py — add 'curator' to _AUX_TASK_SLOTS so the
dashboard REST endpoint accepts it.
- web/src/pages/ModelsPage.tsx — add Curator entry so the dashboard
Models tab renders the task.
agent/curator.py _resolve_review_model() now reads auxiliary.curator
first (canonical), falls back to legacy curator.auxiliary (with an info
log asking users to migrate), then falls back to the main chat model.
Pre-unification users keep working.
Docs updated: docs/user-guide/features/curator.md now points at
`hermes model` → auxiliary → Curator and the dashboard Models tab.
Tests: 6 unit tests on _resolve_review_model (auto default, canonical
slot honored, partial override fallback, legacy fallback with
deprecation log assertion, new-wins-over-legacy, empty-config safety)
plus a cross-registry test that curator is wired into all four sources
of truth. test_aux_tasks_keys_all_exist_in_default_config already
covers the DEFAULT_CONFIG ↔ _AUX_TASKS invariant.
Reported by Voscko on Discord.
UserMessageChunk and AgentMessageChunk do not have a message_id field
in the ACP schema. Passing it silently dropped the kwarg (pydantic
does not raise on unknown init kwargs here) and the subsequent test
assertions on .message_id raised AttributeError. Strip the dead
plumbing (uuid import, message_id= kwarg on both chunk types, unused
session_id/index parameters) and remove the matching .message_id
asserts from the test.
Adds a deterministic pre-check on top of htsh's exception-based fallback:
before calling /content/abstract or /content/overview on a non-pseudo URI,
probe /api/v1/fs/stat. If the server says the URI is a file, route straight
to /content/read instead of eating a failing 500 round-trip.
This is the same idea pty819 and chennest independently landed in PRs
#12757 and #12937 — merged here on top of htsh's broader fix so we keep
pseudo-URI normalization and v0.3.3 browse-shape handling while avoiding
the slow exception path on servers that return a raised 500 every time.
The exception fallback from #5886 stays in place for environments where
fs/stat is unavailable or returns an unfamiliar shape.
Also credits pty819, chennest, and htsh in AUTHOR_MAP so future release
notes attribute them correctly.
OpenViking returns 500 for /content/abstract and /content/overview when URI points to mem_*.md files.
Add resilient fallback to /content/read for non-pseudo summary file URIs while preserving pseudo summary normalization.
Also add regression tests for fallback behavior.
OpenViking v0.3.3 expects directory URIs for abstract/overview reads.
Passing pseudo-files like /.overview.md and /.abstract.md to
/api/v1/content/overview|abstract triggers HTTP 500.
This change normalizes those pseudo-URIs to their parent directory for
abstract/overview requests, preserves full reads, and hardens parsing for
wrapped/unwrapped result payloads and fs list response shapes.
Seed the tips corpus with the knobs users can turn to reduce token
spend: hermes tools / hermes skills config to trim surface area,
/reasoning low|minimal to dial thinking depth down from the medium
default, and hermes models to route auxiliary tasks (vision, compression,
title gen, session_search) to cheaper backends while the main chat model
stays intact.
Requested by @micheltamanda under Teknium's tip-of-the-day tweet.
`hermes dashboard` is a long-lived foreground server that users often
start and forget about, sometimes in a shell they've since closed. We
didn't have a way to stop it — users had to find the PID manually.
Adds two lifecycle flags that reuse the same detection + termination
path the post-`hermes update` cleanup (PR #17832) uses:
hermes dashboard --status
List running hermes dashboard processes with PID + cmdline.
Exit 0, informational.
hermes dashboard --stop
Terminate all running dashboards (3s grace then force-kill survivors).
Exit 0 if none remain, 1 if any couldn't be stopped.
Windows uses `taskkill /F` as before.
Both flags short-circuit before any fastapi/uvicorn import so they work
even on installations where the dashboard extras aren't installed —
useful when you're cleaning up after uninstalling.
The kill helper gained an optional `reason=...` param so the output
reads "(requested via --stop)" instead of the post-update-specific
"running backend no longer matches the updated frontend" wording.
E2E: `hermes dashboard --status` with nothing running prints the
empty message; with a fake `hermes dashboard ...` cmdline spawned via
`exec -a`, `--status` lists it, `--stop` terminates it (exit -15),
and a follow-up `--status` returns empty.
Reshape of PR #17211 (@versun). Lets users wire any local or external
TTS CLI into Hermes without adding engine-specific Python code. Users
declare any number of named providers in config.yaml and switch between
them with tts.provider: <name>, alongside the built-ins (edge, openai,
elevenlabs, …).
Config shape:
tts:
provider: piper-en
providers:
piper-en:
type: command
command: 'piper -m ~/model.onnx -f {output_path} < {input_path}'
output_format: wav
Placeholders: {input_path}, {text_path}, {output_path}, {format},
{voice}, {model}, {speed}. Use {{ / }} for literal braces.
Key behavior:
- Built-in provider names always win — a tts.providers.openai entry
cannot shadow the native OpenAI provider.
- type: command is the default when command: is set.
- Placeholder values are shell-quote-aware (bare / single / double
context), so paths with spaces and shell metacharacters are safe.
- Default delivery is a regular audio attachment. voice_compatible: true
opts in to Telegram voice-bubble delivery via ffmpeg Opus conversion.
- Command failures (non-zero exit, timeout, empty output) surface to
the agent with stderr/stdout included so you can debug from chat.
- Process-tree kill on timeout (Unix killpg, Windows taskkill /T).
- max_text_length defaults to 5000 for command providers; override
under tts.providers.<name>.max_text_length.
Tests: tests/tools/test_tts_command_providers.py — 42 new tests cover
provider resolution, shell-quote context, placeholder rendering with
injection payloads, timeout, non-zero exit, empty output, voice_compatible
opt-in, and end-to-end dispatch through text_to_speech_tool. All 88
pre-existing TTS tests still pass.
Docs: new "Custom command providers" section in
website/docs/user-guide/features/tts.md with three worked examples
(Piper, VoxCPM, MLX-Kokoro), placeholder reference, optional keys,
behavior notes, and security caveat.
E2E-verified live: isolated HERMES_HOME, command provider declared in
config.yaml, text_to_speech_tool dispatches through the registered
shell command and the output file is produced as expected.
Co-authored-by: Versun <me+github7604@versun.org>
`hermes update` previously just printed a warning when it detected a
running `hermes dashboard` process from the previous version, telling
the user to kill and restart it themselves. In practice dashboards get
started and forgotten, so the warning was routinely ignored and users
ended up with a silent frontend/backend mismatch (new JS bundle served
against the old in-memory Python backend, e.g. new auth headers the old
code doesn't recognise → every API call 401s).
The dashboard has no service manager, no PID file, and we don't record
the original launch args (--host, --port, --insecure, --tui, --no-open)
so we can't auto-restart it. But we CAN stop it, which is what the
user wants — the failure mode when the stale process is left alive is
worse than the dashboard just being down.
- POSIX: SIGTERM, poll for ~3s, SIGKILL any survivors.
- Windows: `taskkill /PID <pid> /F`.
- Print each PID's outcome plus a one-line restart hint.
- Detection logic is unchanged (same ps / wmic scan, same guards
against the `pgrep -f` greedy-match trap from #16872 and the
#17049 wmic UnicodeDecodeError fix).
Also split the old monolithic `_warn_stale_dashboard_processes` into
`_find_stale_dashboard_pids` (scan) + `_kill_stale_dashboard_processes`
(kill), keeping the old name as an alias so any external callers still
work.
E2E verified: spawned a fake `hermes dashboard` cmdline via
`exec -a 'hermes dashboard …' sleep 300`, ran
`_kill_stale_dashboard_processes()`, confirmed SIGTERM exit (-15)
and that a post-scan returns an empty PID list.
Three narrow fixes targeting the remaining red checks after #17828:
1. ui-tui/src/app/slash/commands/ops.ts (Docker Build):
/reload-mcp's local params type annotated session_id: string
while ctx.sid is string | null. Widen to string | null —
matches every other rpc call site and the test harness which passes
{ session_id: null }. Fixes TS2322 on line 86. The rpc signature
itself is Record<string, unknown>, so this is purely a local
typing fix, no behavioral change.
2. tests/plugins/test_achievements_plugin.py (13 cascading test failures):
_install_fake_session_db did a raw sys.modules['hermes_state'] =
fake_module without restoration, leaking the fake across xdist
worker boundaries. Downstream tests doing from hermes_state import
SessionDB got a module whose SessionDB was lambda: fake_db
— 6 test_hermes_state.py tests failed with AttributeError: 'function'
object has no attribute '_sanitize_fts5_query' / _contains_cjk,
and 7 test_860_dedup.py tests failed with TypeError: got unexpected
keyword argument 'db_path' (real code calls SessionDB(db_path=...)).
Fix: stash monkeypatch on the plugin_api module object in the
fixture, and have the helper do monkeypatch.setitem(sys.modules,
'hermes_state', fake_module) for auto-restoration at test teardown.
3. tests/hermes_cli/test_web_server.py (WS race):
TestPtyWebSocket::test_pub_broadcasts_to_events_subscribers hit the
30s test timeout on CI. websocket_connect returns after
ws.accept() — but /api/events registers the subscriber in
_event_channels on the NEXT await (inside _event_lock). A
publish immediately after connect could race ahead of registration
and be dropped, and the subsequent receive_text() blocked until
SIGALRM killed the test. Fix: poll _event_channels after the
subscriber connects, before publishing.
Validation:
scripts/run_tests.sh tests/plugins/test_achievements_plugin.py
tests/run_agent/test_860_dedup.py
tests/test_hermes_state.py
tests/hermes_cli/test_web_server.py 338 passed
cd ui-tui && npm run type-check clean
cd ui-tui && npm run build clean
Remaining red checks are pure infra (Nix ubuntu hits
TwirpErrorResponse ResourceExhausted on the GH Actions cache API; Nix
macos bounces between npm build openssl-legacy and cache rate-limits)
and cannot be fixed in the codebase.
Extracted from PR #17211 (@versun) so it can land independently of the
local_command TTS provider redesign.
- Add should_send_media_as_audio(platform, ext, is_voice) in
gateway/platforms/base.py; single source of truth for audio routing.
- Add .flac to recognized audio extensions (MEDIA regex, weixin audio
set, send_message audio set).
- Telegram send_voice() now falls back to send_document for formats
Telegram's Bot API can't play natively (.wav, .flac, ...) instead of
raising; MP3/M4A still go to sendAudio, Opus/OGG still go to sendVoice.
- Route _send_telegram() in send_message_tool through a narrower
_TELEGRAM_SEND_AUDIO_EXTS = {.mp3, .m4a} set.
- cron.scheduler._send_media_via_adapter now delegates the audio
decision to should_send_media_as_audio so it matches the gateway.
- Update the cron live-adapter ogg test to flag [[audio_as_voice]] so
it still routes to sendVoice under the new Telegram-specific policy.
- Tests: unit coverage for should_send_media_as_audio across platforms,
end-to-end MEDIA routing via _process_message_background and
GatewayRunner._deliver_media_from_response, TelegramAdapter.send_voice
fallback for FLAC/WAV.
Co-authored-by: Versun <me+github7604@versun.org>
Fixes the xdist collision that broke CI on PR #17764, and structurally
prevents future plugin-adapter tests from reintroducing it.
Problem
-------
tests/gateway/test_teams.py (new in this PR) and tests/gateway/test_irc_adapter.py
(already on main) both followed the same anti-pattern:
sys.path.insert(0, str(_REPO_ROOT / 'plugins' / 'platforms' / '<name>'))
from adapter import <Adapter>
Every platform plugin ships its own adapter.py, so the bare
'from adapter import ...' races for sys.modules['adapter']. Whichever test
collected first in a given xdist worker won; the other crashed at
collection with ImportError, and the polluted sys.path cascaded into 19
unrelated test failures across tools/, hermes_cli/, and run_agent/ in the
same worker.
Fix
---
1. tests/gateway/_plugin_adapter_loader.py (new): shared helper
load_plugin_adapter('<name>') that imports plugins/platforms/<name>/adapter.py
via importlib.util under the unique module name plugin_adapter_<name>.
Zero sys.path mutation, no possibility of collision.
2. tests/gateway/test_irc_adapter.py and tests/gateway/test_teams.py:
migrated to the helper. All 'from adapter import ...' statements
(including the ones inside test methods) are replaced with module-level
attribute access on the loaded module.
3. tests/gateway/conftest.py: new pytest_configure guard that AST-scans
every test_*.py under tests/gateway/ at session start and fails the
run with a pointer to the helper if any test uses sys.path.insert into
plugins/platforms/ OR a bare 'import adapter' / 'from adapter import'.
Runs on the xdist controller only (skipped in workers). The next plugin
adapter test that tries to reintroduce this pattern gets rejected at
collection time with a clear remediation message.
4. scripts/release.py: add aamirjawaid@microsoft.com -> heyitsaamir to
AUTHOR_MAP so the check-attribution workflow passes.
Validation
----------
scripts/run_tests.sh tests/gateway/ 4194 passed
scripts/run_tests.sh tests/gateway/test_{teams,irc}* 72 passed (both orderings)
scripts/run_tests.sh <11 prev-failing test files> 398 passed
Guard triggers correctly on both Path-operator and string-literal forms
of the anti-pattern.
Replace the Azure portal credential prompts with the teams CLI
workflow: install @microsoft/teams.cli, run teams app create,
paste the output credentials. Matches the setup docs.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Pass cmd/desc in button action data so the card response can
reconstruct the original body. Clicking a button now replaces
only the actions with a status line, keeping the command and
reason text visible.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
The gateway calls send_image_file() for locally cached images
(e.g. from image_gen tools). Without this override the base class
falls back to sending the file path as plain text. Delegate to
send_image() which already handles base64 encoding local paths.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Teams doesn't render markdown image syntax. Send images using the SDK's
Attachment API instead — base64 data URI for local files, direct URL
for remote images.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
_get_platform_tools() correctly fell back to f"hermes-{platform}" for
unknown (plugin) platforms when building toolset_names, but then
unconditionally used PLATFORMS[platform] again for platform_tool_universe,
causing KeyError for any plugin-registered platform like Teams.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Hello! I am the maintainer of the microsoft-teams-apps Python SDK and
I built this Teams adapter to integrate Microsoft Teams into Hermes.
Adds a `plugins/platforms/teams` platform plugin using the new
PlatformRegistry system from #17751. The adapter self-registers via
`register(ctx)` — no hardcoding in run.py, toolsets.py, or any
other core file.
Key features:
- Supports personal DMs, group chats, and channel posts
- Adaptive Card approval prompts with in-place button replacement
(Allow Once / Allow Session / Always Allow / Deny)
- aiohttp webhook server bridged from the Teams SDK to avoid
the fastapi/uvicorn dependency
- ConversationReference caching for correct proactive sends in
non-DM chats
- `interactive_setup()` for `hermes gateway setup` integration
- `platform_hint` for LLM context (Teams markdown subset)
- 34 tests covering adapter init, send, message handling, and
plugin registration
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
PR #17660 landed a sweep of CI fixes but left three loose ends:
1. tests/cli/test_cli_loading_indicator.py::test_reload_mcp_sets_busy_state_
and_prints_status — /reload-mcp gained a prompt-cache-invalidation
confirmation (commit 4d7fc0f37) that was never wired into this test.
The test exercises the loading-indicator path, so pre-approve via
config and go straight into _reload_mcp().
2. tools/mcp_tool.py _make_tool_handler — the added
getattr(server, '_rpc_lock', None) + 'skip the lock if missing'
branch is inconsistent with four sibling call sites that still
direct-access server._rpc_lock. The lock is guaranteed by
MCPServerTask.__init__; falling through to an unlocked
session.call_tool would silently serialize-strip RPCs if the guard
ever triggered. Restore direct access.
3. tui_gateway/server.py _messages_as_conversation — the helper
existed only to catch 'TypeError: include_ancestors unexpected'
from mocked SessionDBs that don't actually exist. The real
SessionDB.get_messages_as_conversation has accepted
include_ancestors since introduction, and every test FakeDB in
the repo already declares the kwarg. Remove the shim, inline the
two call sites.
Dashboard Models page was analytics-only — no way to pick a model as main
for new sessions or override an auxiliary task slot without hand-editing
config.yaml or running a /model slash command inside a chat.
Changes:
- hermes_cli/web_server.py: three REST endpoints (GET /api/model/options,
GET /api/model/auxiliary, POST /api/model/set). Reuses
list_authenticated_providers() from model_switch.py so the REST path
surfaces the same curated model lists as the TUI-gateway model.options
JSON-RPC. POST /api/model/set writes model.provider + model.default for
scope=main, and auxiliary.<task>.{provider,model} for scope=auxiliary
(with task="" meaning 'all 8 slots' and task="__reset__" resetting them
to auto).
- web/src/components/ModelPickerDialog.tsx: accepts an optional loader +
onApply pair so it works without an open chat PTY. ChatSidebar's
gw-WebSocket path still works unchanged (back-compat).
- web/src/pages/ModelsPage.tsx: Model Settings panel at the top showing
main model + collapsible list of 8 auxiliary tasks with per-row Change
buttons and Reset all to auto. Every existing model card gets a
'Use as' dropdown for one-click assignment to main or any aux slot.
Cards badged 'main' or 'aux · <task>' when currently assigned.
- website/docs/user-guide/configuring-models.md: new docs page walking
through both UI paths, aux task override patterns, troubleshooting,
plus REST/CLI alternatives.
- Screenshots under website/static/img/docs/dashboard-models/.
Applies to new sessions only — running sessions keep their model (use
/model slash command to hot-swap a live session). No prompt-cache
invalidation on existing sessions.
Dashboard plugin API routes (web_server._mount_plugin_api_routes) and
gateway event hooks (gateway.hooks.HookRegistry.discover_and_load) both
loaded Python files via importlib.util.spec_from_file_location +
exec_module without registering the resulting module in sys.modules.
That breaks any plugin or hook handler that uses `from __future__ import
annotations` together with a Pydantic BaseModel / dataclass / anything
that introspects `__module__`: at first request Pydantic tries to
resolve string-form type hints against the defining module's namespace,
can't find it by name, and raises:
PydanticUserError: TypeAdapter[...] is not fully defined;
you should define ... and all referenced types,
then call `.rebuild()` on the instance.
This is what broke the kanban dashboard's 'triage' button — POST
/api/plugins/kanban/tasks validated against CreateTaskBody (a Pydantic
model in a file using `from __future__ import annotations`) and
returned 500 on every click.
The fix, applied symmetrically to both loaders:
1. Compute module_name once.
2. Register the module in sys.modules BEFORE exec_module.
3. On exec_module failure, pop the half-initialized stub so subsequent
reloads don't pick up broken state.
GETs were unaffected because they don't build a body TypeAdapter, which
is why this only surfaced when users started POSTing.
* feat(plugins): bundle hermes-achievements, scan full session history
Ships @PCinkusz's hermes-achievements dashboard plugin (https://github.com/PCinkusz/hermes-achievements) as a bundled plugin at plugins/hermes-achievements/ and fixes a bug in the scan path that made the plugin only see the first 200 sessions — making lifetime badges (50k tool calls, 75k errors, etc.) unreachable on long-running installs.
Changes:
- plugins/hermes-achievements/: vendor v0.3.1 verbatim (manifest, dist/, plugin_api.py, tests, docs, README).
- plugins/hermes-achievements/dashboard/plugin_api.py:
* scan_sessions(): limit=None now scans ALL sessions via SQLite LIMIT -1. Previously capped at 200, so users with 8000+ sessions saw ~2% of their history.
* evaluate_all(): first-ever scans run in a background thread so the dashboard request path never blocks. Stale snapshots serve immediately while a background refresh runs. force=True still blocks synchronously for manual /rescan.
* _build_pending_snapshot(), _start_background_scan(), _run_scan_and_update_cache(): supporting plumbing + idempotent thread spawn.
- tests/plugins/test_achievements_plugin.py: new tests covering the 200-cap regression, the background-scan first-run flow, stale-serve-plus-background-refresh, forced sync rescan, and scan-thread idempotency.
- website/docs/user-guide/features/built-in-plugins.md: lists hermes-achievements in the bundled-plugins table and documents API endpoints, state files, and performance characteristics.
E2E validated against a real 8564-session ~6.4GB state.db:
* Cold scan: 13m 19s (one-time, backgrounded — UI never blocks)
* Warm rescan: 1.47s (8563/8564 sessions reused from checkpoint cache)
* 57/60 achievements unlocked, 3 discovered — aggregates like total_tool_calls=259958, total_errors=164213, skill_events=368243 correctly surface lifetime badges that the 200-cap made unreachable.
Original credit: @PCinkusz (MIT-licensed). Upstream repo remains the staging ground for new badges; this bundle keeps the dashboard feature parity with Hermes core changes.
* feat(achievements): publish partial snapshots during cold scan
Previously a cold scan on a large session DB (13min on 8564 sessions)
showed zero badges for the entire duration, then every badge at once
when the scan completed. A dashboard refresh mid-scan was indistinguishable
from a fresh install with no history.
Now the scanner publishes a partial snapshot to _SNAPSHOT_CACHE every
250 sessions, so each refresh during a cold scan surfaces more badges
incrementally.
Mechanism:
- scan_sessions() takes an optional progress_callback fired every
progress_every sessions with (sessions_so_far, scanned, total).
- _compute_from_scan() is extracted from compute_all() and gains an
is_partial flag that skips writing to state.json — we don't want
to record unlocked_at based on a half-complete aggregate that a
later session might rebalance.
- _run_scan_and_update_cache() installs a publisher callback that
builds a partial snapshot, marks it mode='in_progress', and writes
it to the cache with age=0 so the UI keeps polling /scan-status
and picks up the final snapshot when the scan completes.
- Manual /rescan (force=True) disables partial publishing — the
caller is blocking on the final result anyway.
E2E against real 8564-session state.db (polled cache every 10s):
t=10s: cache empty
t=20s: 250/8564 scanned, 35 unlocked, 25 discovered
t=40s: 500/8564 scanned, 42 unlocked, 18 discovered
t=60s: 1000/8564 scanned, 49 unlocked, 11 discovered
...
Tests: 9/9 pass (2 new — partial snapshot publication + no-persist-on-partial).
Upstream unittest suite: 10/10 pass.
* feat(achievements): in-progress scan banner with live % progress
Previously the dashboard showed zero badges silently during long cold
scans (13min on 8564 sessions). The backend was publishing partial
snapshots every 250 sessions, but the bundled UI didn't surface any
indicator that a scan was running — it just rendered the main page
with whatever counts were currently published and no way for the user
to know more progress was coming.
UI changes (dist/index.js, dist/style.css):
- Added a scan-in-progress banner rendered between the hero and stats
when scan_meta.mode is 'pending' or 'in_progress'. Shows:
BUILDING ACHIEVEMENT PROFILE…
Scanned 1,750 of 8,564 sessions · 20%. Badges unlock as more history streams in.
with a pulsing teal indicator and a filling teal/cyan progress bar.
Disappears the moment the backend flips to 'full' or 'incremental'.
- Added an auto-poller via useEffect — while scanInFlight is true the
page re-fetches /achievements every 4s WITHOUT toggling the loading
skeleton, so unlock counts tick up visibly without the user refreshing.
The effect cleans itself up when the scan finishes.
- Added refresh() (re-fetch, no loading flip) alongside the existing
load() (full reload, used by the Rescan button).
Attribution preserved:
- Added a header comment to index.js crediting @PCinkusz
(https://github.com/PCinkusz/hermes-achievements, MIT) as the
original author, noting the banner is a layered addition on top
of the original dist bundle.
- Matching header comment in style.css, flagging the new
.ha-scan-banner* rules as the local addition.
Live-verified end to end:
- Spun up `hermes dashboard --port 9229 --no-open` against a fresh
HERMES_HOME symlinked to the real 8564-session state.db.
- Opened /achievements in a browser, confirmed the banner renders with
live progress: 'Scanned 1,000 of 8,564 sessions · 11%' → updates to
'1,250 ... · 14%' → '1,750 ... · 20%' without user interaction,
matching the backend's partial publications.
- Stats row simultaneously climbed from 35 → 49 → 53 unlocked as
more history streamed in.
- Vision analysis of the rendered page confirms the banner styling
matches the rest of the dashboard (dark card bg, teal accent, same
small-caps typography, pulsing indicator reusing ha-pulse keyframes).
The _CODEX_AUX_MODEL constant had already rotated twice in 6 weeks
(gpt-5.3-codex -> gpt-5.2-codex -> now broken again at gpt-5.2-codex)
because ChatGPT-account Codex gates which models it accepts via an
undocumented, shifting allow-list that OpenAI publishes no changelog
for. Any pinned default will keep going stale. Issue #17533 reports
the current breakage: every ChatGPT-account auxiliary fallback fails
with HTTP 400 "model is not supported" and the 60s pause loop degrades
long sessions.
Rather than reset the clock with another stale pin (PR #17544 proposes
gpt-5.2-codex -> gpt-5.4), remove the hardcoded second-order Codex
fallback entirely:
- Delete `_CODEX_AUX_MODEL`.
- Drop `_try_codex` from `_get_provider_chain()` (the auto chain now
ends at api-key providers; 4 rungs instead of 5).
- Rename `_try_codex() -> _build_codex_client(model)` and require an
explicit model from the caller. No more guessing.
- `resolve_provider_client("openai-codex", model=None)` now warns and
returns (None, None) instead of silently guessing a stale model ID.
- Remove `_try_codex` from the `provider="custom"` fallback ladder
(same stale-constant trap).
- `_resolve_strict_vision_backend("openai-codex")` routes through
`resolve_provider_client` so the caller's explicit model is honored.
Codex-main users are unaffected: Step 1 of `_resolve_auto` already
uses `main_provider` + `main_model` directly and passes the user's
configured Codex model through `resolve_provider_client`, which never
touched `_CODEX_AUX_MODEL`. Per-task overrides (`auxiliary.<task>.provider/model`)
continue to work and are the supported way to route specific aux tasks
through Codex.
Users whose main provider fails with a payment/connection error and
who have ONLY ChatGPT-account Codex auth will now see the 60s pause
without a stale-model-rejection noise line in between -- same outcome,
cleaner failure.
Closes#17533. Supersedes #17544 (which resets the clock on the
same stale-constant problem).
Keep context-1m-2025-08-07 in OAuth requests by default so 1M-capable
subscriptions retain full context. When Anthropic rejects a request with
400 'long context beta is not yet available for this subscription',
disable the beta for the rest of the session, rebuild the client, and
retry once.
Addresses #17680 (thanks @JayGwod for the clean reproduction) without
forcing every OAuth user off the 1M context window.
Changes:
- agent/error_classifier.py: new FailoverReason.oauth_long_context_beta_forbidden;
pattern matches 400 + 'long context beta' + 'not yet available'. Narrow
enough that the existing 429 tier-gate pattern keeps its own reason.
- agent/anthropic_adapter.py: _common_betas_for_base_url,
build_anthropic_client, build_anthropic_kwargs gain drop_context_1m_beta
kwarg. Default=False (1M stays). OAuth OAUTH_ONLY_BETAS unchanged.
- agent/transports/anthropic.py: build_kwargs forwards the flag.
- run_agent.py: self._oauth_1m_beta_disabled flag, retry-once guard,
recovery branch next to the image-shrink path. _rebuild_anthropic_client
honors the flag. The main build_kwargs call site threads it through for
fast-mode extra_headers.
- hermes_cli/doctor.py, hermes_cli/models.py: sibling OAuth /v1/models
probes get the same reactive retry — previously they'd falsely report
the Anthropic API as unreachable for affected subscriptions.
Tests: 2190 tests/agent/ + 94 adjacent integration tests pass. New unit
tests cover the classifier pattern (including the collision guard against
the 429 tier-gate) and the drop_context_1m_beta adapter behavior (default
keeps 1M, flag strips only 1M while preserving every other beta).
Platform plugins shipped in-repo under plugins/platforms/ should be
available out of the box — users shouldn't have to add 'irc-platform'
to plugins.enabled before they can pick IRC from the gateway setup menu.
Adds a new ``kind: platform`` plugin type that mirrors the existing
``kind: backend`` auto-load semantics:
- Bundled (shipped in the hermes-agent repo): auto-load unconditionally.
- User-installed (~/.hermes/plugins/): still opt-in via plugins.enabled
so untrusted code doesn't silently run.
Changes:
* hermes_cli/plugins.py: add 'platform' to _VALID_PLUGIN_KINDS, document
the new kind in the PluginManifest docstring, extend the bundled auto-
load rule from 'backend only' to 'backend or platform'.
* plugins/platforms/irc/plugin.yaml: declare kind: platform.
* hermes_cli/gateway.py: remove the now-redundant
_load_bundled_platform_plugins_for_enumeration() helper and the
_enable_plugin_for_platform() helper. The setup menu's _all_platforms()
just calls discover_plugins() and reads the registry — bundled
platforms are already loaded at that point. Drops the 'needs_enable'
flag and the 'plugin disabled — select to enable' status string.
* hermes_cli/setup.py: relax the "gateway is configured" detector used
during OpenClaw migration. Switching to _platform_status() in an
earlier commit tightened the check to require an exact "configured"
match, dropping platforms whose status is "enabled, not paired",
"partially configured", "configured + E2EE", etc. Now any non-"not
configured" status counts — the user has already started setup there
and we shouldn't force the section to rerun.
* tests/hermes_cli/test_setup_irc.py: drop the TestIRCPluginDisabledFlow
class and test_configure_platform_enables_disabled_plugin_first — the
no-longer-existent flow they were testing.
* tests/hermes_cli/test_setup_openclaw_migration.py: patch both
setup.get_env_value and gateway.get_env_value in the 4 gateway-section
tests that reach _platform_status() through the unified setup flow;
switch WHATSAPP_ENABLED to the literal "true" in the registry-parity
test so WhatsApp's value-shape validator matches.
Verified via fresh-install smoke (empty plugins.enabled, no env vars):
IRC plugin loads, Platform('irc') resolves, _all_platforms() lists IRC
with status 'not configured'. 160 targeted tests pass.
feat(gateway): refine Platform._missing_ and platform-connected dispatch
Restricts plugin-name acceptance to bundled plugin scan + registry
(no arbitrary string -> enum-pollution), pulls per-platform connectivity
checks into a _PLATFORM_CONNECTED_CHECKERS lambda map with a clean
_is_platform_connected method, and adds tests covering the checker map,
plugin platform interface, and IRC setup wizard.
Nix-built hermes only copied skills/ into the output, so bundled platform
plugins weren't discoverable when running `nix run` (IRC invisible, no
plugin.yaml files present). Mirror the bundled-skills pattern:
- packages.nix: cleanSourceWith plugins/, copy to
$out/share/hermes-agent/plugins, set HERMES_BUNDLED_PLUGINS on every
wrapper.
- checks.nix: new bundled-plugins check verifying the directory, a
sample manifest, and the wrapper env var.
- hermes_cli.plugins.get_bundled_plugins_dir(): central helper that
honors HERMES_BUNDLED_PLUGINS with a dev-checkout fallback. Used by
plugins.py, plugins_cmd.py, gateway.py, and web_server.py so every
call site resolves the same path.
Merge the two gateway setup paths (hermes setup gateway + hermes gateway
setup) to use a single _unified_platforms() list that merges built-in
_PLATFORMS with dynamically registered plugin entries from
platform_registry.
- Add setup_fn field to PlatformEntry for plugin setup flows
- _unified_platforms() merges built-ins with registry entries by key
- setup_gateway() now uses unified list instead of hardcoded
_GATEWAY_PLATFORMS tuple list
- gateway_setup() uses same unified list, plugin entries appear
alongside built-ins with no [plugin] suffix
- _platform_status() handles plugin platforms via registry check_fn
- Plugin platforms with setup_fn get called directly; plugins without
get a generic env-var display fallback
IRC and other plugin platforms now appear automatically in the setup
menu when registered via platform_registry.register().
feat(gateway): surface disabled platform plugins in setup and auto-enable on select
Platform plugins under plugins/platforms/* (IRC, etc.) were gated behind
plugins.enabled, so `hermes gateway setup` wouldn't list them until the
user ran `hermes plugins enable <name>` first. Now the setup menu always
surfaces them as "plugin disabled — select to enable", and picking one
adds it to plugins.enabled before running its setup flow.
Along the way, unify the two gateway setup flows so `hermes setup gateway`
and `hermes gateway setup` both read from the same platform list (built-in
_PLATFORMS + platform_registry entries), dispatch through a single
_configure_platform() helper, and share _platform_status(). Deletes the
dead bespoke wrappers in setup.py (_setup_whatsapp, _setup_weixin,
_setup_email, etc.) that duplicated logic now covered by the registry
path or _setup_standard_platform.
Also:
- PlatformEntry gains a plugin_name field so the registry knows which
plugin owns each entry (required for auto-enable).
- PluginContext.register_platform auto-stamps plugin_name from the
manifest so plugins don't have to pass it explicitly.
- PluginManager now scans plugins/platforms/* as its own category root,
one level below the bundled plugin scan.
- Fix IRC plugin discovery: rename PLUGIN.yaml → plugin.yaml (the
scanner is case-sensitive) and add the missing __init__.py that
_load_directory_module requires.
Plugin platforms now get full toolset support without any entries in
toolsets.py.
tools_config._get_platform_tools(): Falls back to 'hermes-<name>'
when the platform isn't in the static PLATFORMS dict. No more
KeyError for plugin platforms.
toolsets.resolve_toolset(): Auto-generates a toolset for plugin
platforms (hermes-<name>) containing _HERMES_CORE_TOOLS plus any
tools the plugin registered into a matching toolset name. This means
a plugin can call ctx.register_tool(toolset='irc', ...) and those
tools will be included in the hermes-irc toolset automatically.
webhook.py: Registry-aware cross-platform delivery.
run_agent.py: Platform hints from plugin registry.
IRC adapter: Token lock + platform hint.
Removed dead token-empty-warning extension.
Updated docs.
Closes remaining functional gaps and adds documentation.
webhook.py: Cross-platform delivery now checks the plugin registry
for unknown platform names instead of hardcoding 15 names in a tuple.
Plugin platforms can receive webhook-routed deliveries.
prompt_builder: Platform hints (system prompt LLM guidance) now fall
back to the plugin registry's platform_hint field. Plugin platforms
can tell the LLM 'you're on IRC, no markdown.'
PlatformEntry: Added platform_hint field for LLM guidance injection.
IRC adapter: Added acquire_scoped_lock/release_scoped_lock in
connect/disconnect to prevent two profiles from using the same IRC
identity. Added platform_hint for IRC-specific LLM guidance.
Removed dead token-empty-warning extension for plugin platforms
(plugin adapters handle their own env vars via check_fn).
website/docs/developer-guide/adding-platform-adapters.md:
- Added 'Plugin Path (Recommended)' section with full code examples,
PLUGIN.yaml template, config.yaml examples, and a table showing all
18 integration points the plugin system handles automatically
- Renamed built-in checklist to clarify it's for core contributors
gateway/platforms/ADDING_A_PLATFORM.md:
- Added Plugin Path section pointing to the reference implementation
and full docs guide
- Clarified built-in path is for core contributors only
PII redaction: build_session_context_prompt() now checks the plugin
registry's pii_safe flag in addition to the hardcoded _PII_SAFE_PLATFORMS
frozenset. Plugin platforms that set pii_safe=True (e.g. phone-based
messaging bridges) get their user IDs redacted before LLM context.
Token empty warnings: the empty-token diagnostic at config load now
checks the plugin registry's required_env when a platform isn't in the
hardcoded _token_env_names dict. Catches 'enabled but empty' for
plugin platforms too.
Extends the platform plugin interface from Phase 1 to cover every
touchpoint where built-in platforms have hardcoded behavior.
- allowed_users_env / allow_all_env: per-platform auth env vars
- max_message_length: smart-chunking for send_message tool
- pii_safe: session PII redaction flag
- emoji: CLI/gateway display
- allow_update_command: /update access control
send_message tool (tools/send_message_tool.py):
- Replaced hardcoded platform_map dict with Platform() call
- Added _send_via_adapter() for plugin platforms — routes through
live gateway adapter when available
- Registry-aware max message length for smart chunking
Cron delivery (cron/scheduler.py):
- Replaced hardcoded 15-entry platform_map with Platform() call
- Plugin platforms now work as cron delivery targets
User authorization (gateway/run.py _is_user_authorized):
- Registry fallback: checks PlatformEntry.allowed_users_env and
allow_all_env when platform not in hardcoded maps
- Plugin platforms get per-platform auth support
_UPDATE_ALLOWED_PLATFORMS: checks registry allow_update_command flag
Channel directory: includes plugin platforms in session enumeration
Orphaned config warning: descriptive message when plugin platform is
in config but no plugin registered it
Gateway weakref: _gateway_runner_ref for cross-module adapter access
hermes status: shows plugin platforms with (plugin) tag
hermes gateway setup: plugin platforms appear in menu with setup hints
hermes_cli/platforms.py: get_all_platforms() merges with registry,
platform_label() falls back to registry for plugin names
- 8 new tests (extended fields, cron resolution, platforms merge)
- Updated 3 tests for new Platform() based resolution
- 2829 passed, 24 pre-existing failures, zero new failures
Adds a platform adapter plugin interface so anyone can create new gateway
platforms (IRC, Viber, Line, etc.) as drop-in plugins without modifying
core gateway code.
- PlatformEntry dataclass: name, label, adapter_factory, check_fn,
validate_config, required_env, install_hint, source
- PlatformRegistry singleton with register/unregister/create_adapter
- _create_adapter() in gateway/run.py checks registry first, falls
through to existing if/elif chain for built-in platforms
- Platform._missing_() accepts unknown string values, creating cached
pseudo-members so Platform('irc') is Platform('irc') holds true
- GatewayConfig.from_dict() now parses plugin platform names from
config.yaml without rejecting them
- get_connected_platforms() delegates to registry for unknown platforms
- PluginContext.register_platform() for plugin authors
- Mirrors the existing register_tool() / register_hook() pattern
- Full async IRC adapter using stdlib asyncio (zero external deps)
- Connects via TLS, handles PING/PONG, nick collision, NickServ auth
- Channel messages require addressing (nick: msg), DMs always dispatch
- Markdown stripping for IRC-clean output, message splitting for
512-byte line limit
- Config via config.yaml extra dict or IRC_* env vars
- Platform enum dynamic members (identity stability, case normalization)
- PlatformRegistry (register, unregister, create, validation, factory)
- GatewayConfig integration (from_dict parsing, get_connected_platforms)
- IRC adapter (init, send, protocol parsing, markdown, requirements)
No existing platform adapters were migrated — the if/elif chain is
untouched. This is Phase 1: prove the interface with a real plugin.
Reloading MCP servers rebuilds the tool set for the active session, which
invalidates the provider prompt cache (tool schemas are baked into the
system prompt). The next message re-sends full input tokens — can be
expensive on long-context or high-reasoning models.
To surface that cost, /reload-mcp now routes through a new slash-confirm
primitive with three options: Approve Once / Always Approve / Cancel.
'Always Approve' persists approvals.mcp_reload_confirm: false so future
reloads run silently.
Coverage:
* Classic CLI (cli.py) — interactive numbered prompt.
* TUI (tui_gateway + Ink ops.ts) — text warning on first call; `now` /
`always` args skip the gate; `always` also persists the opt-out.
* Messenger gateway — button UI on Telegram (inline keyboard), Discord
(discord.ui.View), Slack (Block Kit actions); text fallback on every
other platform via /approve /always /cancel replies intercepted in
gateway/run.py _handle_message.
* Config key: approvals.mcp_reload_confirm (default true).
* Auto-reload paths (CLI file watcher, TUI config-sync mtime poll) pass
confirm=true so they do NOT prompt.
Implementation:
* tools/slash_confirm.py — module-level pending-state store used by all
adapters and by the CLI prompt. Thread-safe register/resolve/clear.
* gateway/platforms/base.py — send_slash_confirm hook (default 'Not
supported' → text fallback).
* gateway/run.py — _request_slash_confirm helper + text intercept in
_handle_message (yields to in-progress tool-exec approvals so
dangerous-command /approve still unblocks the tool thread first).
Tests:
* tests/tools/test_slash_confirm.py — primitive lifecycle + async
resolution + double-click atomicity (16 tests).
* tests/hermes_cli/test_mcp_reload_confirm_gate.py — default-config
shape + deep-merge preserves user opt-out (5 tests).
Targeted runs (hermetic): 89 passed (slash-confirm, config gate,
existing agent cache, existing telegram approval buttons).
Each AIAgent.__init__() was unconditionally starting a daemon thread to
pre-warm the OpenRouter model metadata cache. In gateway mode a new
AIAgent is created for every incoming message, so one OS thread leaked
per request. After ~1 000 messages the process hit the Linux thread
limit and raised RuntimeError: can't start new thread for all subsequent
requests.
Add a module-level threading.Event (_openrouter_prewarm_done) that is
set before the thread is started. Subsequent AIAgent instantiations
skip the spawn entirely; fetch_model_metadata() is cached for 1 hour so
the single background call is sufficient.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
PR #15027 (5 days ago) shipped TELEGRAM_GROUP_ALLOWED_USERS as a chat-ID
allowlist. #17686 correctly renames that to sender user IDs and moves
chat IDs to TELEGRAM_GROUP_ALLOWED_CHATS. Without a shim, any user on
PR #15027's guidance would silently start rejecting group traffic on
upgrade.
- gateway/run.py: in _is_user_authorized, if TELEGRAM_GROUP_ALLOWED_USERS
contains values starting with '-' (chat-ID-shaped), honor them as chat
IDs and log a one-shot deprecation warning pointing users at the new
TELEGRAM_GROUP_ALLOWED_CHATS var.
- tests/gateway/test_unauthorized_dm_behavior.py: three new tests cover
legacy chat-ID values authorizing the listed chat, not crossing to
other chats, and mixed sender/chat values in the same var.
- website/docs/user-guide/messaging/telegram.md: rewrite the Group
Allowlisting section to document the new user/chat split + migration
note. Remove stale '/thread_id' suffix claim (code never parsed it).
- website/docs/reference/environment-variables.md: document all three
Telegram allowlist env vars.
Salvage-follow-up to @shannonsands's /reload-skills PR. Trims the feature to
match the design: user-initiated rescan, no prompt-cache reset, no new
schema surface, no phantom user turn, and the next-turn note carries each
added/removed skill's 60-char description (not just its name).
Changes vs the original PR:
* Drop the in-process skills prompt-cache clear in reload_skills(). Skills
are invoked at runtime via /skill-name, skills_list, or skill_view —
they don't need to live in the system prompt for the model to use them.
Keeping the cache intact preserves prefix caching across the reload so
/reload-skills pays no cache-reset cost. (MCP has to break the cache
because tool schemas must be known at conversation start; skills do not.)
* Drop the skills_reload agent tool and SKILLS_RELOAD_SCHEMA from
tools/skills_tool.py, plus the four skills_reload enumerations in
toolsets.py. No new schema surface — agents can already see a freshly-
installed skill via skill_view / skills_list the moment it's on disk.
* Replace the phantom 'role: user' turn injection with a one-shot queued
note. CLI uses self._pending_skills_reload_note (same pattern as
_pending_model_switch_note, prepended to the next API call and cleared).
Gateway uses self._pending_skills_reload_notes[session_key]. The note
is prepended to the NEXT real user message in this session, so message
alternation stays intact and nothing out-of-band is persisted to the
transcript.
* reload_skills() now returns added/removed as
[{'name': str, 'description': str}, ...] (description truncated to 60
chars — matches the curator / gateway adapter budget). The injected
next-turn note formats each entry as 'name — description' so the model
can actually reason about which new skills to call without running
skills_list first.
* Only emit the note when the diff is non-empty. On empty diff, print
'No new skills detected' and do nothing else.
* Tests rewritten to cover the queue semantics, the description payload,
and a regression guard that the prompt-cache snapshot is preserved.
Adds a public reload path for the in-process skill caches so newly
installed (or removed) skills become visible mid-session without a
gateway restart. Mirrors the shape of /reload-mcp.
Three surfaces:
* /reload-skills slash command — CLI (cli.py) and gateway (gateway/run.py),
with /reload_skills alias for Telegram autocomplete and an explicit
Discord registration.
* skills_reload agent tool (tools/skills_tool.py) — lets agents/subagents
pick up freshly-installed skills via tool call.
* agent.skill_commands.reload_skills() — shared helper that clears
_skill_commands, _SKILLS_PROMPT_CACHE (in-process LRU), and the
on-disk .skills_prompt_snapshot.json, then returns an added/removed
diff plus the new total count.
Tested:
* tests/agent/test_skill_commands_reload.py (9 cases)
* tests/cli/test_cli_reload_skills.py (3 cases)
* tests/gateway/test_reload_skills_command.py (4 cases)
Use case: NemoClaw / OpenShell-style sandboxed orchestrators that drop
skills into ~/.hermes/skills mid-session, plus agentic flows where the
agent itself installs a skill via the shell tool and needs it bound
without a gateway restart. The Python helper
clear_skills_system_prompt_cache(clear_snapshot=True) already exists
internally — this PR just exposes it via slash command and tool.
- SQL: add `model != ''` to both queries in /api/analytics/models so
sessions with empty-string model (pre-existing data integrity,
confirmed in production DB: ~107 sessions) no longer render as
blank-header cards.
- ModelsPage: drop the arbitrary slashIdx < 20 length gate in
shortModelName / modelProvider. The gate was fragile for longer
vendor prefixes (e.g. `deepseek-ai/...`). Strip on the first /
unconditionally. Rename modelProvider -> modelVendor to avoid
confusion with the billing provider column.
- scripts/release.py: add AUTHOR_MAP entry for yatesjalex.
- New /models page in left nav (after Analytics)
- New /api/analytics/models endpoint with per-model token/cost/session
breakdown, cache read/reasoning tokens, tool calls, avg tokens/session,
and capabilities from models.dev (vision/tools/reasoning/context window)
- Model cards with stacked token distribution bar, capability badges,
provider badges, cost info, and relative time
- Summary stats bar (models used, total tokens, est. cost, sessions)
- Period selector (7d/30d/90d) with refresh
- i18n support (en + zh)
Broad drift audit against origin/main (b52b63396).
Reference pages (most user-visible drift):
- slash-commands: add /busy, /curator, /footer, /indicator, /redraw, /steer
that were missing; drop non-existent /terminal-setup; fix /q footnote
(resolves to /queue, not /quit); extend CLI-only list with all 24
CLI-only commands in the registry
- cli-commands: add dedicated sections for hermes curator / fallback /
hooks (new subcommands not previously documented); remove stale
hermes honcho standalone section (the plugin registers dynamically
via hermes memory); list curator/fallback/hooks in top-level table;
fix completion to include fish
- toolsets-reference: document the real 52-toolset count; split browser
vs browser-cdp; add discord / discord_admin / spotify / yuanbao;
correct hermes-cli tool count from 36 to 38; fix misleading claim
that hermes-homeassistant adds tools (it's identical to hermes-cli)
- tools-reference: bump tool count 55 -> 68; add 7 Spotify, 5 Yuanbao,
2 Discord toolsets; move browser_cdp/browser_dialog to their own
browser-cdp toolset section
- environment-variables: add 40+ user-facing HERMES_* vars that were
undocumented (--yolo, --accept-hooks, --ignore-*, inference model
override, agent/stream/checkpoint timeouts, OAuth trace, per-platform
batch tuning for Telegram/Discord/Matrix/Feishu/WeCom, cron knobs,
gateway restart/connect timeouts); dedupe the Cron Scheduler section;
replace stale QQ_SANDBOX with QQ_PORTAL_HOST
User-guide (top level):
- cli.md: compression preserves last 20 turns, not 4 (protect_last_n: 20)
- configuration.md: display.platforms is the canonical per-platform
override key; tool_progress_overrides is deprecated and auto-migrated
- profiles.md: model.default is the config key, not model.model
- sessions.md: CLI/TUI session IDs use 6-char hex, gateway uses 8
- checkpoints-and-rollback.md: destructive-command list now matches
_DESTRUCTIVE_PATTERNS (adds rmdir, cp, install, dd)
- docker.md: the container runs as non-root hermes (UID 10000) via
gosu; fix install command (uv pip); add missing --insecure on the
dashboard compose example (required for non-loopback bind)
- security.md: systemctl danger pattern also matches 'restart'
- index.md: built-in tool count 47 -> 68
- integrations/index.md: 6 STT providers, 8 memory providers
- integrations/providers.md: drop fictional dashscope/qwen aliases
Features:
- overview.md: 9 image models (not 8), 9 TTS providers (not 5),
8 memory providers (Supermemory was missing)
- tool-gateway.md: 9 image models
- tools.md: extend common-toolsets list with search / messaging /
spotify / discord / debugging / safe
- fallback-providers.md: add 6 real providers from PROVIDER_REGISTRY
(lmstudio, kimi-coding-cn, stepfun, alibaba-coding-plan,
tencent-tokenhub, azure-foundry)
- plugins.md: Available Hooks table now includes on_session_finalize,
on_session_reset, subagent_stop
- built-in-plugins.md: add the 7 bundled plugins the page didn't
mention (spotify, google_meet, three image_gen providers, two
dashboard examples)
- web-dashboard.md: add --insecure and --tui flags
- cron.md: hermes cron create takes positional schedule/prompt, not
flags
Messaging:
- telegram.md: TELEGRAM_WEBHOOK_SECRET is now REQUIRED when
TELEGRAM_WEBHOOK_URL is set (gateway refuses to start without it
per GHSA-3vpc-7q5r-276h). Biggest user-visible drift in the batch.
- discord.md: HERMES_DISCORD_TEXT_BATCH_SPLIT_DELAY_SECONDS default
is 2.0, not 0.1
- dingtalk.md: document DINGTALK_REQUIRE_MENTION /
FREE_RESPONSE_CHATS / MENTION_PATTERNS / HOME_CHANNEL /
ALLOW_ALL_USERS that the adapter supports
- bluebubbles.md: drop fictional BLUEBUBBLES_SEND_READ_RECEIPTS env
var; the setting lives in platforms.bluebubbles.extra only
- qqbot.md: drop dead QQ_SANDBOX; add real QQ_PORTAL_HOST and
QQ_GROUP_ALLOWED_USERS
- wecom-callback.md: replace 'hermes gateway start' (service-only)
with 'hermes gateway' for first-time setup
Developer-guide:
- architecture.md: refresh tool/toolset counts (61/52), terminal
backend count (7), line counts for run_agent.py (~13.7k), cli.py
(~11.5k), main.py (~10.4k), setup.py (~3.5k), gateway/run.py
(~12.2k), mcp_tool.py (~3.1k); add yuanbao adapter, bump platform
adapter count 18 -> 20
- agent-loop.md: run_agent.py line count 10.7k -> 13.7k
- tools-runtime.md: add vercel_sandbox backend
- adding-tools.md: remove stale 'Discovery import added to
model_tools.py' checklist item (registry auto-discovery)
- adding-platform-adapters.md: mark send_typing / get_chat_info as
concrete base methods; only connect/disconnect/send are abstract
- acp-internals.md: ACP sessions now persist to SessionDB
(~/.hermes/state.db); acp.run_agent call uses
use_unstable_protocol=True
- cron-internals.md: gateway runs scheduler in a dedicated background
thread via _start_cron_ticker, not on a maintenance cycle; locking
is cross-process via fcntl.flock (Unix) / msvcrt.locking (Windows)
- gateway-internals.md: gateway/run.py ~12k lines
- provider-runtime.md: cron DOES support fallback (run_job reads
fallback_providers from config)
- session-storage.md: SCHEMA_VERSION = 11 (not 9); add migrations
10 and 11 (trigram FTS, inline-mode FTS5 re-index); add
api_call_count column to Sessions DDL; document messages_fts_trigram
and state_meta in the architecture tree
- context-compression-and-caching.md: remove the obsolete 'context
pressure warnings' section (warnings were removed for causing
models to give up early)
- context-engine-plugin.md: compress() signature now includes
focus_topic param
- extending-the-cli.md: _build_tui_layout_children signature now
includes model_picker_widget; add to default layout
Also fixed three pre-existing broken links/anchors the build warned
about (docker.md -> api-server.md, yuanbao.md -> cron-jobs.md and
tips#background-tasks, nix-setup.md -> #container-aware-cli).
Regenerated per-skill pages via website/scripts/generate-skill-docs.py
so catalog tables and sidebar are consistent with current SKILL.md
frontmatter.
docusaurus build: clean, no broken links or anchors.
Self-review caught several errors in the previous commit:
Frontmatter
- Replace non-standard `requires_runtime` / `requires_tooling` fields with
the documented `compatibility:` field (parsed by tools/skills_tool.py).
- Drop the `audit-v5` author tag I added unnecessarily.
MODEL_LOADERS catalog
- Remove `IPAdapterUnifiedLoader` (input `preset` is an enum, not a file).
- Remove `IPAdapterInsightFaceLoader` and `InsightFaceLoader` (input
`provider` is a GPU backend selector, not a model file). These would have
flagged enum values like "STANDARD" or "CUDA" as missing model files.
- Add "NB:" comment explaining `BasicGuider` has no `cfg` input
(the original PARAM_PATTERNS entry would never have matched).
- Remove `SamplerCustomAdvanced.noise_seed` from PARAM_PATTERNS — that
node takes a NOISE input from RandomNoise, not a seed field directly.
NODE_TO_PACKAGE registry slugs
- Verified all 18 packages against api.comfy.org and fixed:
- `comfyui-essentials` → `comfyui_essentials` (underscore, not hyphen)
- `comfyui-gguf` → `ComfyUI-GGUF` (case-sensitive)
- `comfyui-photomaker-plus` → `ComfyUI-PhotoMaker-Plus`
- `comfyui-wanvideowrapper` → `ComfyUI-WanVideoWrapper`
- ComfyUI-HunyuanVideoWrapper isn't on the registry; surface a git-URL
install hint via new NODE_TO_GIT_URL fallback so the user can install
via ComfyUI-Manager's /manager/queue/install endpoint.
Wrong class names
- `Canny` → `CannyEdgePreprocessor` (controlnet-aux registers the latter,
the former never appears in /object_info).
- Add `Zoe_DepthAnythingPreprocessor` and `AnimalPosePreprocessor` while
fixing controlnet-aux.
- Remove `Reroute (rgthree)` (rgthree's Reroute is JS-only — no Python
class, never appears in /object_info).
- Add `Display Int (rgthree)` (sibling of Display Any).
- Move `UltralyticsDetectorProvider` from `comfyui-impact-pack` to
`comfyui-impact-subpack` (separate package, registered there).
Tests
- Update test_packages_are_safe_for_shell to accept case-mixed slugs (the
registry uses both ComfyUI- and comfyui_ prefixes inconsistently). Replaced
the lowercase-only assertion with a shell-safe regex check.
- 117 tests still pass (105 unit + 8 cloud + 4 cross-host).
Attribution
- Add `SHL0MS@users.noreply.github.com` mapping to scripts/release.py
AUTHOR_MAP so check-attribution CI passes.
The audit of v4.1 surfaced ~70 issues across the five scripts and three
reference docs — most user-visible (silent file overwrites, status-error
misclassified as success, X-API-Key leaked to S3 on /api/view redirect,
Cloud endpoints that 404 because they were renamed). v5.0.0 fixes those
and fills the gaps that previously forced users to write their own glue
(WebSocket monitoring, batch/sweep, img2img upload helper, dep auto-fix,
log fetch, health check, example workflows).
Critical fixes
- run_workflow.py: poll_status now checks status_str==error BEFORE
completed:true, so a failed run no longer reports success
- run_workflow.py: download_output streams to disk via safe_path_join,
preserves server subfolder structure (no silent overwrites), and
retries with exponential backoff
- run_workflow.py: refuses to overwrite a link with a literal in
inject_params (would silently break wiring)
- _common.py: _StripSensitiveOnRedirectSession (subclasses
requests.Session.rebuild_auth) drops X-API-Key/Cookie on cross-host
redirects — fixes a real key-leak path through Cloud's signed-URL
download flow. Tested
- Cloud routing (verified live): /history → /history_v2,
/models/<f> → /experiment/models/<f>, plus folder aliases for the
unet ↔ diffusion_models and clip ↔ text_encoders rename
- check_deps.py: distinguishes 200/empty vs 404 folder_not_found vs
403 free-tier; emits concrete fix_command per missing dep
- extract_schema.py: prompt vs negative_prompt determined by tracing
KSampler.{positive,negative} connections (incl. through Reroute /
Primitive nodes) instead of meta-title heuristic; symmetric
duplicate-name resolution; cycle-safe trace_to_node
- hardware_check.py: multi-GPU pick-best, Apple variant detection,
Rosetta detection, WSL2, ROCm --json, disk-space check, optional
PyTorch probe; powershell preferred over deprecated wmic
- comfyui_setup.sh: prefers pipx → uvx → pip --user (with PEP-668
fallback); idempotent — skips relaunch if server already up;
configurable port/workspace; persistent log; SIGINT trap
New scripts
- run_batch.py — count or sweep (cartesian product), parallel up to
cloud tier limit
- ws_monitor.py — real-time WebSocket viewer; saves preview frames
- auto_fix_deps.py — runs comfy node install / model download for
whatever check_deps reports missing (with --dry-run)
- health_check.py — single command that runs the verification checklist
(comfy-cli + server + checkpoints + optional smoke test that cancels
itself to avoid burning compute)
- fetch_logs.py — pull traceback / status messages for a prompt_id
Coverage expansion
- Param patterns now cover Flux (BasicScheduler, BasicGuider,
RandomNoise, ModelSamplingFlux), SD3, Wan/Hunyuan/LTX video,
IPAdapter, rgthree, easy-use, AnimateDiff
- Embedding refs in CLIPTextEncode strings extracted as model deps
- ckpt_name / vae_name / lora_name / unet_name now controllable so
workflows can be retargeted per run
Examples
- workflows/{sd15,sdxl,flux_dev}_txt2img.json
- workflows/sdxl_{img2img,inpaint}.json
- workflows/upscale_4x.json
- workflows/{animatediff_video,wan_video_t2v}.json + README
Tests
- 117 tests (105 unit + 8 cloud integration + 4 cross-host security)
- Cloud tests auto-skip without COMFY_CLOUD_API_KEY; verified end-to-end
against live cloud API
Backwards compatibility
- All existing CLI flags continue to work; new behavior is opt-in
(--ws, --input-image, --randomize-seed, --flat-output, etc.)
Pull the top-level + chat parser construction out of main() into
hermes_cli/_parser.py so relaunch.py can introspect parser._actions to
discover which flags exist and whether they take values, instead of
maintaining a parallel hand-rolled (flag, takes_value) tuple list.
- _parser.py: build_top_level_parser() returns (parser, subparsers,
chat_parser); side-effect-free import.
- main.py: ~290 lines of inline parser construction collapsed to a
helper call. Other subparsers stay inline (dispatch is bound to
module-level cmd_* functions).
- _parser._inherited_flag(parser, ...): wraps parser.add_argument and
sets action.inherit_on_relaunch = True. Used in place of
parser.add_argument for the 25 flags (top-level + chat) that need to
carry over.
- _parser.PRE_ARGPARSE_INHERITED_FLAGS: holds --profile/-p, which
isn't on argparse (consumed earlier by main._apply_profile_override).
- relaunch.py: drops _CRITICAL_DESTS and _PRE_ARGPARSE_FLAGS; the table
builder now filters by getattr(action, 'inherit_on_relaunch', False).
- test_ignore_user_config_flags.py: brittle inspect.getsource grep
replaced with proper parser introspection.
- test_relaunch.py: introspection sanity tests added.
Salvaged from PR #17549; added top-level -t/--toolsets flag to
_parser.py so #17623 (fix(tui): honor launch toolsets) behavior is
preserved on current main.
Co-authored-by: ethernet <arilotter@gmail.com>
Extract all os.execvp('hermes', ...) calls into a utility so flags like
--tui, --dev, --profile, --model, --provider, et al. survive session
resume and post-setup relaunch.
- resolve_hermes_bin: prefers sys.argv[0] when callable, then PATH,
then falls back to '${sys.executable} -m hermes_cli.main' (fixes nix
run relaunches)
- build_relaunch_argv: allowlists critical flags so they carry over
- cmd_sessions browse now calls relaunch(['--resume', <id>])
- _apply_profile_override skips redundant work when HERMES_HOME is
already set (child inherits parent profile)
- setup.py replaces _resolve_hermes_chat_argv with relaunch_chat()
- added comprehensive tests for flag extraction and binary resolution
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
If a concurrent RPC mutates _sessions while session.delete is iterating
it (e.g. a parallel session.create on the thread pool), the bare except
swallowed the RuntimeError and let the delete proceed against a row
that may still be live. Snapshot via list(_sessions.values()) and
return an error when even that raises, instead of treating "couldn't
check" as "no active sessions."
Single-key confirm matches how the picker already accepts 1-9 to
resume — no separate y/n keymap to learn — and "press d again" is
self-documenting next to the cursor.
Pressing `d` on the highlighted row in the resume picker prompts
`delete? y/n`; `y` deletes the session (DB row + on-disk transcript
files), anything else cancels. The active session is excluded from
deletion server-side.
Adds a new `session.delete` JSON-RPC handler that wraps
`SessionDB.delete_session`, forwarding the per-profile `sessions/`
directory so transcripts get cleaned up alongside the row.
vision_analyze used Path('./temp_vision_images') — a relative path that
resolved against cwd. Under Docker the image's WORKDIR is /opt/hermes,
which is root-owned and only chmoded a+rX (read + traversal). Since
#5811 landed (run as non-root hermes UID 10000, Apr 12), remote-URL
vision calls fail with PermissionError on mkdir.
Switch to get_hermes_dir('cache/vision', 'temp_vision_images'): resolves
to $HERMES_HOME/cache/vision/ (= /opt/data/cache/vision/ in Docker —
the user-owned volume mount). Existing installs with the old dir keep
using it via the get_hermes_dir back-compat path; no migration needed.
Only site in the codebase that stored runtime files via Path('./...').
Reported via Discord: https://juick.com/i/p/3089079.jpg → Telegram →
gateway → [Errno 13] Permission denied: 'temp_vision_images'.
CI Tests workflow has been red on main for 40+ consecutive runs. This
commit recovers every failure visible in run 25130722163 (most recent
completed run prior to this PR).
Root causes, by group:
Test-mock drift after product landed (fix: update mocks)
- test_mcp_structured_content / test_mcp_dynamic_discovery (6 tests):
product added _rpc_lock (#02ae15222) and _schedule_tools_refresh
(#1350d12b0) without updating sibling test files. Install a real
asyncio.Lock inside the fake run-loop and patch at _schedule_tools_refresh.
- test_session.py: renamed normalize_whatsapp_identifier → canonical_
whatsapp_identifier upstream; keep a local alias so the legacy tests
keep working.
- test_run_progress_topics Slack DM test: PR #8006 made Slack default
tool_progress=off; explicitly set it to 'all' in the test fixture so
the progress-callback path still runs. Also read tool_progress_callback
at call time rather than freezing it in FakeAgent.__init__ — production
assigns it AFTER construction.
- test_tui_gateway_server session-create/close race: session.create now
defers _start_agent_build behind a 50ms timer — wait for the build
thread to enter _make_agent before closing, otherwise the orphan-
cleanup path never runs.
- test_protocol session.resume: product get_messages_as_conversation now
takes include_ancestors kwarg; accept **_kwargs in the test stub.
- test_copilot_acp_client redaction: redactor is OFF by default (snapshots
HERMES_REDACT_SECRETS at import); patch agent.redact._REDACT_ENABLED=True
for the duration of the test.
- test_minimax_provider: after #17171, dots in non-Anthropic model names
stay dots even with preserve_dots=False. Assert the new invariant
rather than the old 'broken for MiniMax' behavior.
- test_update_autostash: updater now scans `ps -A` for dashboard PIDs;
the test's catch-all subprocess.run stub needed stdout/stderr fields.
- test_accretion_caps: read_timestamps dict is populated lazily when
os.path.getmtime succeeds. Use .get("read_timestamps", {}) to tolerate
CI filesystems where the stat races file creation.
Change-detector tests (fix: rewrite as structural invariants)
- test_credential_sources_registry_has_expected_steps: was a frozen set
comparison that broke when minimax-oauth was added. Rewrite as an
invariant check (every step has description, no dupes, core steps
present) per AGENTS.md 'don't write change-detector tests'.
xdist ordering / test pollution (fix: reset state, use module-local patches)
- test_setup vercel: sibling test saved VERCEL_PROJECT_ID='project' to
os.environ via save_env_value() and never cleared it. monkeypatch.delenv
the VERCEL_* vars in the link-file test.
- test_clipboard TestIsWsl: GitHub Actions is on Azure VMs whose real
/proc/version often contains 'microsoft'. Patching builtins.open with
mock_open didn't reliably intercept hermes_constants.is_wsl's call in
xdist workers that had already cached _wsl_detected=True from an
earlier test. Patch hermes_constants.open directly and add
teardown_method to reset the cache after each test.
Pytest-asyncio cancellation hangs (fix: bound product await with timeout)
- test_session_split_brain_11016 (3 params) + test_gateway_shutdown
cancel-inflight: under pytest-asyncio 1.3.0, 'await task' and
'asyncio.gather(cancelled_tasks)' can stall for 30s when the cancelled
task's finally block awaits typing-task cleanup. Bound both with
asyncio.wait_for(..., timeout=5.0) and asyncio.shield — the stragglers
are released from adapter tracking and allowed to finish unwinding in
the background. This is also a legitimate hardening: a wedged finally
shouldn't stall the caller's dispatch or a gateway shutdown.
Orphan UI config (fix: merge tiny tab into messaging category)
- test_web_server test_no_single_field_categories: the telegram.reactions
config field lived in its own 'telegram' schema category with no
siblings. Fold it under 'discord' via _CATEGORY_MERGE so the dashboard
doesn't render an orphan single-field tab.
Local verification: 38/38 originally-failing tests pass; 4044/4044
gateway tests pass; 684/684 targeted subset (all 16 touched test files)
passes.
Reset sticky mouse/focus/paste terminal modes before the TUI starts and during graceful shutdown paths so stale tab state from prior crashes cannot poison the next session.
Detect leaked SGR mouse-report fragments in CLI input, strip them, and reset terminal modes in-place so scroll and typing recover without reopening the tab. Add regression tests for escaped, visible, and bare leak forms.
Route Option/Alt or Ctrl wheel input through a gated precision path that scrolls at most one row per short interval, while preserving the existing accelerated behavior for plain wheel input. Keep precision active briefly after modifier release so queued wheel events from the same gesture do not jump into acceleration mid-stream.
curl is a ubiquitous tool both for users running ad-hoc commands inside
the container (debugging, health checks, quick HTTP probes) and for
agent workflows — many bundled skills and hub skills lean on curl for
HTTP calls, API exploration, and installer bootstrapping. Its absence
causes silent workflow failures with "curl: command not found" until
the user manually apt-installs it.
Add curl to the single apt-get install layer alongside the other base
utilities (build-essential, nodejs, git, openssh-client, etc.) so it
ships in the image with zero extra layers and negligible size impact
(~400 KB).
- Dockerfile: add curl to the apt-get install list
Decode Shift, Meta, and Ctrl bits from SGR and legacy X10 wheel event button bytes so TUI input handlers can distinguish modified wheel gestures from plain scrolling.
check_for_updates() looked at __file__.parent.parent for a .git dir to
diff against origin/main. A nix-built hermes lives in /nix/store with
no .git there, so the check fell through to whatever editable-install
dev checkout last populated ~/.hermes/.update_check, producing stale
"X commits behind" warnings right after a fresh `nix run --refresh`.
Embed the locked flake rev into the wrapper as HERMES_REVISION (only
on
clean builds — dirty refs don't represent any upstream commit). When
set, banner.py compares it to upstream main via `git ls-remote`
instead
of inspecting a local checkout, and the cache key includes the rev so
nix updates invalidate immediately. Without local history we can't
count commits, so the message is a plain "update available" with no
suggested command — nix users may install via `nix run`, profile,
system flake, or home-manager, and we don't know which.
Also bump web/package-lock.json npmDepsHash via `nix run
.#fix-lockfiles`.
* fix(tui): offload manual compaction RPC
Route TUI session compression through the existing long-handler pool so slow compaction does not block other gateway RPCs.
* fix(tui): show compaction progress immediately
Print a local status line before the compress RPC starts so slow manual compaction does not look like a no-op.
* feat(tui): rich /compress feedback parity with CLI
Show pre-compaction message count and rough token estimate immediately, emit a status update so the bottom bar reflects ongoing compaction, and report a multi-line summary (headline + token delta + optional note) using the shared summarize_manual_compression helper.
* fix(tui): show live compaction estimate in transcript
Mirror compression progress status into the transcript so users see the backend message count and token estimate while /compress is still running.
* fix(tui): single live compaction line with spinner glyph
Drop the redundant local "compressing context..." placeholder and prefix the live backend status line with a braille spinner glyph so /compress reads as a single in-progress row.
* fix(tui): address review nits on /compress feedback
Reuse the precomputed token estimate inside _compress_session_history so the gateway does not redo the O(n) work while holding history_lock, keep the status bar pinned during long manual compactions instead of auto-restoring after 4s, and drop the redundant noop bullet that doubled with the system role glyph.
* fix(tui): release history_lock during compaction LLM call
Move the snapshot/commit pattern into _compress_session_history so the lock is held only across the in-memory bookkeeping, not during agent._compress_context. Also emit a final neutral status update from session.compress so the pinned compressing indicator clears even on errors.
* fix(tui): rebuild prompt cleanly + sync session_key after compress
Pass system_message=None so AIAgent._compress_context rebuilds the system prompt without nesting the cached identity block. Reuse the handler's pre-snapshotted history inside _compress_session_history to avoid a second O(n) copy under the lock. After compaction, when AIAgent._compress_context rotates session_id, sync the gateway session_key, migrate approval notify + yolo state, restart the slash worker, and clear the stale pending title. Mirrors HermesCLI._manual_compress.
* Avoid /compress lock re-entry in slash side effects.
Stop pre-locking history before _compress_session_history in slash command mirroring, keep session-key sync parity with manual compression, and add a regression test that asserts /compress is invoked without holding history_lock.
* fix(tui): word-wrap composer input
Wrap composer input at word boundaries and anchor the good-vibes heart to the full composer row.
* test(tui): cover composer word wrap edge
Add regression coverage for moving the next word instead of splitting it at the composer edge.
* fix(tui): honor launch toolsets
Carry chat --toolsets through the TUI launcher so TUI sessions use the same per-session tool scope as the classic CLI.
* fix(tui): parse top-level toolsets flag
Allow top-level hermes --tui --toolsets to reach the implicit chat session, matching chat subcommand behavior.
* fix(tui): validate launch toolsets
Filter invalid HERMES_TUI_TOOLSETS entries and fall back to configured CLI toolsets when the override contains no valid toolsets.
* fix(tui): avoid config load for builtin toolsets
Honor built-in HERMES_TUI_TOOLSETS values before loading config and treat all/* as the all-toolsets sentinel.
* fix(cli): honor toolsets in oneshot mode
Forward top-level --toolsets into oneshot agent construction so the flag is not silently ignored outside the TUI path.
* fix(cli): validate oneshot toolsets
Reject invalid-only oneshot toolset overrides before output redirection and clarify TUI fallback warnings.
* fix(cli): preserve all-toolsets sentinel
Map explicit all/* oneshot toolset overrides to the all-toolsets sentinel and replace locals() checks in TUI toolset loading.
* fix(cli): warn on extra all-toolset entries
Warn when all/* toolset overrides include additional ignored entries so typos are still visible.
* fix(tui): honor plugin toolset overrides
Discover plugin toolsets before rejecting unresolved explicit toolset overrides and read raw config for MCP name validation.
* fix(tui): reuse toolset argument normalizer
Share top-level TUI toolset argument parsing with the oneshot path to avoid duplicate normalization logic.
* fix(cli): reject disabled mcp toolsets
Validate explicit toolset overrides against enabled MCP servers only and clarify top-level toolset flag help.
* fix(cli): distinguish disabled mcp from unknown toolsets
Report disabled MCP servers separately from unknown toolset entries and stub plugin discovery in invalid-name tests for determinism.
shutil.copytree from default ~/.hermes duplicated ~/.hermes/profiles into
the new profile, causing nested profiles/.../profiles/... and huge disk use.
Match export behavior (_DEFAULT_EXPORT_EXCLUDE_ROOT) by ignoring the sibling
profiles tree at the source root.
Made-with: Cursor
Intended placement per PR #17610 discussion — comfyui belongs in
skills/creative/ alongside other creative built-ins (touchdesigner-mcp,
pretext, sketch), not in optional-skills/.
Pure directory rename, no content changes. History preserved via git mv.
The skip_pre_tool_call_hook flag was added to prevent double-firing of
pre_tool_call when run_agent._invoke_tool pre-checks for a block
directive and then dispatches via handle_function_call. But the
implementation added an else: branch that fired invoke_hook again for
'observers', without noticing that get_pre_tool_call_block_message() in
hermes_cli.plugins already fires invoke_hook('pre_tool_call', ...) as
part of its block-directive poll.
Result: every tool call ran through the run_agent loop fired the hook
twice — reported by community users whose observer / audit plugins
logged each tool invocation twice with identical timestamps.
Fix: delete the else: branch. The single-fire contract is now:
- skip=False (direct handle_function_call): hook fires once inside
get_pre_tool_call_block_message().
- skip=True (run_agent._invoke_tool path): caller fires the hook
once via get_pre_tool_call_block_message(); handle_function_call
must not fire it again.
Tightened the existing skip-flag test (renamed to
test_skip_flag_prevents_double_fire) to assert pre_tool_call fires
zero times when skip=True, and added
test_run_agent_pattern_fires_pre_tool_call_exactly_once to lock in
end-to-end that the full block-check + dispatch sequence fires the
hook exactly once.
Adds Step 0 'Ask Local vs Cloud' as the very first onboarding step, with a
scripted question that spells out the hardware requirements for local
(6 GB VRAM NVIDIA, ROCm AMD on Linux, or M1+ Mac with 16 GB unified)
and routes Cloud users straight to Path A without a hardware check.
Hardware check becomes Step 1, run only when the user picked local.
Layers a programmatic hardware-feasibility check on top of the v4 skill
so the agent doesn't silently push users toward a local install they
can't actually run. The official comfy-cli supports --nvidia / --amd /
--m-series / --cpu, but has no guard against "4 GB laptop GPU on SDXL"
or "Intel Mac falling back to CPU" — both route to comfy-cli paths in
the original table and then fail on first workflow.
- scripts/hardware_check.py: detect OS/arch/GPU (NVIDIA nvidia-smi,
AMD rocm-smi, Apple M1+ via arm64+sysctl, Intel Arc via clinfo),
VRAM, system/unified RAM. Emits JSON
{verdict: ok|marginal|cloud, recommended_install_path, comfy_cli_flag}
with practical thresholds: discrete GPU >=6 GB VRAM minimum,
Apple Silicon >=16 GB unified memory minimum, Intel Mac -> cloud,
no accelerator -> cloud. comfy_cli_flag maps directly to
`comfy install` so the agent can stitch the whole flow together.
- scripts/comfyui_setup.sh: runs hardware_check.py first when no
explicit flag is passed. If verdict=cloud, refuses to install
locally, prints Comfy Cloud URL + an override command, exits 2.
Otherwise auto-selects the right --nvidia/--amd/--m-series flag
for `comfy install`. Surfaces marginal-verdict notes to the user.
- SKILL.md Setup & Onboarding: adds mandatory Step 0 "Check If This
Machine Can Run ComfyUI Locally" ahead of the Path A-E selection.
Documents the verdict thresholds inline, ties verdict + comfy_cli_flag
to the install paths, and updates the path-choice table so
"verdict: cloud" is the first row. Quick-Start "Detect Environment"
block extended to include the hardware check. Verification
checklist gains a hardware-check gate.
- Frontmatter setup.help rewritten to point at hardware_check.py
first. Version bumped 4.0.0 -> 4.1.0.
Capture the reusable layout and animation lessons from the advanced Pretext demo so the skill teaches measured obstacle fields, morphing geometry, and polished browser examples.
Cron is a built-in Hermes feature (CLI `hermes cron`, `cronjob` agent
tool, gateway ticker, scheduler in cron/scheduler.py) but croniter has
been gated behind the [cron] optional extra. Users who do a plain
`pip install hermes-agent` can create jobs via /cron but any recurring
cron schedule silently returns next_run_at=None (HAS_CRONITER=False),
which then gets wrapped into a 'state=error' message only after a tick.
Move croniter into core dependencies so scheduled jobs work out of the
box on any install path. The [cron] extra is kept as an empty
passthrough so existing `pip install hermes-agent[cron]` installs and
the [all]/[termux] extras continue to resolve.
Also update the now-stale user-facing error message in
`compute_next_run()` that still tells users to install `hermes-agent[cron]`.
Salvaged from #17234 (authored by @txbxxx) with a corrected premise:
the original PR claimed [cron] wasn't in [all], but it is (pyproject.toml
line 112). The real UX problem is the plain no-extras install path,
which this fix addresses.
Add a dedicated 'Pinning a skill' section that covers both gating
layers — curator auto-transitions AND the agent's skill_manage tool
— so users know what the flag actually protects against after
PR #17562. Updates the one-line claim in 'How it runs' to cross-link
the new section instead of only mentioning auto-transitions.
Extend curator's pin flag from 'skip auto-transitions' to 'no agent
edits at all'. All five skill_manage mutation actions (edit, patch,
delete, write_file, remove_file) now refuse pinned skills with a
message pointing the user at `hermes curator unpin <name>`.
Motivation: pin used to only stop the curator's own maintenance pass
from touching a skill. Nothing prevented the main agent from editing
or deleting a pinned skill via skill_manage in-session. This gives
users a hard fence against unwanted agent edits — same semantics as
curator pinning, extended to the write tool.
Create is unaffected (you can't pin a name that doesn't exist yet,
and name collisions already error out). Broken sidecars fail open
rather than lock the agent out.
The schema description advertises the new refusal so models know
not to route around it with rename/recreate tricks.
Skill catalog pages (bundled/optional) were drowning out real user-guide
and reference docs in search results. There are ~3100 of them and they
match on almost every generic term.
- Add `ignoreFiles` regexes to docusaurus-search-local for
`user-guide/skills/bundled/` and `user-guide/skills/optional/`.
The two human-written catalog indexes (`reference/skills-catalog`,
`reference/optional-skills-catalog`) remain indexed.
- Add a new feature page `user-guide/features/curator.md` covering the
curator subsystem merged in #16049 and refined in #17307 (per-run
reports): how it runs, config, CLI (`hermes curator status/run/pin/
restore/...`), `.usage.json` telemetry, archival semantics, and
recovery. Slotted into the Core features sidebar next to Skills.
Search index size dropped from 5822 docs to 2704 in the main section;
`user-guide/features/curator` is indexed.
Close integration gaps discovered by auditing qwen-oauth's file coverage.
These are surfaces the original salvage missed — they all existed on
main and were added in the 747 commits since PR #15203 was opened.
Coverage added:
- agent/credential_pool.py: seed pool from auth.json providers.minimax-oauth
so `hermes auth list` reflects logged-in state and
`hermes auth remove minimax-oauth <N>` works through the standard flow.
- agent/credential_sources.py: register RemovalStep for minimax-oauth
with suppression-aware `_clear_auth_store_provider`.
- agent/models_dev.py: PROVIDER_TO_MODELS_DEV mapping (-> 'minimax' family).
- hermes_cli/providers.py: HermesOverlay entry (anthropic_messages transport,
oauth_external auth_type, api.minimax.io/anthropic base).
- hermes_cli/model_normalize.py: add to _MATCHING_PREFIX_STRIP_PROVIDERS so
`minimax-oauth/MiniMax-M2.7` in config.yaml gets correctly repaired.
- hermes_cli/status.py: render MiniMax OAuth block in `hermes doctor`
(logged-in / region / expires_at / error).
- hermes_cli/web_server.py: register in OAUTH_PROVIDER_REGISTRY + dispatch
branch in _resolve_provider_status so the dashboard auth page shows it.
- website/docs/integrations/providers.md: full 'MiniMax (OAuth)' section.
- website/docs/reference/cli-commands.md: --provider enum.
- website/docs/user-guide/features/fallback-providers.md: fallback table row.
- scripts/release.py AUTHOR_MAP: amanning3390 mapping (CI gate).
Add comprehensive documentation for the minimax-oauth provider.
New file: website/docs/guides/minimax-oauth.md
- Overview table (provider ID, auth type, models, endpoints)
- Quick start via 'hermes model'
- Manual login via 'hermes auth add minimax-oauth'
- --region global|cn flag reference
- The PKCE OAuth flow explained step-by-step
- hermes doctor output example
- Configuration reference (config.yaml shape, region table, aliases)
- Environment variables note: MINIMAX_API_KEY is NOT used by
minimax-oauth (OAuth path uses browser login)
- Models table with context length note
- Troubleshooting section: expired token, timeout, state mismatch,
headless/remote sessions, not logged in
- Logout command
Updated: website/docs/getting-started/quickstart.md
- Add MiniMax (OAuth) to provider picker table as the recommended
path for users who want MiniMax models without an API key
Updated: website/docs/user-guide/configuration.md
- Add 'minimax-oauth' to the auxiliary providers list
- Add MiniMax OAuth tip callout in the providers section
- Add minimax-oauth row to the provider table (auxiliary tasks)
- Add MiniMax OAuth config.yaml example in Common Setups
Updated: website/docs/reference/environment-variables.md
- Annotate MINIMAX_API_KEY, MINIMAX_BASE_URL, MINIMAX_CN_API_KEY,
MINIMAX_CN_BASE_URL as NOT used by minimax-oauth
- Add minimax-oauth to HERMES_INFERENCE_PROVIDER allowed values
Wire MiniMax-M2.7 and MiniMax-M2.7-highspeed into the model catalog,
CLI model picker, and agent auxiliary/metadata subsystems.
Changes:
- hermes_cli/models.py:
- Add 'minimax-oauth' to _PROVIDER_MODELS with MiniMax-M2.7 and
MiniMax-M2.7-highspeed
- Add ProviderEntry('minimax-oauth', 'MiniMax (OAuth)', ...) to
CANONICAL_PROVIDERS near existing minimax entries
- Add aliases: minimax-portal, minimax-global, minimax_oauth in
_PROVIDER_ALIASES
- hermes_cli/main.py:
- Add 'minimax-oauth' to provider_labels dict
- Insert 'minimax-oauth' into providers list in
select_provider_and_model() near the other minimax entries
- Add 'minimax-oauth' to --provider argparse choices
- Add _model_flow_minimax_oauth() function: ensures login via
_login_minimax_oauth(), resolves runtime credentials, prompts for
model selection, saves model choice and config
- Add dispatch elif branch for selected_provider == 'minimax-oauth'
- agent/auxiliary_client.py:
- Add 'minimax-oauth': 'MiniMax-M2.7-highspeed' to
_API_KEY_PROVIDER_AUX_MODELS
- Add 'minimax-oauth' to _ANTHROPIC_COMPAT_PROVIDERS set
- agent/model_metadata.py:
- Add 'minimax-oauth' to _PROVIDER_PREFIXES frozenset
- MiniMax-M2.7 context length (200_000) already covered by the
existing 'minimax' substring match in DEFAULT_CONTEXT_LENGTHS
compute_next_run() ignored the last_run_at parameter for cron-type
schedules, always computing from _hermes_now() instead. This was
inconsistent with interval jobs which DO use last_run_at as the anchor.
After a crash or restart, cron jobs would compute next_run_at from
the arbitrary restart time rather than the actual last execution time.
While the stale detection in get_due_jobs() catches most cases, using
last_run_at as the croniter base eliminates edge cases and makes the
behavior consistent across schedule types.
Salvaged from #9014 (authored by @beenherebefore) onto current main.
The original PR branch was 2+ weeks stale and would have reverted
substantial unrelated work (jobs_file_lock, workdir/context_from/
enabled_toolsets, issue #16265 state=error recovery). Kept just the
7-line substantive fix and the regression test.
Bare `float(os.getenv("HERMES_CRON_TIMEOUT", 600))` in `run_job()` raises
a `ValueError` when the env var is set to a non-numeric string (e.g. "abc").
Replace it with the same defensive try/except pattern already used by
`_get_script_timeout()` for `HERMES_CRON_SCRIPT_TIMEOUT`: log a warning
and fall back to the 600 s default instead of crashing.
Also update the existing env-var tests to exercise the new code path and
add two new tests — one for an invalid value, one for an empty string.
Fixes#11319
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Closes#4759, closes#4381.
Mutating actions (patch, edit, write_file, remove_file, delete) used to
refuse skills that lived under `skills.external_dirs` with 'Skill X is in
an external directory and cannot be modified. Copy it to your local skills
directory first.' Faced with that error, the agent would fall back to
action='create', which always writes under ~/.hermes/skills/ — producing
a silent duplicate of the external skill in the local store.
Fix: drop the read-only gate. `skills.external_dirs` is configured by the
user; if they pointed it at a directory, they already said 'these are my
skills, treat them the same.' Filesystem permissions handle the genuine
read-only case (write fails, agent sees the error).
- New _containing_skills_root() resolves whichever dir actually contains
the skill; _delete_skill uses it to bound empty-category cleanup so an
external root is never rmdir'd.
- _create_skill behavior is unchanged: new skills still land in local
SKILLS_DIR only. Fewer moving parts.
- Seven new TestExternalSkillMutations tests covering patch/edit/write_file/
remove_file/delete/create against a mocked two-root layout + a category
rmdir-safety check.
When a user authenticates a built-in provider via env var (e.g. DASHSCOPE_API_KEY
triggers the built-in 'alibaba' row) AND defines a custom_providers entry
pointing at the same endpoint, the picker previously emitted two rows for one
endpoint. The built-in row already carries the canonical slug, curated model
list, and correct auth wiring, so the shadow custom entry is redundant.
Adds a _builtin_endpoints set populated as sections 1/2/2b emit rows. Each
entry is the provider's effective base URL (env override via base_url_env_var
wins over the static inference_base_url, so DASHSCOPE_BASE_URL-overridden
endpoints dedup correctly). Section 4 skips any grouped custom entry whose
base_url matches.
Intentionally does NOT repurpose model_catalog.enabled as a 'hide built-ins'
flag. That config controls the remote curated-manifest fetch (documented on
the model-catalog reference page) and overloading it would silently change
behavior for users who disable it for network/privacy reasons.
Three new tests:
- shadow dedup fires when endpoint matches static inference_base_url
- dedup does NOT hide custom entries on genuinely distinct endpoints
- dedup honors the base_url_env_var override path
Covers the #16748 fix:
- unsigned thinking blocks synthesised from reasoning_content survive replay
- non-latest assistant turns keep their thinking (DeepSeek validates every turn)
- signed Anthropic blocks are stripped (DeepSeek can't validate them)
- cache_control is stripped from thinking blocks
- OpenAI-compat base (api.deepseek.com without /anthropic) is NOT matched
- non-DeepSeek third parties (minimax) keep the generic strip-all behaviour
DeepSeek's /anthropic endpoint requires thinking blocks to be replayed
in multi-turn conversations for reasoning continuity. The existing code
classified api.deepseek.com as a generic third-party endpoint and stripped
ALL thinking blocks, causing HTTP 400 from DeepSeek.
Fix: add _is_deepseek_anthropic_endpoint() detector (following the Kimi
precedent) and a dedicated branch that strips only signed Anthropic blocks
while preserving unsigned ones synthesised from reasoning_content.
This follows the exact same pattern as the Kimi exemption (issue #13848)
and does not change behavior for any other third-party endpoint (Azure,
Bedrock, MiniMax, etc.).
FixesNousResearch/hermes-agent#16748
Follow-up to the cherry-picked PR #17447. The original flush spawned a
bare threading.Thread for the buffer-flush path, overwriting
self._sync_thread — which is aliased to the long-lived writer thread.
Two consequences:
1. No serialization with the writer queue. If old-session retains were
still queued in _retain_queue, the flush ran concurrently with the
writer and both threads could call aretain_batch against the same
document_id.
2. The pre-spawn 'self._sync_thread.join(timeout=5.0)' tried to join the
long-lived writer, which never exits, so the join was a no-op that
just timed out — never actually serialized anything.
Fix: enqueue the flush closure on _retain_queue via _ensure_writer +
put(). Natural FIFO ordering behind any pending retains, no new thread,
no broken join. Shutdown-aware so it doesn't enqueue after teardown.
Tests updated to drain via _retain_queue.join() instead of the stale
_sync_thread.join(). Added regression guard
test_flush_serializes_behind_pending_retains_via_writer_queue that
blocks the writer mid-retain to prove the flush waits in FIFO behind
the old retain.
Also seeds _retain_queue / _shutting_down / stubbed _ensure_writer on
the bare-object test helper in test_memory_session_switch.py so that
path doesn't blow up under the new queue-enqueue.
tests/plugins/memory/test_hindsight_provider.py + tests/agent/test_memory_session_switch.py: 103/103 passing.
Two data-loss / leak gaps in HindsightMemoryProvider.on_session_switch
introduced by #17409.
1. Buffered turns silently lost when retain_every_n_turns > 1.
on_session_switch unconditionally cleared _session_turns without
flushing. Users who batched every N>1 turns and switched mid-batch
(/reset, /new, /resume, /branch, or context compression) had those
buffered turns disappear. Same data-loss class as the shutdown race,
different lifecycle event.
Note commit_memory_session() -> on_session_end() runs *before*
on_session_switch on /reset, but Hindsight doesn't implement
on_session_end so the buffer survives that step and dies at clear
time. /resume, /branch, and compression skip commit_memory_session
entirely so an on_session_end impl wouldn't help them anyway.
Fix: snapshot the old _session_id, _document_id, _parent_session_id,
_turn_index, and _session_turns; spawn one final retain that lands
under the OLD document_id; then rotate state. Metadata is built
synchronously against the old self._* so session_id / lineage tags
on the flushed item all reference the prior session consistently.
2. Stale _prefetch_result leaks across switch.
If queue_prefetch ran in the old session and the result hadn't been
consumed by prefetch() yet, on_session_switch left the cached recall
text in place. The next session's first prefetch() call would return
text mined from the prior session's bank/query.
Fix: join any in-flight _prefetch_thread (3s bounded — matches
shutdown()), then clear _prefetch_result under _prefetch_lock before
rotating session_id.
Tests
-----
- tests/plugins/memory/test_hindsight_provider.py (TestSessionSwitchBufferFlush):
- buffered turns flushed under OLD document_id with OLD lineage tags
- empty buffer => no spurious retain
- _prefetch_result cleared on switch
- in-flight prefetch thread is awaited before clear (no race)
- tests/agent/test_memory_session_switch.py: factory extended to seed the
attrs the new flush path reads (_retain_source, _platform, _bank_id,
prefetch state, etc.) and stub _run_hindsight_operation so existing
switch-state assertions keep passing without network setup.
The ~/.openclaw/ detection banner (#16327) had two problems flagged in #16629:
1. It only pitched 'hermes claw cleanup' (destructive archive) and never
mentioned 'hermes claw migrate' — the actual non-destructive path that
ports config/memory/skills into Hermes.
2. The copy anthropomorphized the bug ('the agent can still get confused',
'dutifully reads') and framed OpenClaw as a competitor to eliminate
('instead of Hermes's').
Rewrite so migrate leads, cleanup is a clearly-labelled follow-up with a
warning that archiving breaks OpenClaw for users still running it.
Closes#16629
Address Copilot review on PR #16666:
1. **Duplicate event on every tool start** — both ``tool_progress_callback``
and ``tool_start_callback`` fire side-by-side in ``run_agent.py``, so
wiring both into chat completions emitted *two* ``hermes.tool.progress``
events per real tool call. Drop the legacy ``_on_tool_progress`` emit
entirely; ``_on_tool_start`` now produces a single unified event that
carries the legacy ``tool``/``emoji``/``label`` fields plus the new
``toolCallId``/``status`` correlation fields. Label is computed inline
via ``build_tool_preview`` so callers do not need to pre-format it.
2. **Weak per-event correlation in the regression test** — the previous
assertion checked that a ``toolCallId`` appeared *somewhere* in the
aggregate, which would have passed even if ``running`` lacked the id.
Collect ``(status, toolCallId)`` per event and assert each event
carries the correct pair, plus exactly two events on the wire (no
silent duplication regression).
The two existing chat-completions tool-progress tests are updated to fire
``tool_start_callback`` instead of ``tool_progress_callback``, matching
production reality where ``run_agent`` always pairs them.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Adds Vercel Sandbox as a supported Hermes terminal backend alongside
existing providers (Local, Docker, Modal, SSH, Daytona, Singularity).
Uses the Vercel Python SDK to create/manage cloud microVMs, supports
snapshot-based filesystem persistence keyed by task_id, and integrates
with the existing BaseEnvironment shell contract and FileSyncManager
for credential/skill syncing.
Based on #17127 by @scotttrinh, cherry-picked onto current main.
Adds two API server endpoints for external UIs and orchestrators:
- GET /v1/capabilities — machine-readable feature discovery so clients
can detect which Runs API / SSE / auth features this Hermes version
supports before depending on them.
- GET /v1/runs/{run_id} — pollable run status so dashboards can check
queued/running/completed/failed/cancelled/stopping state without
holding an SSE connection open.
Also moves request validation ahead of run allocation so invalid
payloads no longer leave orphaned entries in _run_streams waiting for
the TTL sweep.
task_id is intentionally kept as "default" for the Runs API to
preserve the shared-sandbox model used by CLI, gateway, and the
existing _run_agent_with_callbacks path. session_id is surfaced in
run status for external-UI correlation only.
Salvage of PR #17085 by @Magaav.
The guard that drops Anthropic's `thinking` kwarg for Kimi endpoints was
matched on `https://api.kimi.com/coding` only. Users configuring a
custom Kimi-compatible gateway (or an official Moonshot host) with
`api_mode: anthropic_messages` fall through to the generic third-party
path, which strips thinking blocks AND still sends
`thinking={enabled,...}` → upstream rejects with HTTP 400
"reasoning_content is missing in assistant tool call message at index N"
on the next request after a tool call.
Replace `_is_kimi_coding_endpoint` callers (history replay + thinking
kwarg gate) with `_is_kimi_family_endpoint(base_url, model)` that also
matches the `api.kimi.com` / `moonshot.ai` / `moonshot.cn` hosts and
Kimi/Moonshot family model names (`kimi-`, `moonshot-`, `k1.`, `k2.`,
…) for custom / proxied endpoints. Keeps the UA-header check in
`build_anthropic_client` URL-only — the `claude-code/0.1.0` header is
an official-Kimi contract.
Plumbs optional `model` through `convert_messages_to_anthropic` so
the unsigned reasoning_content→thinking block synthesised for Kimi's
history validation survives the third-party signature-stripping pass
on custom hosts too.
Closes#17057.
The cron schema contracts deliver as a string ("local", "origin",
"telegram", "telegram:chat_id[:thread_id]", or comma-separated combos),
but MCP clients and scripts sometimes pass an array like ['telegram'].
Before this change, the list was written to jobs.json verbatim, and
the scheduler's str(deliver).split(',') then tried to resolve the
literal string "['telegram']" as a platform — returning None and
logging 'no delivery target resolved for deliver=[\'telegram\']'.
Fix on both ends:
- tools/cronjob_tools.py: normalize deliver at the API boundary on
create and update, so storage is always a string.
- cron/scheduler.py: normalize deliver in _resolve_delivery_targets,
so existing jobs.json entries with list-form deliver are handled
gracefully without requiring users to edit the file.
Closes#17139
The normalize_model_name() function unconditionally converted dots to
hyphens in all model names. This caused non-Anthropic models (e.g.
gpt-5.4) to be mangled to gpt-5-4 when routed through the Anthropic
adapter path, resulting in HTTP 404 from the backend.
Now only applies dot-to-hyphen conversion for models starting with
"claude-" or "anthropic/", which are the actual Anthropic model IDs.
Fixes NousResearch/hermes-agent#17171
Related: #7421, #13061, #16417
The plugin used to spawn one daemon thread per sync_turn() to do the
aretain_batch network write. On CLI exit, that pattern raced interpreter
shutdown — the last retain could reach aiohttp after asyncio's
"cannot schedule new futures" guard had fired, producing noisy logs and
silently losing the final unsaved turn:
WARNING ... Hindsight sync failed: cannot schedule new futures after
interpreter shutdown
ERROR asyncio: Unclosed client session
client_session: <aiohttp.client.ClientSession object at 0x...>
Switch to a single-writer model: each provider owns one long-lived
writer thread plus a queue. sync_turn() snapshots state and enqueues a
job; the writer drains sequentially. Once shutdown() is called:
- new sync_turn() / queue_prefetch() calls are dropped, not enqueued
- a sentinel wakes the writer so it finishes in-flight work
- shutdown joins the writer (10s) before nulling the client
Also register an idempotent atexit hook from the first sync_turn(), so
exit paths that don't go through MemoryManager.shutdown_all() (Ctrl-C,
abrupt exit) still get a chance to drain.
Tests: keep _sync_thread as a legacy alias to the writer, swap join()
calls to _retain_queue.join() (canonical wait-for-drain), add a new
TestShutdownRace suite covering single-writer reuse, post-shutdown drop,
queue draining, and shutdown idempotency.
Pass encoding='utf-8', errors='ignore' and guard against result.stdout
being None so _scan_gateway_pids() no longer crashes with
UnicodeDecodeError + AttributeError on Windows systems whose default
code page is not UTF-8 (e.g. cp936 on zh-CN). The parser only matches
the ASCII prefixes CommandLine= and ProcessId=, so dropping undecodable
bytes is safe.
Closes#17049.
Two fix-ups for #17123:
1. Reword the inline comment in `_warn_stale_dashboard_processes` to
accurately describe the failure mode (locale-dependent decoder, not a
"default UTF-8 decoder") and identify `errors="ignore"` as the
load-bearing protection. Per Copilot's review.
2. Switch `TestWindowsWmicEncoding` from `patch("hermes_cli.main.sys")`
to `monkeypatch.setattr(sys, "platform", "win32")` — the codebase's
canonical pattern (e.g. `tests/hermes_cli/test_auth_ssl_macos.py`).
The MagicMock-replacement approach passed locally on Python 3.12 but
the platform-equality check failed under CI's xdist+Python 3.11,
leaving both new tests red despite the fix being present.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
`hermes update` calls `_warn_stale_dashboard_processes()` to warn about
dashboard processes still running the pre-update Python backend. On
Windows, that scan shells out to `wmic process get ProcessId,CommandLine
/FORMAT:LIST` with `text=True` and no explicit encoding.
`wmic` emits text in the system code page (e.g. cp936 on zh-CN locales),
not UTF-8. Without an explicit `encoding=`, Python's default UTF-8
decoder crashes the subprocess reader thread with
`UnicodeDecodeError: 'utf-8' codec can't decode byte 0xd0 ...`. In
Python 3.11 that crash is silently absorbed: `subprocess.run()` returns
a `CompletedProcess` with `result.stdout = None`, the next line calls
`result.stdout.split("\n")`, and `hermes update` aborts with the
exact `AttributeError: 'NoneType' object has no attribute 'split'`
trace reported in #17049.
Fix: pass `encoding="utf-8", errors="ignore"` so undecodable bytes
cannot take down the reader thread (the parsing only matches the ASCII
prefixes `CommandLine=` and `ProcessId=`, so dropping non-UTF-8 bytes
is safe), and short-circuit when `result.stdout is None` as a defensive
guard for environments where the reader thread still fails for other
reasons.
This is the same root cause as #17074 (which patches
`hermes_cli/gateway._scan_gateway_pids` for the `hermes setup` path).
That PR does not touch `_warn_stale_dashboard_processes`, so
`hermes update` remains broken on the same locales until this lands.
Regression test in `tests/hermes_cli/test_update_stale_dashboard.py`:
- `test_wmic_invoked_with_utf8_ignore_errors` asserts the explicit
encoding/errors kwargs reach `subprocess.run`.
- `test_wmic_returns_none_stdout_does_not_crash` simulates the
reader-thread-crashed `result.stdout=None` aftermath and asserts the
function returns silently instead of raising AttributeError.
Both new tests fail against clean origin/main (7d4648461) reproducing
the original AttributeError; both pass with this patch. The remaining
3 failures in `tests/hermes_cli/test_cmd_update.py` and
`test_update_autostash.py` are pre-existing baselines on origin/main —
they reproduce identically without this change and are unrelated to
the wmic scan.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
QR-login connects an iLink bot identity (...@im.bot), not a scriptable
personal WeChat account. iLink typically does not deliver ordinary WeChat
group events to these bots, so WEIXIN_GROUP_POLICY / WEIXIN_GROUP_ALLOWED_USERS
often have no effect regardless of value.
- Setup wizard: print iLink-bot caveat before the group-policy prompt; relabel
the allowlist input as 'group chat IDs (not member user IDs)'; note that
'open' / 'allowlist' only take effect if iLink delivers group events.
- Adapter: log a WARNING at connect() when WEIXIN_GROUP_POLICY is non-disabled
so the limitation is surfaced in gateway logs, not just docs.
- Docs: add a top-of-page warning callout to weixin.md explaining the iLink
bot identity, narrow the 'DM and group messaging' feature line to DM-only
with a group caveat, tighten the Group Policy section and troubleshooting
row, and clarify WEIXIN_GROUP_ALLOWED_USERS as group IDs (not user IDs)
in weixin.md and environment-variables.md.
Closes#17094
Widen #17163 to the sibling file tools/transcription_tools.py, which had
the same class of bug. STT provider call sites and the _get_provider
selection gate called os.getenv(...) directly and missed keys that only
lived in ~/.hermes/.env.
Same pattern as tts_tool.py: one guarded top-level import of
get_env_value (falls back to os.getenv on ImportError), then every
API-key and paired-base-URL lookup swapped over.
Call sites migrated:
- _transcribe_groq — GROQ_API_KEY
- _transcribe_mistral — MISTRAL_API_KEY
- _transcribe_xai — XAI_API_KEY, XAI_STT_BASE_URL
- _get_provider — GROQ/MISTRAL/XAI_API_KEY in explicit + auto branches
Module-level defaults (DEFAULT_STT_MODEL, GROQ_BASE_URL, etc.) stay on
os.getenv — they're import-time constants, not runtime config, and the
dotenv fallback would add no value there.
New regression tests in tests/tools/test_transcription_dotenv_fallback.py
(8 cases) mirror briandevans' TTS tests: per-provider dotenv-key
forwarding, selection-gate dotenv visibility, and an end-to-end probe
that patches hermes_cli.config.load_env to simulate ~/.hermes/.env
carrying the key while os.environ does not.
Wrap the new top-level `from hermes_cli.config import get_env_value`
in try/except ImportError and fall back to a thin os.getenv shim, so
importing tools.tts_tool keeps working in environments where
hermes_cli.config is unavailable. This matches the existing tolerance
in `_load_tts_config()` (tools/tts_tool.py) and the same
import-fallback pattern in tools/tool_backend_helpers.py::fal_key_is_configured.
Also update the TestDotenvFallbackPerProvider docstring to accurately
describe the mocking strategy: per-provider tests patch
`tools.tts_tool.get_env_value` directly, while the regression-guard
tests cover the lower-level `hermes_cli.config.load_env` integration.
Addresses Copilot review on #17163.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
TTS provider tools (elevenlabs, xai, minimax, mistral, gemini) called
os.getenv("X_API_KEY") directly, which bypassed Hermes's dotenv bridge in
hermes_cli.config. Users who keep their TTS keys only in ~/.hermes/.env saw
"X_API_KEY not set" errors even though the rest of the stack
(agent/credential_pool, hermes_cli/auth) already resolves keys through
get_env_value() — same class of bug as #15914 fixed for those modules.
Switch every TTS env-var lookup (API keys, base URLs, and
check_tts_requirements gates) to get_env_value, which checks os.environ
first and then ~/.hermes/.env. Behaviour for users with keys exported in
the shell is unchanged; users with dotenv-only keys now succeed. The two
diagnostics prints in __main__ are migrated for consistency.
Regression test (tests/tools/test_tts_dotenv_fallback.py):
- per-provider: each backend reads the dotenv key when only
~/.hermes/.env carries it (5 providers).
- end-to-end: with hermes_cli.config.load_env returning the key and
os.environ empty, _generate_minimax_tts and check_tts_requirements
both succeed; reverting tools/tts_tool.py back to os.getenv makes all
7 tests fail with "MINIMAX_API_KEY not set" / similar.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* docs(anthropic): correct OAuth scope to Max plan + extra usage credits only
The previous docs pass (#17399) overstated what Anthropic OAuth works
with. In practice Hermes can only route against a Claude Max plan that
has purchased extra usage credits — the base Max allowance is not
consumed, and Claude Pro is not supported at all. Without Max + extra
credits, users must fall back to an ANTHROPIC_API_KEY (pay-per-token).
Updates the four pages touched in #17399:
- integrations/providers.md
- user-guide/features/credential-pools.md
- reference/environment-variables.md
- getting-started/quickstart.md
* fix(aux): skip kimi-coding in vision auto-detect (closes#17076)
Kimi Coding Plan's /coding endpoint (Anthropic Messages wire) has no
image_in capability — Kimi's own docs confirm and suggest switching to
a vision-capable model. Vision lives on the separate Kimi Platform
(api.moonshot.ai, OpenAI-wire, pay-as-you-go). When the user has
kimi-coding as main provider and auxiliary.vision.provider=auto,
resolve_vision_provider_client was handing back an AnthropicAuxiliaryClient
wrapped around /coding which 404'd on every vision request.
Add a _PROVIDERS_WITHOUT_VISION frozenset ({kimi-coding, kimi-coding-cn})
and gate the main-provider vision branch on membership. On a skip the
auto-detect falls through to OpenRouter → Nous like any other
main-provider-unavailable case.
Explicit per-task overrides (auxiliary.vision.provider=kimi-coding) are
unaffected — the skip only applies when the caller is in auto mode.
Tests: 4 new targeted tests in TestVisionAutoSkipsKimiCoding covering
the skip path, CN variant, explicit-override passthrough, and a guard
against accidental skip-list widening.
_update_cwd() uses a bare open(self._cwd_file).read() that never
closes the file descriptor. This method runs on every terminal
command execution, so the fd leaks accumulate in long sessions.
Use a with statement so the fd is released promptly.
Fixes#15552 (standalone resubmission)
Regression test for the ret=-2 / errmsg='unknown error' disambiguation:
- ret=-2 or errcode=-2 with 'unknown error' → stale session (True)
- ret=-2 with 'freq limit' or other errmsg → rate limit (False)
- ret=-14 → not matched here (handled by SESSION_EXPIRED_ERRCODE path)
- Success codes and missing errmsg → False
The Weixin adapter only recognized errcode=-14 as a session-expired
signal. However, iLink also returns ret=-2 with errmsg="unknown error"
for the same underlying condition (stale session). The adapter treated
ret=-2 as a rate-limit, exhausting retries with the same stale
context_token instead of refreshing the session.
Added _is_stale_session_ret() helper that distinguishes ret=-2 with
"unknown error" from genuine rate limits. Updated both the poll loop
and _send_text_chunk to use the helper.
FixesNousResearch/hermes-agent#17228
_run_async() bridges sync tool handlers to async code. When the handler
is invoked from inside a running event loop (gateway / nested async),
it spawns a worker thread and blocks on future.result(timeout=300).
Before this change, a coroutine that ran past 300s leaked its worker
thread:
- future.cancel() is a no-op on a running ThreadPoolExecutor future
(cancel only works on not-yet-started work).
- pool.shutdown(wait=False, cancel_futures=True) let the caller
proceed but the worker kept running the coroutine until it
returned on its own.
Every tool timeout leaked one thread. In long-lived gateway / RL
sessions this is cumulative.
The fix replaces bare asyncio.run() with a worker wrapper that
creates its own event loop. On timeout, _run_async schedules
task.cancel() on that loop via call_soon_threadsafe, then shuts the
pool down with wait=False so the caller returns immediately. The
coroutine observes CancelledError at its next await and the worker
thread exits cleanly.
Also switches logger.error() to logger.exception() in the top-level
handle_function_call() except block so tool failures produce full
stack traces in errors.log instead of just the message.
Related: #17420 (contributor flagged the leak; the original fix used
pool.shutdown(wait=True) which would have converted the leak into a
hang — caller blocks forever on the same stuck coroutine). Credit
for identifying the leak goes to the contributor.
Co-authored-by: 0z! <162235745+0z1-ghb@users.noreply.github.com>
Wrap each adapter.connect() in asyncio.wait_for() so one platform hanging
during startup or reconnect cannot block the others. Telegram's 8-retry
connect loop (~140s worst case) previously prevented Feishu from ever
starting when Telegram was network-restricted — common for users in
regions where Telegram is blocked.
Default timeout is 30s; override via HERMES_GATEWAY_PLATFORM_CONNECT_TIMEOUT
(0 disables). Applied to both startup and the reconnect watcher so a
platform that hangs mid-retry also does not stall retries for others.
Fixes#17242
When a background terminal process spawns a descendant daemon that
inherits the stdout pipe (e.g. 'hermes update' triggering a gateway
systemctl restart), the reader thread's stdout.read() never returns EOF
and its finally: block never runs. session.exited stays False forever,
so process(action='poll') returns 'running' indefinitely even though
the direct child exited long ago.
Issue #17327: Feishu user polled 74 times over 7 minutes before killing
the gateway manually.
Fix: add _reconcile_local_exit() that checks the direct Popen.poll()
before trusting session.exited. If the direct child has exited, drain
any immediately-readable bytes non-blocking and flip session.exited.
Called from poll() and wait(). The stuck reader thread remains blocked
but is a daemon thread and gets reaped with the process.
Safe no-op for env/PTY sessions, already-exited sessions, and live
children (returns None from Popen.poll()).
Fixes#6672
Memory providers now receive on_session_switch() whenever AIAgent.session_id
rotates mid-process — /resume, /branch, /reset, /new, and context
compression. Before this, providers that cached per-session state in
initialize() (Hindsight's _session_id, _document_id, accumulated
_session_turns, _turn_counter) kept writing into the old session's
record after the agent had moved on.
MemoryProvider ABC
------------------
- New optional hook on_session_switch(new_session_id, *,
parent_session_id='', reset=False, **kwargs) with no-op default for
backward compat. reset=True signals /reset or /new — providers should
flush accumulated per-session buffers. reset=False for /resume,
/branch, compression where the logical conversation continues.
MemoryManager
-------------
- on_session_switch() fans the hook out to every registered provider.
Isolated try/except per provider — one bad provider can't block others.
- Empty/None new_session_id is a no-op to avoid corrupting provider state
during shutdown paths.
run_agent.py
------------
- _sync_external_memory_for_turn now passes session_id=self.session_id
into sync_all() and queue_prefetch_all(). Providers with defensive
session_id updates in sync_turn (Hindsight already had this at
plugins/memory/hindsight/__init__.py:1199) now actually receive the
current id.
- Compression block at ~L8884 already notified the context engine of
the rollover; now also calls
_memory_manager.on_session_switch(reason='compression').
cli.py
------
- new_session() fires reset=True, reason='new_session' so providers
flush buffers.
- _handle_resume_command fires reset=False, reason='resume' with the
previous session as parent_session_id.
- _handle_branch_command fires reset=False, reason='branch' with the
parent session_id already captured for the DB parent link.
gateway/run.py
--------------
- _handle_resume_command now evicts the cached AIAgent, mirroring
/branch and /reset. The next message rebuilds a fresh agent whose
memory provider initialize() runs with the correct session_id —
matches the pattern the gateway already uses for provider state
cross-session transitions.
Hindsight reference implementation
----------------------------------
- plugins/memory/hindsight/__init__.py adds on_session_switch that:
updates _session_id, mints a fresh _document_id (prevents
vectorize-io/hindsight#1303 overwrite), and clears _session_turns /
_turn_counter / _turn_index so in-flight batches don't flush under
the new document id. parent_session_id only overwritten when provided
(avoids clobbering on a bare switch).
Tests
-----
- tests/agent/test_memory_session_switch.py: new dedicated file. ABC
default no-op, manager fan-out, failure isolation, empty-id no-op,
session_id propagation through sync_all/queue_prefetch_all, Hindsight
state transitions for every reset/non-reset case, parent preservation.
- tests/cli/test_branch_command.py: new test verifying /branch fires
the hook with correct parent_session_id + reset=False + reason.
- tests/gateway/test_resume_command.py: new test verifying /resume
evicts the cached agent.
- tests/run_agent/test_memory_sync_interrupted.py: updated existing
assertions to account for the session_id kwarg on sync_all and
queue_prefetch_all.
E2E verified (real imports, tmp HERMES_HOME):
- /resume: session_id updates, doc_id fresh, buffers cleared, parent set
- /branch: session_id forks, parent links to original
- /new: reset=True clears accumulated state
- compression: reason='compression' propagated, lineage preserved
- Empty id: no-op, state preserved
- Legacy provider without on_session_switch: no crash
Reported by @nicoloboschi (Hindsight maintainer); related scope-widening
comment by @kidonng extending coverage to compression.
MiniMax's /anthropic endpoint documents cache_control support (0.1x read
pricing, 5-min TTL) for MiniMax-M2.7, M2.5, M2.1, M2. PR #12846 gated
third-party Anthropic-wire caching on 'claude' in model name, which left
MiniMax's own model family re-paying full input tokens every turn.
Opt in explicitly via provider id (minimax / minimax-cn) or host match
(api.minimax.io / api.minimaxi.com). Narrow allowlist mirroring the
existing Qwen/Alibaba branch below; leaves room for a capability-based
surface (ProviderConfig.supports_anthropic_cache) if a third provider
needs it.
Closes#17332
Fixes#16825. Sessions using MiniMax-M2.7 via minimax-cn showed
estimated_cost_usd=0.0 and cost_status='unknown' because neither
provider had a billing route or pricing entry. Adds official_docs_snapshot
entries ($0.30/M input, $1.20/M output) for both minimax and minimax-cn,
and adds explicit routing in resolve_billing_route so both resolve to
billing_mode='official_docs_snapshot' instead of falling through to 'unknown'.
_send_yuanbao() already supported media_files= and the user-facing
error strings already advertised yuanbao support, but there was no
dispatch branch in _send_to_platform() actually routing to it. Target
yuanbao in send_message previously fell through to
"Direct sending not yet implemented".
- Add yuanbao media-chunk branch (mirrors Signal/Matrix: media on
final chunk only).
- Add yuanbao elif in the non-media loop.
Salvage of #17411; SKILL.md description change and redundant
sidebars.ts entry dropped, indentation/trailing-whitespace cleaned up.
- _markdown_to_signal docstring claimed SPOILER support but the regex list
never handled ``||...||``. Correct the docstring to match the four
actually-supported styles (BOLD / ITALIC / STRIKETHROUGH / MONOSPACE).
Signal's SPOILER bodyRange would need dedicated ``||spoiler||`` parsing
and is left for a follow-up.
- scripts/release.py: add exiao's noreply email to AUTHOR_MAP so the
contributor-attribution gate accepts their cherry-picked commit.
Three Signal adapter improvements that depend on the no-edit-mode
plumbing from the previous commit.
1. Native formatting (markdown -> Signal bodyRanges)
Signal renders markdown as literal characters (**bold**, `code`, #
heading), which looks broken. Added _markdown_to_signal(text) that
strips markdown syntax and emits Signal-native bodyRanges as
start:length:STYLE entries. Offsets are computed in UTF-16 code
units so non-BMP emoji stay aligned. Supports BOLD, ITALIC, STRIKE,
MONO, and headings mapped to BOLD. Fenced code and inline code are
handled; link syntax is unwrapped to visible text + URL.
Includes edge-case fixes reported previously:
- Bullet lists ("* item") no longer misidentified as italics
- URLs containing underscores no longer italicized around the dot
2. Reply-quote context
Parses dataMessage.quote on inbound messages and populates
MessageEvent.raw_message with sender + timestamp_ms. This lets the
gateway's existing [Replying to: "..."] injector (gateway/run.py)
work on Signal, matching Telegram/Matrix behavior.
3. Processing reactions
Overrides on_processing_start -> hourglass and on_processing_complete
-> checkmark via the sendReaction JSON-RPC using targetAuthor and
targetTimestamp pulled from raw_message. Uses the ProcessingOutcome
enum introduced in the previous commit.
Also sets SUPPORTS_MESSAGE_EDITING = False on SignalAdapter so the
no-edit streaming path activates.
Tests: 40+ new tests in tests/gateway/test_signal_format.py covering
markdown conversion, UTF-16 offset correctness with non-BMP emoji,
bullet-list and URL false-positive regressions, reply-quote extraction,
and reaction payload shape. Regression extensions to test_signal.py.
The previous docs pass (#17399) overstated what Anthropic OAuth works
with. In practice Hermes can only route against a Claude Max plan that
has purchased extra usage credits — the base Max allowance is not
consumed, and Claude Pro is not supported at all. Without Max + extra
credits, users must fall back to an ANTHROPIC_API_KEY (pay-per-token).
Updates the four pages touched in #17399:
- integrations/providers.md
- user-guide/features/credential-pools.md
- reference/environment-variables.md
- getting-started/quickstart.md
Users have been asking what they're billed for when they authenticate
Anthropic via OAuth in Hermes. Clarify in the provider docs that OAuth
routes through Anthropic's Claude Code subscription path — consuming
the extra Claude Code usage included with their Pro or Max plan — and
that an ANTHROPIC_API_KEY is pay-per-token against that key's org
instead.
Touches:
- integrations/providers.md: new info admonition in Anthropic (Native)
section, plus provider-table row.
- user-guide/features/credential-pools.md: OAuth comment line.
- reference/environment-variables.md: Provider Auth (OAuth) intro.
- getting-started/quickstart.md: provider-picker table row.
Completes the cfg_get migration started in PR #17304. Covers the
remaining hermes_cli/ and plugins/ config-access sites that the first
PR intentionally left opportunistic.
Migrated (33 sites across 14 files):
hermes_cli/setup.py 13 sites (terminal.*, agent.*, display.*, compression.*, tts.*)
hermes_cli/tools_config.py 7 sites (tts.*, browser.*, web.*, platform_toolsets.*)
hermes_cli/plugins_cmd.py 3 sites (plugins.*, memory.*, context.*)
plugins/memory/honcho/cli.py 3 sites (hosts.*)
hermes_cli/web_server.py 1 site (dashboard.*)
hermes_cli/skills_config.py 1 site (platform_disabled)
hermes_cli/plugins.py 1 site (plugins.disabled)
hermes_cli/status.py 1 site (terminal.backend)
hermes_cli/mcp_config.py 1 site (mcp_servers.*)
hermes_cli/webhook.py 1 site (platforms.webhook)
plugins/memory/__init__.py 1 site (memory.provider)
plugins/memory/hindsight/ 1 site (banks.hermes)
plugins/memory/holographic/ 1 site (plugins.hermes-memory-store)
run_agent.py 1 site (auxiliary.compression)
The helper supports non-literal keys too, so e.g.
cfg.get('hosts', {}).get(HOST, {})
becomes
cfg_get(cfg, 'hosts', HOST, default={})
Migration bugs caught and fixed during this PR:
1. An AST-based batch rewrite naïvely captured the first word token in
a chain, which corrupted 'self._config.get(...).get(...)' into
'self.cfg_get(_config, ...)' (dropping 'self.', creating a broken
method call). Plugins/memory/hindsight caught it via its test suite.
Fixed manually to 'cfg_get(self._config, ...)'.
2. Import-extension heuristic rewrote multi-line parenthesized imports
('from X import (\n A,\n B,\n)') as
'from X import cfg_get, (' — syntactically broken. Fixed by inserting
cfg_get as the first name inside the parentheses.
Combined with PR #17304, the cfg_get migration now covers:
PR #17304 (first batch): 20 sites in tools/ + gateway/
PR #17317 (this one): 33 sites in hermes_cli/ + plugins/ + run_agent.py
Total: 53 sites migrated. Remaining ~8 sites are either:
- Function-call chains (e.g. '_load_stt_config().get(...).get(...)')
that would need double-evaluation or a local binding to migrate
cleanly — intentionally deferred.
- JSON response-navigation (e.g. 'response_data.get('data',{}).get('web'))
which is unrelated to config access and shouldn't use cfg_get.
Verified:
- 412/412 tests/plugins/ pass (including the hindsight test that caught
the self.X regex bug before commit)
- 3181/3189 tests/hermes_cli/ pass (8 pre-existing failures on main,
verified by git-stash comparison)
- Live 'hermes status' and 'hermes config' render correctly (exercise
the migrated terminal.backend, tts.provider, browser.cloud_provider,
compression.threshold, display.tool_progress sites)
- Live 'hermes chat': 1 turn + /quit, zero errors in 11-line log window
No semantic changes — cfg_get was already proven to be a 1:1 match for
the original .get("X",{}).get("Y",default) pattern in PR #17304.
Every curator pass now emits a dated report directory under
`~/.hermes/logs/curator/{YYYYMMDD-HHMMSS}/` with two files:
- `run.json` — machine-readable full record (before/after snapshot,
state transitions, all tool calls, model/provider, timing, full LLM
final response untruncated, error if any)
- `REPORT.md` — human-readable markdown: model + duration header,
auto-transition counts, LLM consolidation stats, archived-this-run
list, new-skills-this-run list, state transitions, the full LLM
final summary, and a recovery footer pointing at the archive + the
`hermes curator restore` command
Reports live under `logs/curator/`, not inside `skills/` — they're
operational telemetry, not user-authored skill data, and belong
alongside `agent.log` / `gateway.log`.
Internals:
- `_run_llm_review()` now returns a dict (final, summary, model,
provider, tool_calls, error) instead of a bare truncated string so
the reporter has full fidelity
- Report writer is fully best-effort — any failure logs at DEBUG and
never breaks the curator itself. Same-second rerun gets a numeric
suffix so reports can't clobber each other
- Report path stamped into `.curator_state` as `last_report_path`
- `hermes curator status` surfaces a "last report:" line so users
can immediately open the latest run
Tests (all green):
- 7 new tests in tests/agent/test_curator_reports.py covering: report
location (logs not skills), both files written, run.json shape and
diff accuracy, markdown structure, error path still writes, state
transitions captured, same-second runs get unique dirs
- Existing test_run_review_synchronous_invokes_llm_stub updated to
stub the new dict-returning _run_llm_review signature
Live E2E: ran a synchronous pass against a 1-skill test collection
with a stubbed LLM; report written correctly, state stamped with
last_report_path, markdown human-readable, run.json machine-parseable.
The "cfg.get('X', {}).get('Y', default)" pattern appears 50+ times
across tools/, gateway/, and plugins/. Each call site manually handles
the same three gotchas:
1. Missing intermediate key → empty dict → chain works
2. Non-dict value at intermediate position → AttributeError
(uncaught in most sites, so a misconfigured YAML crashes the tool)
3. cfg is None → AttributeError
Introduces cfg_get(cfg, *keys, default=None) in hermes_cli/config.py
as the canonical helper. Handles all three uniformly, returns default
only when the final key is *absent* (matches dict.get semantics —
explicit None values are preserved, falsy values like 0 / False / ''
are preserved).
Named cfg_get rather than cfg_path to avoid shadowing the existing
'cfg_path = _hermes_home / "config.yaml"' local variable that appears
in gateway/run.py, cron/scheduler.py, hermes_cli/main.py, etc.
Migrated 20 call sites as the first-batch proof-of-value:
gateway/run.py 10 sites (agent/display subtrees)
tools/browser_tool.py 3 sites
tools/vision_tools.py 2 sites
tools/browser_camofox.py 1 site
tools/approval.py 1 site
tools/skills_tool.py 1 site
tools/skill_manager_tool.py 1 site
tools/credential_files.py 1 site
tools/env_passthrough.py 1 site
The remaining ~30 sites across plugins/ and smaller tool files can be
migrated opportunistically — the helper is now available and the
pattern is established.
Fixed a latent bug along the way: tools/vision_tools.py had its
cfg_get usage at line 560 inside a function that locally re-imports
'from hermes_cli.config import load_config', but the AST-based
migration script wrote the top-level cfg_get import to a different
function scope, leaving line 560's cfg_get as a NameError silently
swallowed by the surrounding try/except. Test
test_vision_uses_configured_temperature_and_timeout caught it. Fixed
by including cfg_get in the function-local import.
Verified:
- 7880/7893 tests/tools/ + tests/gateway/ + tests/hermes_cli/test_config
tests pass; all 13 failures pre-existing on main (MCP, delegate,
session_split_brain — verified earlier in the sweep).
- All 20 migrated sites AST-verified to have cfg_get in scope (either
module-level or function-local).
- Live 'hermes chat' smoke: 2 turns + /model switch + tool calls +
/quit, zero errors. Agent correctly counted 20 cfg_get hits across
8 tool files — matching the migration.
Semantic parity verified against the original pattern across 8 edge
cases (missing keys, None values, falsy values, empty strings, string
instead of dict, None cfg, nested levels).
Add opt-in terminal.docker_run_as_host_user config flag that passes
--user $(id -u):$(id -g) to the Docker backend so files written into
bind-mounted directories (/workspace, /root, docker_volumes entries) are
owned by the host user instead of root.
When enabled on POSIX platforms, also drops SETUID/SETGID caps since the
container no longer needs gosu/su to switch users. Falls back cleanly on
platforms without os.getuid (e.g. native Windows Docker) with a warning.
Wired through all three config.yaml -> TERMINAL_* env-var bridges:
- cli.py env_mappings (CLI + TUI startup)
- gateway/run.py _terminal_env_map (gateway / messaging platforms)
- hermes_cli/config.py _config_to_env_sync (`hermes config set`)
Also fixes docker_mount_cwd_to_workspace silently failing in gateway
mode -- it was missing from gateway/run.py's _terminal_env_map.
Adds tests/tools/test_terminal_config_env_sync.py to guard against
future drift between the three bridges (same bug class shipped twice
in one month).
Bundled Hermes image won't work with this flag since its entrypoint
expects to start as root for the usermod/gosu hermes flow; works with
the default nikolaik/python-nodejs image and plain Debian/Ubuntu.
Copilot caught that clearing inFlight on a transient normal-memory tick could
allow a second dump/eviction to start before the first async tick completed.
Only clear dumped on normal; let the in-flight tick's finally remove its own
level.
Tests:
- cd ui-tui && npm run type-check && npm run build
Copy profile dashboard changes onto a fresh branch under the vincez-hms-coder account.
Includes:
- Profiles dashboard route and sidebar entry
- Profile lifecycle REST endpoints
- SOUL.md read/write support
- i18n labels and helper text updates
- Targeted profile API tests
Test plan:
- pytest tests/hermes_cli/test_web_server.py -k profile -q
- cd web && npm run build
Based on three live test runs against 346 agent-created skills on the
author's own setup (~6.5 min, opus-4.7, 86 API calls), the curator
prompt needed three sharpenings before it consistently produced real
umbrella consolidation instead of passive audit output:
**Umbrella-first framing.** The original 'decide keep/patch/archive/
consolidate' framing lets opus default to 'keep' whenever two skills
aren't byte-identical. The new prompt explicitly tells the reviewer
that pairwise distinctness is the wrong bar — the right question is
'would a human maintainer write this as N separate skills, or one
skill with N labeled subsections?' Expect 10-25 prefix clusters; merge
each into an umbrella via one of three methods.
**Three concrete consolidation methods.** (a) Merge into an existing
umbrella (patch the broadest skill, archive siblings); (b) Create a
new umbrella SKILL.md (skill_manage action=create); (c) Demote
session-specific detail into references/, templates/, or scripts/
under the umbrella via skill_manage action=write_file, then archive
the narrow sibling. This matches the support-file vocabulary the
review-prompt side already uses (PR #17213).
**Two observed bailouts pre-empted:** 'usage counters are zero so I
can't judge' (rule 4: judge on content, not use_count) and 'each has
a distinct trigger' (rule 5: pairwise distinctness is the wrong bar).
**Config-aware parent inheritance.** _run_llm_review() was building
AIAgent() without explicit provider/model, hitting an auto-resolve
path that returned empty credentials → HTTP 400 'No models provided'
against OpenRouter. Fork now inherits the user's main provider and
model (via load_config + resolve_runtime_provider) before spawning —
runs on whatever the user is currently on, OAuth-backed or
pool-backed included.
**Unbounded iteration ceiling.** max_iterations=8 was way too low for
an umbrella-build pass over hundreds of skills. A live pass takes
50-100 API calls (scanning, clustering, skill_view'ing candidates,
patching umbrellas, mv'ing siblings). Raised to 9999 — the natural
stopping criterion is 'no more clusters worth processing', not an
arbitrary tool-call budget.
**Tests updated:** test_curator_review_prompt_has_invariants accepts
DO NOT / MUST NOT and drops 'keep' from the required-verb set (the
umbrella-first prompt correctly deemphasizes 'keep' as a first-class
decision label since passive keep-everything is the failure mode
being prevented). Added test_curator_review_prompt_is_umbrella_first
asserting the umbrella framing, class-level thinking, references/
+ templates/ + scripts/ support-file mentions, and the 'use_count
is not evidence of value' pre-emption. Added
test_curator_review_prompt_offers_support_file_actions asserting
skill_manage action=create and action=write_file are both named.
**Live validation on author's setup:**
- Run 1 (old prompt): 3 archives, stopped after surveying — typical passive outcome
- Run 2 (consolidation prompt): 44 archives, 3 patches, surfaced the 50-skill mlops reorg duplicate bug but didn't umbrella
- Run 3 (this prompt): 249 archives + 18 new class-level umbrellas created, reducing agent-created skills from 346 → 118 with every archived skill's content preserved as references/ under its umbrella. Pinned skill untouched. Full report in PR description.
Long-running gateways need the curator to fire on cadence without
restarts. Piggy-back on the existing cron ticker thread (which already
runs image/document cache cleanup every hour on the same pattern)
instead of spawning a dedicated timer thread.
- New CURATOR_EVERY = 60 ticks (poll hourly at default 60s interval).
The inner config.interval_hours gate controls the real cadence, so
60 of these 60 hourly pokes are cheap no-ops and one runs the review.
- Removed the boot-time call added in the prior commit — the ticker
covers boot + every hour thereafter. Avoids double-running.
Handles the weekly-default-on-24/7-gateway gap flagged in review.
Weekly is closer to how skill churn actually works — most agent-created
skills don't change multiple times per day, so a daily review is pure
cost without benefit. Bumping the default to 7 days reduces aux-model
spend while still catching drift and staleness on the timescales that
matter (30d stale, 90d archive).
Changes:
- DEFAULT_INTERVAL_HOURS: 24 -> 168 (7 days)
- config.yaml default: interval_hours: 24 -> 24 * 7
- CLI status line renders as '7d' when interval is a whole-day multiple
- Test `test_old_run_eligible` decoupled from the exact default: it now
uses 2 * get_interval_hours() so future tweaks don't break it
Previous invariants only gated the primary entry points
(apply_automatic_transitions, archive_skill, CLI pin). Several paths
were unprotected:
- bump_view / bump_use / bump_patch / set_state / set_pinned wrote
usage records unconditionally, which is confusing noise in
.usage.json even though the review list filtered them out
- restore_skill did not check whether a bundled skill now shadows
the archived name
- CLI unpin was asymmetric with CLI pin — it had no gate
Fixes:
- _mutate() (the shared counter / state writer) now drops silently
when the skill is not agent-created. .usage.json never gains a
record for a bundled or hub-installed skill.
- restore_skill() refuses to restore under a name that is now
bundled or hub-installed (would shadow upstream).
- CLI unpin gate matches CLI pin.
New tests:
- 5 provenance-guard tests on skill_usage (one per mutator)
- 1 end-to-end test that hammers every mutator at a bundled skill
and a hub skill, asserts both are untouched on disk, and asserts
the sidecar stays clean
- 2 CLI tests proving pin/unpin refuse bundled skills symmetrically
64/64 tests passing (29 skill_usage + 27 curator + 8 new guards).
The LLM review prompt mentioned bespoke `archive_skill` and `pin_skill`
tools that are not registered as model tools. Swap the prompt to rely
on the real surface:
- skill_manage action=patch — for patching and consolidation
- terminal — to `mv` skill dirs into .archive/
Also drop `pin` from the model's decision list — pinning is a user
opt-out for `hermes curator pin <skill>`, not something the model
should do autonomously.
Decision list is now: keep / patch / consolidate / archive.
Tests updated: prompt-invariant test now asserts the existing tools
are referenced and that bespoke tool names do NOT appear. New test
prevents `pin` from being re-added as a model decision.
Adds the Curator — an auxiliary-model background task that periodically
reviews AGENT-CREATED skills and keeps the collection tidy: tracks usage,
transitions unused skills through active → stale → archived, and spawns
a forked AIAgent to consolidate overlaps and patch drift.
Default: enabled, inactivity-triggered (no cron daemon). Runs on CLI
startup and gateway boot when the last run is older than interval_hours
(default 24) AND the agent has been idle for min_idle_hours (default 2).
Invariants (all load-bearing):
- Never touches bundled or hub-installed skills (.bundled_manifest +
.hub/lock.json double-filter)
- Never auto-deletes — archive only. Archives are recoverable
via `hermes curator restore <skill>`
- Pinned skills bypass all auto-transitions
- Uses the aux client; never touches the main session's prompt cache
New files:
- tools/skill_usage.py — sidecar .usage.json telemetry, atomic writes,
provenance filter
- agent/curator.py — orchestrator: config, idle gating, state-machine
transitions (pure, no LLM), forked-agent review prompt
- hermes_cli/curator.py — `hermes curator {status,run,pause,resume,
pin,unpin,restore}` subcommand
- tests/tools/test_skill_usage.py — 29 tests
- tests/agent/test_curator.py — 25 tests
Modified files (surgical patches):
- tools/skills_tool.py — bump view_count on successful skill_view
- tools/skill_manager_tool.py — bump patch_count on skill_manage
patch/edit/write_file/remove_file; forget record on delete
- hermes_cli/config.py — add curator: section to DEFAULT_CONFIG
- hermes_cli/commands.py — add /curator CommandDef with subcommands
- hermes_cli/main.py — register `hermes curator` subparser via
register_cli() from hermes_cli.curator
- cli.py — /curator slash-command dispatch + startup hook
- gateway/run.py — gateway-boot hook (mirrors CLI)
Validation:
- 54 new tests across skill_usage + curator, all passing in 3s
- 346 tests across all touched files' neighbors green
- 2783 tests across hermes_cli/ + gateway/test_run_progress_topics.py green
- CLI smoke: `hermes curator status/pause/resume` work end-to-end
Companion to PR #16026 (class-first skill review prompt) — together
they form a loop: the review prompt stops near-duplicate skill creation
at the source, and the curator prunes/consolidates what still accumulates.
Refs #7816.
Relative entries in skills.external_dirs were resolved against the
process cwd via Path.resolve(), making them silently fail when Hermes
was launched from a different directory.
Resolve relative paths against get_hermes_home() for consistent
behavior across CLI, gateway, and cron contexts. Absolute paths
and env-var/tilde expansion are unchanged.
Commit 3c42064e made config.yaml the single source of truth for
TERMINAL_CWD, but the config bridge passes cwd values verbatim to
os.environ. When a user sets terminal.cwd: ~/ in config.yaml, the
literal string '~/'' reaches subprocess.Popen, which the kernel
rejects because it does not expand shell tilde syntax.
This patch adds three defensive layers:
1. gateway/run.py — expanduser at config bridge time so TERMINAL_CWD
is always an absolute path.
2. tools/terminal_tool.py — expanduser when reading TERMINAL_CWD in
_get_env_config(), guarding against stale or manually-set env vars.
3. tools/environments/local.py — expanduser in LocalEnvironment before
passing cwd to subprocess.Popen, the final safety net.
Includes regression tests in test_config_cwd_bridge.py for nested
terminal.cwd, top-level cwd alias, and precedence ordering.
Refs: 3c42064e
Finish the Copilot review cleanup for lazy prompt submission:
- prompt.submit now claims session.running before returning success, preserving
the existing RPC-level session busy error so the frontend can queue.
- agent-init timeout/failure now emits a normal error event instead of writing a
second JSON-RPC response for an already-settled request id.
Tests:
- python -m py_compile tui_gateway/server.py tui_gateway/entry.py
- cd ui-tui && npm run type-check && npm run build
- scripts/run_tests.sh tests/tui_gateway/test_protocol.py::test_sess_found tests/tools/test_code_execution_modes.py tests/tools/test_code_execution.py
- cd ui-tui && npm test -- --run src/__tests__/useSessionLifecycle.test.ts src/__tests__/useConfigSync.test.ts
The known-key splitter in `_sanitize_env_lines` used substring matching
to find concatenated KEY=VALUE pairs. When a registered key was a suffix
of another (LM_API_KEY is a suffix of GLM_API_KEY), the shorter key's
needle would match inside the longer one, causing the sanitizer to
rewrite `GLM_API_KEY=...` as `G\nLM_API_KEY=...` and silently break
Z.AI/GLM auth (and similarly `GLM_BASE_URL` -> `G\nLM_BASE_URL`).
Drop matches whose needle range is fully contained within a longer
overlapping match. Two regression tests cover the suffix-collision case
and confirm a real concatenation that happens to start with the longer
key still splits where it should.
Fixes#17138
Respond to Copilot's lazy-start review: session metadata/history/usage do not
need a constructed AIAgent, so keep them on the no-wait session path. This
preserves the deferred startup model and avoids blocking simple session RPCs on
agent initialization.
Tests:
- python -m py_compile tui_gateway/server.py tui_gateway/entry.py
- cd ui-tui && npm run type-check && npm run build
- scripts/run_tests.sh tests/tui_gateway/test_protocol.py::test_sess_found tests/tools/test_code_execution_modes.py tests/tools/test_code_execution.py
- cd ui-tui && npm test -- --run src/__tests__/useSessionLifecycle.test.ts src/__tests__/useConfigSync.test.ts
Classic CLI exposes ``/reload`` (re-reads ~/.hermes/.env into
``os.environ`` via ``hermes_cli.config.reload_env``) so newly added API
keys take effect without restarting the session. The TUI was missing
the parity command, so users had to Ctrl+C out and ``hermes --tui``
again whenever they added or rotated a credential.
Three small wires:
* New ``reload.env`` JSON-RPC method in ``tui_gateway/server.py`` that
delegates to ``hermes_cli.config.reload_env`` and returns the count
of vars updated.
* New ``/reload`` slash command in ``ui-tui/src/app/slash/commands/ops.ts``
matching the existing ``/reload-mcp`` pattern (native RPC, no slash
worker).
* Drop ``cli_only=True`` from the ``reload`` ``CommandDef`` in
``hermes_cli/commands.py`` so help/menus surface it in the TUI too.
``reload_env`` itself is environment-agnostic.
Same caveat as classic CLI: the *currently constructed* agent's
credential pool / provider routing does not auto-rebuild. Users who
want a brand-new credential resolution should follow with ``/new``.
Tests:
* New ``test_reload_env_rpc_calls_hermes_cli_reload_env`` confirms
RPC delegates and reports the count.
* New ``test_reload_env_rpc_surfaces_errors`` confirms exceptions are
rendered as JSON-RPC errors.
* ``createSlashHandler.test.ts`` slash-parity matrix extended with
``['/reload', 'reload.env', {}]`` so we can't regress the routing.
Validation:
scripts/run_tests.sh tests/test_tui_gateway_server.py — 92/92.
scripts/run_tests.sh tests/hermes_cli/test_commands.py — 128/128.
cd ui-tui && npm run type-check — clean; npm test --run — 390/390.
After PR #7885 (97b0cd51e) added content-side segment breaks for
natural mid-turn assistant messages, the tool-progress task in
gateway/run.py was not updated to match. progress_msg_id and
progress_lines persisted for the whole run, so after a tool batch
produced bubble B1 followed by content bubble C1, the next tool.started
kept editing the OLD bubble B1 above C1 — making the chat appear out
of order on Telegram, Discord, and Slack.
Add on_new_message callback to GatewayStreamConsumer, fired at the
four sites where a fresh content bubble lands on the platform:
- _send_or_edit first-send branch (NOT edits)
- _send_commentary
- _send_new_chunk (overflow split)
- each successful chunk of _send_fallback_final
Gateway supplies a lambda that enqueues ('__reset__',) into the
progress_queue. send_progress_messages() handles the marker in both
the main loop and the CancelledError drain path, clearing
progress_msg_id, progress_lines, and the dedup state so the next
tool.started opens a fresh bubble below the new content.
Result: each tool batch appears in chronological order below the
preceding content. When no content appears between tool batches,
tools still group in one bubble (CLI-style compactness).
Co-authored-by: teknium1 <teknium@users.noreply.github.com>
init_session() runs a login shell bootstrap that sources profile scripts
(.bashrc, .bash_profile, etc.) before capturing pwd. If any profile
script changes the working directory, the captured cwd overwrites the
configured terminal.cwd value — so terminal commands run in the wrong
directory despite the TUI banner showing the configured path.
Add an explicit 'builtin cd' to the configured cwd in the bootstrap
script, after profile sourcing but before pwd capture, ensuring the
configured terminal.cwd is always what gets recorded.
Fixes#14044
* Reject unsupported schemes (anything outside http/https/ws/wss) in
cli.py /browser connect before probing or persisting, matching the
gateway's existing 4015 path.
* Defend gateway browser.manage against `{"url": null}` and
non-string urls: empty/null falls back to DEFAULT_BROWSER_CDP_URL,
non-string returns a 4015 instead of slipping into the generic
5031 catch via TypeError on `"://" in url`.
* Add regression tests for both null-url fallback and non-string
rejection.
* Gate `browser.progress` emit on truthy `session_id`. The TUI
prints `messages` from the response when there's no session, so
emitting events too would double-render. Now: with a session →
events stream live; without one → bundled messages only.
* Resolve `system = platform.system()` once in `_browser_connect`
and thread it through `try_launch_chrome_debug` and
`_failure_messages` → `manual_chrome_debug_command`, so the
generated hint is consistent (and tests are deterministic) on
any host.
* Add `test_browser_manage_connect_no_session_skips_progress_events`
to lock in the gating behavior.
Fixes from Copilot's two passes on PR #17238:
* Validate parsed URL once: reject missing host, invalid port, and
unsupported scheme up front so malformed inputs (e.g. http://:9222
or http://localhost:abc) don't fall through to a generic 5031.
* Tighten _is_default_local_cdp to require a discovery-style path so
ws://127.0.0.1:9222/devtools/browser/<id> is not collapsed to bare
http://127.0.0.1:9222 (which would lose the path and break the
connect).
* Move browser.manage into _LONG_HANDLERS so the up-to-10s
launch-and-retry loop runs on the RPC pool instead of blocking the
main dispatcher.
* try_launch_chrome_debug uses Windows-appropriate detach kwargs
(creationflags=DETACHED_PROCESS|CREATE_NEW_PROCESS_GROUP) instead
of POSIX-only start_new_session=True.
* manual_chrome_debug_command uses subprocess.list2cmdline on
Windows so the printed instruction is cmd.exe-compatible.
* Mirror host/port validation in cli.py /browser connect so the
classic CLI never persists an invalid BROWSER_CDP_URL.
Split browser.manage into a small dispatcher with named connect/disconnect
helpers, fold _http_ok / _probe_urls / _normalize_cdp_url out of the nested
probe loop, collapse the failure-message scaffolding, and DRY the chrome
candidate path tables. Behaviour and event shape unchanged.
Emit browser.progress JSON-RPC notifications during the connect work and render them in the TUI as system transcript lines, so users see the same step-by-step status the base CLI prints instead of nothing for ~1m followed by a final result.
Return CLI-style browser connect status messages from the gateway and render them in the TUI so local Chrome launch attempts are visible instead of ending in a silent delayed failure.
Detect an actual Chrome/Chromium executable before printing a manual CDP launch command, including common WSL-mounted Windows browser paths, so /browser connect does not suggest google-chrome when it is unavailable.
Share Chrome CDP launch helpers between the classic CLI and TUI so default /browser connect uses loopback consistently, retries local Chrome launch, and reports a copyable manual-start command instead of claiming a dead connection.
Clean up the remaining review nits:
- let the deferred @hermes/ink import retry after a transient failure instead
of memoizing a rejected promise forever
- keep memory-monitor in-flight state inside a finally so future exceptions
cannot suppress that memory level indefinitely
- use read_raw_config for the TUI MCP cold-start probe instead of full
load_config()
- keep input.detect_drop for explicit relative path prefixes (./ and ../)
while preserving the no-RPC fast path for ordinary plain prompts
Tests:
- python -m py_compile tui_gateway/server.py tui_gateway/entry.py
- cd ui-tui && npm run type-check && npm run build
- scripts/run_tests.sh tests/tui_gateway/test_protocol.py::test_sess_found tests/tools/test_code_execution_modes.py tests/tools/test_code_execution.py
- cd ui-tui && npm test -- --run src/__tests__/useSessionLifecycle.test.ts src/__tests__/useConfigSync.test.ts
A cleanup review found that adding prompt.submit to _LONG_HANDLERS made the RPC
pool own the full first-turn wait even though the handler itself already spawns
a turn thread. Keep prompt.submit inline and make it return immediately:
- look up the session without waiting
- kick the lazy agent build
- spawn a short waiter thread that blocks on agent_ready, then starts the
existing turn dispatcher
This keeps stdin dispatch responsive, avoids occupying a bounded pool worker for
a normal chat turn, and preserves the lazy-start hydration behavior.
Tests:
- python -m py_compile tui_gateway/server.py
- cd ui-tui && npm run type-check && npm run build
- scripts/run_tests.sh tests/tui_gateway/test_protocol.py::test_sess_found tests/tools/test_code_execution_modes.py tests/tools/test_code_execution.py
- cd ui-tui && npm test -- --run src/__tests__/useSessionLifecycle.test.ts src/__tests__/useConfigSync.test.ts
Copilot correctly flagged two concurrency windows:
- memoryMonitor could re-enter while awaiting the lazy @hermes/ink import or
heap dump, producing duplicate imports/dumps under sustained pressure.
- _start_agent_build used a check-then-set guard without synchronization, so
concurrent agent-backed RPCs could start duplicate agent builders.
Fix both with single-flight guards: cache the dynamic import promise and track
per-level dump in-flight state in memoryMonitor, and protect the TUI agent build
flag with a per-session lock.
Tests:
- python -m py_compile tui_gateway/server.py
- cd ui-tui && npm run type-check && npm run build
- cd ui-tui && npm test -- --run src/__tests__/useSessionLifecycle.test.ts src/__tests__/useConfigSync.test.ts
- scripts/run_tests.sh tests/tui_gateway/test_protocol.py::test_sess_found tests/tools/test_code_execution_modes.py tests/tools/test_code_execution.py
The lazy startup panel could remain stuck on the placeholder when no first
prompt was submitted because agent construction only started from _sess(). Keep
session.create cheap, but schedule _start_agent_build shortly after returning
the placeholder so tools/skills hydrate automatically.
Also replace the ugly placeholder bar rows with compact unicode-animations
braille loaders for the tools and skills sections.
Tests:
- python -m py_compile tui_gateway/server.py
- cd ui-tui && npm run type-check && npm run build
- cd ui-tui && npm test -- --run src/__tests__/useSessionLifecycle.test.ts src/__tests__/useConfigSync.test.ts
- scripts/run_tests.sh tests/tui_gateway/test_protocol.py::test_sess_found tests/tools/test_code_execution_modes.py tests/tools/test_code_execution.py
Match classic CLI perceived startup behavior: show the TUI shell and composer
before constructing the full AIAgent. session.create now returns a lightweight
placeholder session with lazy=true and no longer starts _make_agent eagerly.
The first method that needs the agent triggers _start_agent_build() via _sess();
prompt.submit is routed through the RPC worker pool so that the initial wait for
agent construction does not block the stdio dispatcher.
The intro panel renders skeleton rows for tools/skills while the real
session.info payload is absent, then hydrates to the real tools/skills panel once
AIAgent initialization completes. Also skip the startup /voice status probe and
avoid the input.detect_drop RPC for ordinary plain-text prompts to keep early
startup/first-submit paths cheap.
Measurements on macOS Terminal.app:
- Previous full ready p50 after earlier PR commits: ~1537ms
- Lazy skeleton panel p50: ~794ms
- Original baseline full ready p50: ~1843ms
So the visible startup surface is now ~743ms faster than the prior PR state and
~1.05s faster than the original baseline. First prompt still pays the same agent
construction cost if it races the background/skeleton state, matching classic
CLI's deferred behavior.
Tests:
- python -m py_compile tui_gateway/server.py
- cd ui-tui && npm run type-check && npm run build
- scripts/run_tests.sh tests/tui_gateway/test_protocol.py::test_sess_found tests/tools/test_code_execution_modes.py tests/tools/test_code_execution.py
- cd ui-tui && npm test -- --run src/__tests__/useSessionLifecycle.test.ts src/__tests__/useConfigSync.test.ts
The background skill-review prompts (_SKILL_REVIEW_PROMPT and the **Skills**
half of _COMBINED_REVIEW_PROMPT) steered the reviewer toward passive
behavior — most passes concluded 'Nothing to save.' even when the session
produced real lessons. User-preference corrections (style, format,
legibility, verbosity) were especially lost: they were read as memory
signals only, so skills never carried the fix.
This rewrite changes the stance:
- **Active-update bias.** The reviewer now treats inaction as a missed
learning opportunity. 'Nothing to save.' remains an explicit escape
but is no longer framed as the most-common outcome.
- **User-preference corrections are first-class skill signals.** Style,
tone, format, legibility, verbosity complaints — and the actual
phrasings users use ('stop doing X', 'this is too verbose', 'I hate
when you Y', 'remember this') — now warrant patching the skill that
governs the task, not just writing to memory.
- **Loaded-skill-first preference order.** When a skill was loaded via
/skill-name or skill_view during the session, the reviewer patches
THAT one first. It was in play; it's the right place.
- **Four-step ladder: patch-loaded → patch-umbrella → support-file →
create.** Support files are explicitly enumerated as three kinds:
* references/<topic>.md — session-specific detail OR condensed
knowledge banks (quoted research, API docs excerpts, domain notes)
* templates/<name>.<ext> — starter files to copy and modify
* scripts/<name>.<ext> — statically re-runnable actions
- **Name-veto for CREATE.** New skill names MUST be class-level — no PR
numbers, error strings, codenames, library-alone names, or session
artifacts ('fix-X / debug-Y / audit-Z-today'). If the proposed name
only fits today's task, fall back to one of the patch/support-file
options.
- **Memory scope clarified.** 'who the user is and what the current
situation and state of your operations are' — MEMORY.md is
situational/state, USER.md is identity/preferences.
- **Curator handoff.** Reviewer flags overlap; the background curator
handles consolidation at scale. Single-session reviewer doesn't
attempt umbrella-rebalancing.
Tests: tests/run_agent/test_review_prompt_class_first.py upgraded to
assert the new behavioral contracts (active bias, user-correction
signals, loaded-skill-first, support-file kinds, name-veto, memory
framing, curator handoff). 17 tests, all pass.
Co-authored-by: teknium1 <teknium@users.noreply.github.com>
Adds a 'pretext' skill under skills/creative/ for building cool browser
demos with @chenglou/pretext — the 15KB DOM-free text-layout library by
Cheng Lou.
The skill documents pretext as a creative primitive (not plumbing): text
flowing around obstacles, text-as-geometry games, proportional ASCII
surfaces, shatter/particle typography, editorial multi-column, kinetic
type, and multiline shrink-wrap. Each pattern pairs with copy-pasteable
snippets in references/patterns.md.
Two single-file HTML templates, both verified in a browser:
templates/hello-orb-flow.html
Minimal starter: long paragraph flows around a mouse-tracked orb
using layoutNextLineRange + a per-row corridor-width function.
templates/donut-orbit.html
Full 3D Sloane torus with orbit controls (drag to rotate, scroll to
zoom, idle auto-rotate). Each 'luminance pixel' is a real grapheme
sampled in reading order from a prose corpus via pretext's
prepareWithSegments + layoutWithLines + Intl.Segmenter. Amber-on-
black CRT aesthetic, z-buffer keyed by screen cell, 60fps.
Related skills: p5js, claude-design, excalidraw, architecture-diagram.
Three modules independently implemented the same "preserve head+tail of
a secret, mask the middle" logic with slightly different behaviors that
had started to drift:
hermes_cli/config.py redact_key — 12-char floor, 4+4, DIM '(not set)'
hermes_cli/status.py redact_key — 12-char floor, 4+4, plain '(not set)' ← drift
hermes_cli/dump.py _redact — 12-char floor, 4+4, empty string
The visible bug: 'hermes status' displayed the '(not set)' placeholder
in plain text while 'hermes config' showed it in dim text. Same concept,
inconsistent UI.
Introduces mask_secret() in agent/redact.py as the canonical helper,
with head/tail/floor/placeholder/empty kwargs. The three call sites
become one-line wrappers that differ only in the 'empty' handling:
config.redact_key → mask_secret(k, empty=color('(not set)', Colors.DIM))
status.redact_key → mask_secret(k, empty=color('(not set)', Colors.DIM))
dump._redact → mask_secret(v) # empty → ''
agent.redact._mask_token (log redactor, different policy: 18-char floor,
6+4 visible, '***' on empty) also ports to mask_secret but retains its
own empty-case handling to preserve the historical '***' return.
Net: the three display-time redactors now agree on formatting, the
canonical helper lives in one place, and future tweaks (e.g. adding
bullet-point masking, changing the head/tail widths) happen once.
Verified:
- 3/3 tests/hermes_cli/test_web_server.py::TestRedactKey pass
- 89/89 agent/tests/test_redact.py + tests/tools/test_browser_secret_exfil.py
+ tests/hermes_cli/test_redact_config_bridge.py pass
- Live 'hermes status', 'hermes config', 'hermes dump' all render the
same way they did before (verified against actual env with real
keys: OpenRouter, Firecrawl, Browserbase, FAL, Tinker all show
'prefix...suffix'; Kimi shows '***' at <12 chars; unset shows
'(not set)' uniformly).
Co-authored-by: teknium1 <teknium@users.noreply.github.com>
TUI session readiness was still laggy after the gateway-ready fixes. Profiling
session.create -> session.info showed the slow phase is background AIAgent
construction (~1.1s). A cProfile run of tui_gateway.server::_make_agent showed
model_tools/tool discovery importing tools.code_execution_tool, whose
module-level EXECUTE_CODE_SCHEMA calls _get_execution_mode(), which imported
cli.CLI_CONFIG.
That pulled the classic interactive CLI stack (prompt_toolkit/Rich and REPL
setup) into every agent startup path, including hermes --tui where it is not
used. Replace that with hermes_cli.config.read_raw_config(), which is cached and
reads only the raw code_execution section. Existing defaults still apply when
the key is absent.
Measurements on macOS Terminal.app:
- import run_agent: ~466ms -> ~347ms
- model_tools import: ~418ms -> ~272ms
- _make_agent: ~1452ms -> ~1239ms
- session.create -> session.info: ~1167ms -> ~999ms
- full hermes --tui ready p50: ~1655ms -> ~1537ms
Tests:
- scripts/run_tests.sh tests/tools/test_code_execution_modes.py tests/tools/test_code_execution.py
Match the buffered-stdin rearm cadence to IN_PASTE state so large pastes do not spin the normal escape timeout while waiting for readable data to drain.
Keep the latest prompt sticky while the viewport is in live assistant output beyond history, and clear stale sticky state at the real bottom using fresh scroll height.
detect_dangerous_command() and detect_hardline_command() were calling
re.search(pattern, text, re.IGNORECASE | re.DOTALL) inline — Python's
re._cache (512 patterns) amortizes compile cost on the warm path, but:
1. The first terminal() call per process pays the full compile fan-out
for all 59 patterns (12 HARDLINE + 47 DANGEROUS). Measured at
~2.6 ms per detect_dangerous_command() call after re.purge().
2. The re._cache is LRU — unrelated regex work elsewhere in the agent
(response parsing, text normalization, etc.) can evict our patterns
and silently re-compile them on the next terminal() call.
Precompiling at module load eliminates both costs:
detect_dangerous_command:
cold 2.613 ms → 0.298 ms (-88%)
warm 0.042 ms → 0.004 ms (-90%)
detect_hardline_command:
cold ~0.6 ms → 0.006 ms
warm 0.011 ms → 0.002 ms
Savings are per terminal() call. Agents with heavy terminal use see
compound savings; the bigger value is the stability guarantee (no
re._cache eviction can silently re-introduce the 2.6 ms cold cost
mid-session).
Implementation:
- HARDLINE_PATTERNS_COMPILED and DANGEROUS_PATTERNS_COMPILED built at
module load from the existing (pattern, description) tuples, using
shared _RE_FLAGS = re.IGNORECASE | re.DOTALL.
- detect_* functions now iterate the compiled list and call pattern_re.search(text).
- Original HARDLINE_PATTERNS and DANGEROUS_PATTERNS lists kept as-is
(other code in the file uses them for key derivation /
_PATTERN_KEY_ALIASES).
Verified:
- 160/161 tests/tools/test_approval*.py pass (1 pre-existing heartbeat
test flake on main).
- 349/349 tests/tools/ 'approval or terminal or dangerous' pass.
- Live hermes chat smoke: 3 benign terminal commands + 1 rm -rf /tmp/
(clarify prompt fired — approval path still works) + 1 sudo (sudo
password prompt fired — DANGEROUS pattern match still works). 23
log lines in the smoke window, zero errors.
Co-authored-by: teknium1 <teknium@users.noreply.github.com>
Address two Copilot review comments on PR #17175.
- `wrapForFrac` doc said "additive operators or whitespace" but the
implementation also matches `*` and `/`. The wider behaviour is the
one we want (nested products and fractions need parens to disambiguate
inline `/`), so the doc is updated to match instead of tightening the
regex.
- `fenceOpenAt` was flagged as "overly conservative" vs. `markdown.tsx`,
which falls back to paragraph rendering for unclosed `$$` openers.
Mirroring that fallback in the streaming chunker would prematurely
commit a paragraph rendering of the unclosed opener to the monotonic
stable prefix, where it would be frozen and become wrong the moment
the closer streams in. The asymmetry is deliberate; document why so
it isn't "fixed" again later.
Made-with: Cursor
Replace the removed built-in boot-md hook (#17093) with a how-to that
shows users how to wire up the same behavior themselves via the hooks
system. Uses _resolve_gateway_model() + _resolve_runtime_agent_kwargs()
so the example works against custom endpoints and OAuth providers,
not just the aggregator defaults that the old built-in silently assumed.
Co-authored-by: teknium1 <teknium@users.noreply.github.com>
Validate configured providers against both Hermes runtime provider ids and
catalog-normalized provider ids. This keeps providers like ai-gateway from
being rejected after catalog resolution maps them to models.dev ids.
Keep credential checks and vendor-slug warnings anchored to the runtime id
so doctor reports actionable provider names in follow-up diagnostics.
Two amplifying optimizations to per-turn overhead in the gateway:
1. get_tool_definitions() memoization (model_tools.py)
Keyed on (frozenset(enabled), frozenset(disabled),
registry._generation, config.yaml mtime+size). Only active when
quiet_mode=True (which is every hot-path caller — gateway,
AIAgent.__init__); quiet_mode=False keeps the existing print side
effects. Cached path returns a shallow-copy list sharing read-only
schema dicts.
Measured: 7.5 ms → 0.01 ms per call (~750× speedup). Gateway
constructs fresh AIAgent per message, so this saves ~7 ms/turn before
any LLM work.
2. check_fn() TTL cache (tools/registry.py)
check_fn callables like check_terminal_requirements probe external
state (Docker daemon, Modal SDK, playwright binary). For a long-lived
process, hitting them on every get_definitions() pass was pure waste
— external state changes on human timescales. 30 s TTL so env-var
flips (hermes tools enable X) propagate within a turn or two without
explicit invalidation.
Measured: first call 7.5ms → 1.6ms (check_fn probes now dominate);
subsequent calls ~0.01ms via the upstream memoization.
Invalidation surface:
- registry._generation bumps on register/deregister/register_toolset_alias,
invalidating the memoized definitions automatically.
- config.yaml mtime in the cache key captures user-visible config edits
affecting dynamic schemas (execute_code mode, discord allowlist).
- invalidate_check_fn_cache() exposed for explicit flushes (e.g. after
hermes tools enable/disable).
- tests/conftest.py autouse fixture clears both caches before every test
so env-var monkeypatches don't see stale results.
Also fixes a regression from PR #17046 that I missed:
- tools/web_tools.py — Firecrawl was removed from module scope by the
lazy import, breaking 8 tests that patch 'tools.web_tools.Firecrawl'.
Applied the same _FirecrawlProxy pattern used in auxiliary_client/
run_agent for OpenAI (module-level proxy that looks like the class
but imports the SDK on first call/isinstance; patch() replaces the
attribute as usual).
Verified:
- 49/49 tests/tools/test_web_tools_config.py pass (was 8 failing on main)
- 68/68 tests/tools/test_homeassistant_tool.py pass (was 1 failing in
the full suite due to check_fn TTL cross-test pollution; fixed by
the autouse fixture)
- 3887/3895 tests/tools/ (8 pre-existing fails: 2 delegate, 1 mcp
dynamic discovery, 5 mcp structured content — all confirmed on main)
- 2973/2976 tests/agent/ + tests/run_agent/ (3 pre-existing fails)
- 868/868 tests/run_agent/ (excluding test_run_agent.py which has
pre-existing suite-level issues)
- Live smoke: 2 turns + /model switch + tool calls, zero errors in
agent.log session window.
Co-authored-by: teknium1 <teknium@users.noreply.github.com>
Two targeted fixes on the critical path from `hermes --tui` launch to
`gateway.ready`:
1. **Defer `@hermes/ink` import in memoryMonitor.ts.** The static top-level
import dragged the full ~414KB Ink bundle (React + renderer + all
components/hooks) onto the critical path *before* `gw.start()` could
spawn the Python gateway — serialising ~155ms of Node work in front of
it on every launch. `evictInkCaches` only runs inside the 10-second
tick under heap pressure, so it moves to a lazy dynamic import. First
tick hits the ESM cache because the app entry has long since imported
`@hermes/ink`.
2. **Gate `tools.mcp_tool` import on config in tui_gateway/entry.py.**
Importing the module transitively pulls the MCP SDK + pydantic + httpx
+ jsonschema + starlette formparsers (~200ms). The overwhelming
majority of users have no `mcp_servers` configured, so this runs for
nothing. A cheap `load_config()` check (~25ms) skips the 200ms import
when no servers are declared, with a conservative fallback to the old
behaviour if the config probe itself fails.
## Measurements (macOS Terminal.app, Apple Silicon, n=12)
| Metric | Before (p50) | After (p50) | Δ |
|----------------------------|--------------|-------------|----------|
| Python gateway boot alone | 252–365ms | 105–151ms | −180ms |
| `hermes --tui` banner paint | 686ms | 665ms | −21ms |
| `hermes --tui` → ready | **1843ms** | **1655ms** | **−188ms (−10.2%)** |
| `hermes --tui` → ready p90 | 1932ms | 1778ms | −154ms |
| stdev (ready) | 126ms | 83ms | also more consistent |
## Tests
- `scripts/run_tests.sh tests/tui_gateway/ tests/tools/test_mcp_tool.py`:
195 passed. (The one pre-existing failure in
`test_session_resume_returns_hydrated_messages` reproduces on main —
unrelated, it's a mock-DB kwarg mismatch.)
- `ui-tui` vitest: 430 tests, all pass.
- `npm run type-check` in ui-tui: clean.
## Notes
- Node-side first paint ("banner") didn't move meaningfully because that
latency is dominated by Ink's render pipeline + React mount, not by
which imports load first.
- The win shows up entirely in the time from banner to `gateway.ready`
— exactly where we expected it, since both fixes shorten the Python
gateway's boot path or let it overlap more with Node startup.
- No user-visible behaviour change. Memory monitoring still fires every
10s; MCP still works when `mcp_servers` is configured.
* fix(tui): honor documented mouse_tracking config key
The TUI runtime was reading display.tui_mouse while docs and user-facing
examples pointed users at display.mouse_tracking. That made persistent
mouse-disable config look like a no-op for users trying to restore native
terminal selection/copy behavior on Linux/SSH/tmux terminals.
Use display.mouse_tracking as the canonical key, keep display.tui_mouse as
a legacy fallback, and have /mouse write the documented key. Both gateway
config.get and client-side config sync now share the same precedence: the
canonical key wins, then the legacy key, then default on.
* review(copilot): align mouse tracking config coercion
- Load gateway config once before deriving display.mouse_tracking state.
- Use key-presence precedence on the TUI client too, so canonical
mouse_tracking wins over legacy tui_mouse even when the value is null.
- Treat numeric 0 as disabled on both gateway and client, matching the
existing string "0" handling.
- Widen ConfigDisplayConfig mouse fields because config.get full returns raw
YAML, not normalized booleans.
This PR groups the TUI fixes that restore macOS Terminal usability and clean up the theme/composer regressions:
- copy transcript selections on macOS drag-release so Terminal.app users can copy while mouse tracking is enabled
- copy composer selections on macOS drag-release; composer selection is internal to TextInput and does not use the global Ink selection bus
- keep IDE Cmd+C forwarding setup macOS-only, and make keybinding conflict checks respect simple when-clause overlap/negation
- force truecolor before chalk initializes (unless NO_COLOR / FORCE_COLOR / HERMES_TUI_TRUECOLOR opt-outs apply) so the default banner keeps its gold/amber/bronze gradient in Terminal.app
- move TUI surfaces onto semantic theme tokens and preserve skin prompt symbols as bare tokens with renderer-owned spacing
- render focused placeholders as dim hint text in TTY mode instead of inverse/selected-looking synthetic cursor text
Round 1 of #17174 hit `nix-lockfile-check` failure. Root cause was
NOT a stale hash — the primary `nix (ubuntu-latest)` and
`nix (macos-latest)` builds passed. GitHub's Magic Nix Cache returned
HTTP 418 (rate-limited / throttled) mid-run, so the rebuild bailed
with `some outputs of '/nix/store/...-npm-deps.drv' are not valid,
so checking is not possible` — no `got:` line for the script to
extract.
The script then incorrectly treated this as 'build failed with no
hash mismatch' and exited 1, breaking the lint on every PR whenever
the cache is throttled.
Now we recognize the throttling/cache-disabled signature and skip
that entry with a warning. A real stale hash still surfaces in the
primary `.#$ATTR` build (separate CI job), so we don't lose
coverage.
`web/package-lock.json` was updated by the design-system refactor
(merged via #17007 + follow-ups: spinner / select / badges / buttons)
without bumping `nix/web.nix::npmDeps.hash`, breaking nix builds on
every PR + main since 2026-04-28T18:46.
Hash sourced from the actual `Check flake` failure output:
specified: sha256-AahWmJ9gDQ9pMPa1FYwUjYdO2mOi6JM9Mst27E0vp68=
got: sha256-+B2+Fe4djPzHHcUXRx+m0cuyaopAhW0PcHsMgYfV5VE=
Standalone single-file fix so it can land fast and clear nix on
every other open PR.
* feat(tui): pluggable busy-indicator styles (kaomoji/emoji/unicode/ascii)
The status-bar `FaceTicker` rotated through wide-and-variable kaomoji
glyphs (`(｡•́︿•̀｡)`, `( ͡° ͜ʖ ͡°)`, …) every 2.5s. Real display widths range
from ~5 to ~16 columns, so the rest of the bar (cwd, ctx %, voice,
bg counter) shifted on every cycle. Padding the verb alone (#17116)
helped but didn't address the dominant jitter source — the glyph
itself.
Add four indicator styles, configurable + hot-swappable:
* `kaomoji` (default — preserves the existing vibe; verb is now
pad-stable so the only width churn left is the kaomoji itself).
* `emoji` — single 2-col emoji frame (`⚕ 🌀🤔✨🍵🔮`).
* `unicode` — `unicode-animations` braille spinner (1-col, smooth).
* `ascii` — `| / - \` (1-col, max compat).
Wires:
* `display.tui_status_indicator` in `DEFAULT_CONFIG` (default
`kaomoji`).
* New JSON-RPC `config.set/get indicator` keys, narrow allow-list.
* `applyDisplay` reads the field and patches `UiState.indicatorStyle`,
so the existing `mtime` poll picks up `~/.hermes/config.yaml` edits
within ~5s without a TUI restart.
* `/indicator [style]` slash command (alias `/indicator-style`,
subcommand completion `kaomoji|emoji|unicode|ascii`). Bare form
shows the current style; setter fires `config.set` and
optimistically `patchUiState({ indicatorStyle })` so the live TUI
swaps immediately, matching the `/skin` UX.
* `CommandDef("indicator", ..., subcommands=...)` so classic CLI
autocomplete + TUI `complete.slash` both surface it.
* `FaceTicker` decouples spinner cadence from verb cadence — the
glyph runs at the spinner's authored interval (or `FACE_TICK_MS`
for kaomoji), the verb stays on the original 2.5s cycle, and both
re-arm cleanly when style changes.
Tests:
* `normalizeIndicatorStyle` rejects unknown / non-string input.
* `applyDisplay → tui_status_indicator` covers fan-out + fallback.
* `/indicator <style>` hot-swaps `UiState.indicatorStyle` after a
successful `config.set`.
* `/indicator sparkle` rejects with the usage hint and never hits
the gateway.
* Slash-parity matrix gets `'/indicator'` → `config.get`.
Validation:
cd ui-tui && npm run type-check — clean; npm test --run — 398/398.
scripts/run_tests.sh tests/test_tui_gateway_server.py
tests/hermes_cli/test_commands.py — 220/220.
* chore(tui): drop /indicator-style alias to declutter autocomplete
* fix(tui): drop verb-width pad — /indicator handles glyph jitter directly
* fix(tui): unicode indicator style hides the verb (cleanest option)
* refactor(tui): single source of truth for INDICATOR_STYLES; cleaner error format
Round 1 Copilot review on PR #17150:
- Exported `INDICATOR_STYLES` const tuple from `interfaces.ts`;
`IndicatorStyle` union type is derived from it. `useConfigSync`
builds its validation Set from the tuple, and `session.ts` uses it
for both the usage hint and the runtime allow-list — adding/removing
a style now touches one line.
- Backend `config.set indicator` error message: switched
`sorted(allowed)` list repr to `pick one of ascii|emoji|kaomoji|unicode`
(matches the TUI usage hint), and reports the normalized `raw`
instead of the original `value`. Backend allowed tuple now has a
comment pointing back at `INDICATOR_STYLES` so the two stay aligned.
Note: kept the verb portion unpadded per design intent — fixed-width
padding was the exact UX the `/indicator` command was added to remove.
Stable width comes from the glyph; verbs cycling is part of the kawaii
aesthetic. Reply on the verb thread will explain.
* fix(tui): drop type collapse + gate verb timer + DEFAULT_INDICATOR_STYLE
Round 2 Copilot review on PR #17150:
- `tui_status_indicator?: 'ascii' | ... | string` collapses to `string`
in TS — consumers got no narrowing. Documented as plain `string` with
a comment about runtime validation via `normalizeIndicatorStyle`.
- `FaceTicker` always started a 2.5s verb interval, even for the
`unicode` style which hides the verb entirely. Now gated on
`showVerb` from `renderIndicator` — `unicode` stays calm.
Pre-emptive self-review (avoid round 3):
- Three call sites duplicated the literal `'kaomoji'` default
(uiStore, normalizeIndicatorStyle, slash command). Added
`DEFAULT_INDICATOR_STYLE` to interfaces.ts and threaded it through
so changing the default touches one line.
* fix(tui-gateway): normalize config.get indicator output to match TUI render
Round 4 Copilot review on PR #17150: `config.get` for `indicator`
returned the raw `display.tui_status_indicator` value without
validation, so a hand-edited config.yaml with stray casing or an
unknown style would leave `/indicator` printing one thing while
the TUI rendered the kaomoji default (frontend's
`normalizeIndicatorStyle` does this normalization on receive).
Lifted the allow-list to module scope as `_INDICATOR_STYLES` /
`_INDICATOR_DEFAULT`, reused by both `config.set` and `config.get`.
Comment notes the alignment with `INDICATOR_STYLES` /
`DEFAULT_INDICATOR_STYLE` in interfaces.ts so adding/removing a
style is a one-line change on each end.
Tests cover: known value verbatim, casing/whitespace normalize,
unknown→default, unset→default.
* fix(tui-gateway): preserve falsy-input diagnostics in config.set indicator error
Round 5 Copilot review on PR #17150: `raw = str(value or "").strip().lower()`
collapsed any falsy non-string (`0`, `False`, `[]`) to empty string,
so the error message read `unknown indicator: ` with nothing after —
losing the original input.
Switched to `("" if value is None else str(value)).strip().lower()`
so only `None` (the genuine 'no value' case) becomes blank. Used
`{raw!r}` in the error so the diagnostic is unambiguous (`'0'` vs `0`).
Tests:
- known-value happy path (`'EMOJI'` → `'emoji'`)
- falsy non-string inputs (`0` / `False` / `[]`) surface meaningfully
- `None` keeps the blank-repr error
* feat(tui): expand light-terminal auto-detection (HERMES_TUI_THEME, BG hex)
Modern terminals (Ghostty, Warp, iTerm2) don't set COLORFGBG, so the
auto-light path was effectively COLORFGBG-only and silently broken for
many users. Two pragmatic additions, both opt-in, plus a clearer
priority chain:
1. **`HERMES_TUI_THEME=light|dark`** as a symmetric explicit override.
The existing `HERMES_TUI_LIGHT` is fine but reads as boolean noise;
a named theme env var matches `display.skin` muscle memory.
2. **`HERMES_TUI_BACKGROUND` hex/rgb hint.** Lets advanced users
(or a future OSC11 query helper that caches the answer) state a
ground-truth background colour. Decoded to Rec. 709 luma; ≥ 0.6
counts as light.
Priority order is now fully ordered and explainable:
1. `HERMES_TUI_LIGHT` (1/0/true/false/on/off).
2. `HERMES_TUI_THEME=light|dark`.
3. `HERMES_TUI_BACKGROUND` luminance.
4. `COLORFGBG` last field — light slots 7/15 → light, 0–15 → dark
(authoritative when set, so the new TERM_PROGRAM path can never
stomp on a terminal that already volunteered a dark answer).
5. `TERM_PROGRAM` allow-list — empty by default. The slot is left
in place because folks asked for it but populating it risks
wrongly flipping users on Apple_Terminal / iTerm2 dark profiles
to light. Easy to add per terminal once we have signal.
Tests: 5 new cases in `theme.test.ts` covering theme env, background
hex (3- and 6-char), invalid hex falling through, and COLORFGBG taking
precedence over the future allow-list.
Validation: `npm run type-check` clean, `npm test --run` 392/392.
* review(copilot): tighten theme detection comments + drop unnecessary cast
* review(copilot): strict hex regex so partial garbage doesn't slip into luminance
* test(tui): make TERM_PROGRAM allow-list injectable so precedence is provable
Copilot review on PR #17113: `LIGHT_DEFAULT_TERM_PROGRAMS` is empty
in production, so the prior assertion would have passed even if
`detectLightMode` ignored `COLORFGBG` entirely. That defeats the
test's purpose.
`detectLightMode` now takes the allow-list as an optional second
argument (defaults to the production set). The test injects a set
containing `Apple_Terminal`, asserts the allow-list alone WOULD
return light, then asserts `COLORFGBG: '15;0'` overrides it — the
precedence rule is now exercised, not assumed.
* fix(tui): COLORFGBG empty-trailing-field falls through; isolate DEFAULT_THEME tests
Round 2 Copilot review on PR #17113:
1. `Number(colorfgbg.split(';').at(-1))` returns 0 for an empty trailing
field (e.g. `COLORFGBG='15;'` → bg===0), which would have looked
like an authoritative dark slot and incorrectly blocked the
TERM_PROGRAM allow-list. Added a `/^\d+$/` guard before coercion;
non-numeric trailing fields now fall through.
2. Fixed the misleading '0–6 / 8–15 ranges are dark' comment — the
block returns true for bg===15, so the range is actually 0–6 / 8–14.
3. `DEFAULT_THEME` is computed from `process.env` at module-load.
A developer shell with `HERMES_TUI_THEME=light` (or a bright
`HERMES_TUI_BACKGROUND`) would flip it and break local tests.
The DEFAULT_THEME describe blocks now sterilize the relevant env
vars + dynamically import theme.ts (vi.resetModules pattern from
platform.test.ts). fromSkin tests compare against DARK_THEME
directly to decouple them from ambient env.
* test(tui): isolate ALL env-coupled theme symbols, not just DEFAULT_THEME
Round 3 Copilot review on PR #17113: the static top-level imports of
`fromSkin`, `DARK_THEME`, `LIGHT_THEME` evaluated theme.ts before
`importThemeWithCleanEnv` had a chance to clean the env. Because
`fromSkin` closes over `DEFAULT_THEME`, an ambient `HERMES_TUI_THEME=light`
or bright `HERMES_TUI_BACKGROUND` would still flip the base palette
and cause local-only failures.
Removed the static import entirely. Every test now obtains its theme
symbols via `importThemeWithCleanEnv`, including `detectLightMode`
(for consistency, even though it takes env as a parameter).
`fromSkin` tests assert against the cleaned `DEFAULT_THEME` from the
same dynamic import — preserves the actual contract (skins extend the
ambient base palette) without coupling the test to dev-shell state.
Verified by running with HERMES_TUI_THEME=light + HERMES_TUI_BACKGROUND=#ffffff:
all 20 theme tests still pass.
Self-review (avoid round 4):
- Audited other test files importing DEFAULT_THEME (syntax.test.ts,
streamingMarkdown.test.ts, constants.test.ts) — all just pass it as
a parameter or assert palette property existence (works on both
light + dark), so no env coupling there.
* fix(tui-gateway): harden stdio transport against half-closed pipes + SIGTERM races
`tui_gateway` reports `tui_gateway_crash.log` traces where the main
thread sits in `sys.stdin` while a worker holds `_stdout_lock` mid-
flush, and SIGTERM then calls `sys.exit(0)` while the lock is still
held — the interpreter shutdown stalls behind the wedged write.
Two narrowly scoped hardenings:
**`tui_gateway/transport.py`**
* Move JSON serialisation outside the lock — long messages no longer
block sibling writers while we serialise.
* Treat `BrokenPipeError`, `ValueError` ("I/O on closed file") and
generic `OSError` from both `write` and `flush` as "peer is gone":
return `False` instead of bubbling, matching what `write_json`'s
callers in `entry.py` already expect.
* Split `flush` into its own try block so a stuck flush never strands
a partial write or holds the lock indefinitely on its way out.
* Optional `HERMES_TUI_GATEWAY_NO_FLUSH=1` env knob to skip explicit
`flush()` entirely on environments where a half-closed read pipe
produces an indefinite kernel-level block. Default unchanged.
**`tui_gateway/entry.py`**
* `_log_signal` now spawns a 1-second daemon timer that calls
`os._exit(0)` if the orderly `sys.exit(0)` path is itself stuck
behind a wedged worker. Atexit handlers run inside the grace
window when they can; the timer is the safety net so a deadlocked
flush no longer strands the gateway process.
Tests:
* `test_write_json_closed_stream_returns_false` — ValueError path.
* `test_write_json_oserror_on_flush_returns_false` — OSError on flush
must not strand the lock; the write portion still landed before the
flush failure.
* `test_write_json_no_flush_env_skips_flush` — env knob bypass.
Validation: `scripts/run_tests.sh tests/tui_gateway/test_protocol.py`
(42/42 pass; one pre-existing failure on
`test_session_resume_returns_hydrated_messages` is unrelated to this
change — same `include_ancestors` mock kwarg issue tracked elsewhere).
`scripts/run_tests.sh tests/test_tui_gateway_server.py` 90/90 pass.
* review(copilot): tighten transport hardening comments + test cleanup
* review(copilot): narrow exception capture, configurable grace, simpler no-flush test
* fix(tui-gateway): narrow ValueError to closed-stream; surface UnicodeEncodeError
Copilot review on PR #17118: `UnicodeEncodeError` is a ValueError
subclass, so a non-UTF-8 stdout (mismatched PYTHONIOENCODING / locale)
would have been silently swallowed as 'peer gone' under
`except ValueError`. That hides a real environment bug.
Now:
- UnicodeEncodeError → log with exc_info (warning) and drop the frame
- ValueError where str(e) contains 'closed file' → peer gone, return False
- Any other ValueError → log loudly, drop frame (defensive, but visible)
Same shape applied to flush. Adds two regression tests.
* fix(tui-gateway): reserve write() False for peer-gone; re-raise programming errors
Round 2 Copilot review on PR #17118: `Transport.write()` returning
`False` is documented as 'peer is gone', and `entry.py` reacts by
calling `sys.exit(0)`. But the implementation also returned False
for non-IO conditions (non-JSON-safe payloads, UnicodeEncodeError,
unrelated ValueErrors), so a programming error or local env bug would
present as a clean disconnect — exactly the diagnosis pain we wanted
to eliminate.
Now:
- `json.dumps` failure → re-raises (TypeError/ValueError surfaces in crash log)
- `BrokenPipeError` → False (peer gone)
- `ValueError('...closed file...')` → False (peer gone)
- `UnicodeEncodeError` and any other ValueError → re-raise
- `OSError` → False (existing IO-failure semantics, debug-logged)
Tests updated to assert the re-raise behaviour and added a
non-serializable-payload regression test.
* fix(tui-gateway): narrow OSError to peer-gone errnos; honest test naming
Round 3 Copilot review on PR #17118:
- Docstring claimed False = peer gone, but generic OSError on write/flush
also returned False — meaning ENOSPC/EACCES/EIO would silently exit.
Added `_PEER_GONE_ERRNOS = {EPIPE, ECONNRESET, EBADF, ESHUTDOWN, +WSA}`
and narrowed the OSError handlers; non-peer-gone errnos re-raise.
Docstring now lists OSError as peer-gone branch with the errno set.
- The `_DISABLE_FLUSH` test was named after the env var but actually
patched the module constant. Renamed it to reflect the contract being
tested (skips flush when constant is true) AND added a real
end-to-end test that sets the env var, reloads transport.py, and
asserts the constant flips. Cleanup reload restores defaults so
parallel tests stay isolated.
Self-review (avoid round 4):
- Verified TeeTransport's secondary-swallow stays intentional.
- _log_signal grace path already covered by separate tests.
* fix(tui): honor display.busy_input_mode in TUI v2
The TUI v2 frontend hard-coded `composerActions.enqueue(full)` whenever
`ui.busy` was true. The classic CLI and gateway adapters honor the
`display.busy_input_mode` config key (`interrupt` | `queue` | `steer`),
but Ink ignored it — sending a message during a long-running turn always
landed in the queue regardless of config. The config default is already
`interrupt` (hermes_cli/config.py), so users who explicitly opted into
that experience were silently stuck on the legacy queue path.
This wires the value through the existing config-sync surface:
* `applyDisplay` now reads `display.busy_input_mode`, defaults to
`interrupt` (matching `_load_busy_input_mode` in tui_gateway), and
drops it into a new `UiState.busyInputMode` field.
* `dispatchSubmission` and the queue-edit fall-through call a shared
`handleBusyInput` helper that branches on the mode:
* `queue` — legacy behavior, append to the queue.
* `steer` — call `session.steer`; on rejection, fall back to
queue with a sys note.
* `interrupt` — `turnController.interruptTurn(...)` then `send()`,
so the new prompt actually moves.
* Mtime polling in `useConfigSync` already re-applies `config.full`, so
flipping `display.busy_input_mode` in `~/.hermes/config.yaml` takes
effect on the next 5s tick without restarting the TUI.
Tests:
* `applyDisplay → busy_input_mode` covers normalization + UiState fan-out.
* `normalizeBusyInputMode` mirrors the Python side's allow-list.
Validation:
* `npm run type-check` (in `ui-tui/`) — clean.
* `npm test --run` (in `ui-tui/`) — 394/394.
* review(copilot): narrow busy_input_mode type, preserve queue order on steer fallback
* review(copilot): clarify handleBusyInput comment (option, not return value)
* fix(tui): default busy_input_mode to queue in TUI (CLI keeps interrupt)
In a full-screen TUI users typically author the next prompt while the
agent is still streaming, so an unintended interrupt loses in-flight
typing. TUI fallback now defaults to `queue`; CLI / messaging
adapters keep `interrupt` as the framework default.
Override per-config via `display.busy_input_mode: interrupt` (or
`steer`) — the normalize/wire path is unchanged, only the missing-
value branch differs from the Python default.
uiStore initial value also flipped to `queue` so first-frame render
before `config.full` lands matches the eventual normalized value.
`turnController.recordMessageComplete` and `recordMessageDelta` both
prioritised `payload.rendered` over `payload.text`. `payload.rendered`
is the Rich-Console output `tui_gateway` builds for terminals that
can't render markdown themselves; the TUI already renders markdown via
`<Md>`. Two real bugs follow:
1. **Final answer garbled when `display.final_response_markdown: render`
is set** (#16391). Raw ANSI escape sequences pass through into the
React tree and the user sees overlapping coloured text instead of
their answer.
2. **Streaming silently drops content.** Per-delta `rendered` is an
*incremental* Rich fragment. The previous code did
`this.bufRef = rendered ?? this.bufRef + text`, which on every tick
replaced the whole accumulated buffer with the latest mid-sequence
ANSI fragment. Long replies arrived truncated and looked
half-painted — easy to miss as "model is being terse" instead of a
client bug.
Fix:
* `recordMessageComplete` now prefers `payload.text`, falling back to
`payload.rendered` only when the gateway elected not to send any.
* `recordMessageDelta` always accumulates `text`; `rendered` is ignored
on the streaming path entirely (Ink does its own markdown render via
`<Md>` / `streamingMarkdown.tsx`).
Tests:
* `prefers raw text over Rich-rendered ANSI on message.complete` —
the assistant message reflects raw markdown, not ANSI.
* `falls back to payload.rendered when text is missing` — preserves
the legacy "no `text`, only ANSI" path used by some adapters.
* `always accumulates raw text in message.delta and ignores rendered` —
pre-fix code would have made this assertion fail because each delta
overwrote the buffer.
Validation: `npm run type-check` clean, `npm test --run` 392/392 pass.
* fix(tui): make /browser connect actually take effect on the live agent
Reports were that `/browser connect <url>` (and "changes to CDP url
don't get picked up") didn't propagate to the live agent in `--tui`,
forcing users to fall back to setting `browser.cdp_url` in
`config.yaml` and restarting. Tracing the path on current main shows
the protocol wiring is already correct — `/browser` is registered in
`ui-tui/src/app/slash/commands/ops.ts` and dispatches `browser.manage`
through the gateway RPC, NOT the slash worker (covered by the
`browser.manage` row in `slashParity.test.ts`). But three real gaps
left the experience flaky:
1. `cleanup_all_browsers()` ran AFTER `os.environ["BROWSER_CDP_URL"]`
was rewritten. `_ensure_cdp_supervisor(...)` reads the env to
resolve its target URL, so a tool call landing in that brief window
could re-attach the supervisor to the OLD CDP endpoint just before
we reaped sessions, leaving the agent talking to a dead URL.
Reorder to clean first, swap env, clean again so the supervisor
for the default task is definitively closed.
2. `browser.manage status` reported only the env var, ignoring
`browser.cdp_url` from config.yaml. `_get_cdp_override()` (the
resolver the agent itself uses) consults both — match it so
`/browser status` answers the same question the next
`browser_navigate` will see. Closes a stealth bug where users
saw "browser not connected" while their CDP URL was perfectly
set in config.yaml.
3. `/browser disconnect` only cleared `BROWSER_CDP_URL` and reaped
once, leaving the same swap window as connect. Symmetrical
double-cleanup here too.
Frontend (`ops.ts`):
* Echo "next browser tool call will use this CDP endpoint" on success
so users see immediate confirmation that the gateway accepted the
swap, even before any tool runs.
* Mention `browser.cdp_url` in `config.yaml` in the usage hint and
the not-connected status line. Persistent config is the correct
fix for some terminal-multiplexer / sub-agent flows where env
inheritance is unreliable; surfacing it makes that workaround
discoverable.
Tests (4 new, all hermetic):
* `status` returns the resolved URL when only `browser.cdp_url` is
set in config.yaml.
* `connect` writes env AND cleans before/after, in that order.
* `connect` against an unreachable endpoint does NOT mutate env or
reap.
* `disconnect` removes env and cleans twice.
Validation:
scripts/run_tests.sh tests/test_tui_gateway_server.py — 94/94 pass.
cd ui-tui && npm run type-check — clean; npm test --run — 389/389.
* review(copilot): always defer to _get_cdp_override; normalize bare host:port
* review(copilot): collapse discovery-style CDP paths so /json/version isn't duplicated
* fix(tui): /browser status must not perform CDP discovery I/O
Copilot review on PR #17120: previous version routed through
`tools.browser_tool._get_cdp_override`, which calls
`_resolve_cdp_override` and performs an HTTP probe to /json/version
with a multi-second timeout for discovery-style URLs. That blocks
the TUI on `/browser status` whenever the configured host is slow
or unreachable.
Status now reads env-then-config directly with no network I/O. The
WS normalization still happens in `browser_navigate` for actual
tool calls, so behaviour-on-call is unchanged.
* fix(tui): skip /json/version probe for concrete ws://devtools/browser endpoints
Round 2 Copilot review on PR #17120: hosted CDP providers (Browserbase,
browserless, etc.) return concrete `ws[s]://.../devtools/browser/<id>`
URLs which are already directly connectable but don't serve the HTTP
discovery path. The previous `/json/version` probe rejected these
valid endpoints with 'could not reach browser CDP'.
For `ws[s]://...` URLs whose path starts with `/devtools/browser/` we
now do a TCP-level reachability check (`socket.create_connection`)
instead of the HTTP probe. The actual CDP handshake happens on the
next `browser_navigate` call, so we still surface unreachable hosts
as 5031 errors — just without the false negatives.
Discovery-style URLs (`http://host:port[/json[/version]]`) keep the
HTTP probe path unchanged. Updated existing test + added two new
ones (TCP-only success, TCP unreachable → 5031).
* feat(tui): opt-in auto-resume of the most recent session
`hermes --tui` always forges a fresh session at startup unless the user
sets `HERMES_TUI_RESUME=<id>`. Disconnects, terminal-window crashes,
and accidental Ctrl+D therefore lose every piece of in-flight context
even though `state.db` still has the full history a `/resume` away.
Add an opt-in path that mirrors classic CLI's `hermes -c` muscle
memory: when `display.tui_auto_resume_recent: true` is set in
`~/.hermes/config.yaml`, the TUI looks up the most recent human-facing
session and resumes it instead of starting fresh. Default off so
existing users aren't surprised; explicit `HERMES_TUI_RESUME` always
wins.
Wires:
* New `session.most_recent` JSON-RPC in `tui_gateway/server.py` that
returns the first non-`tool` row from `list_sessions_rich`, or
`{"session_id": null}` when none. Uses the same deny-list as
`session.list` so sub-agent rows can't sneak in.
* `createGatewayEventHandler.handleReady` re-ordered: explicit
`STARTUP_RESUME_ID` first (unchanged), then conditional auto-resume
via `config.get full → display.tui_auto_resume_recent`, then the
legacy `newSession()` fallback. Failures of either RPC fall back
to `newSession()` so the path is always finite.
* Default `display.tui_auto_resume_recent: False` added to
`DEFAULT_CONFIG` in `hermes_cli/config.py` (no `_config_version`
bump per AGENTS.md — deep-merge handles the additive key).
Tests:
* 4 new vitest cases in `createGatewayEventHandler.test.ts` cover
every gate-and-fallback combination (env wins, config off, config
on with hit, config on with miss).
* 3 new pytest cases for `session.most_recent` (denied row skip,
tool-only → null, db-unavailable → null).
Validation:
scripts/run_tests.sh tests/test_tui_gateway_server.py — 93/93.
cd ui-tui && npm run type-check — clean; npm test --run — 393/393.
* review(copilot): fold session.most_recent errors into null + extend ConfigDisplayConfig
* review(copilot): cover RPC-rejection fallbacks in auto-resume tests
* fix(tui): drop stale stream events after ctrl-c interrupt
Once interruptTurn() flips this.interrupted, only recordMessageDelta
short-circuited. recordReasoningDelta/Available, recordToolStart/
Progress/Complete, and recordInlineDiffToolComplete kept populating
turnState until the python loop reached its next _interrupt_requested
check (~1s on busy turns), making it look like ctrl-c was ignored
while late "thinking" + tool calls kept landing in the UI.
Add the same interrupted guard to every stream-side recorder, and
clear the flag at startMessage() so the next turn isn't suppressed
if the previous turn never delivered message.complete.
* fix(tui): guard recordTodos against post-interrupt mutation; fake-timers in test
Copilot review on PR #16706:
1. `recordToolStart` is interruption-guarded, but `tool.start`
handler also calls `recordTodos(payload.todos)` first — so a
late tool.start carrying todos could still mutate `turnState.todos`
after Ctrl-C, leaving ghost rows in the panel. Adds the same
`if (this.interrupted) return` early-exit to `recordTodos` so
*all* tool.start side-effects are dropped post-interrupt.
2. The interrupt test was leaking a real `setTimeout` (interrupt
cooldown) across test files, which could fire later and mutate
uiStore from the wrong test context. Wraps the test in
`vi.useFakeTimers()` + `vi.runAllTimers()` and restores real
timers in finally.
3. Extends the same test with a todos payload on the post-interrupt
tool.start so we have explicit regression coverage for #1.
* fix(tui): guard pushTrail post-interrupt; harden interrupt-test cleanup
Round 2 Copilot review on PR #16706:
1. `tool.generating` events route through `pushTrail`, which was not
interruption-guarded — late events could still write 'drafting …'
into `turnTrail` after Ctrl-C, leaving a stale shimmer in the UI.
Adds the same `if (this.interrupted) return` early-exit.
2. Test cleanup moved `vi.runAllTimers()` into `finally` (before
`vi.useRealTimers()`) so a mid-test assertion failure can't leak
the interrupt-cooldown setTimeout across other test files.
3. Replaced the misleading 'pre-interrupt todos … expected to be
cleared by the interrupt cycle' comment with an accurate one
reflecting current behaviour (interrupt does NOT clear todos).
4. Added an explicit assertion that a post-interrupt `tool.generating`
event does not extend `turnTrail` — regression coverage for #1.
* fix(tui): append gateway stderr tail to start_timeout activity
`gateway.start_timeout` previously published only `cwd` + `python`,
which made TUI startup failures hard to disambiguate. The user saw
`gateway startup timed out · /path/to/python /repo · /logs to inspect`
with no signal whether the actual cause was a wrong python interpreter,
a missing dependency, or a config parse failure.
Plumb a 20-line stderr tail through the event so the most useful lines
land directly in the TUI activity feed, capped to the last 8 non-empty
lines for readability:
* `gatewayClient.ts` — collect `getLogTail(20)` when the readyTimer
fires and attach it as `payload.stderr_tail`.
* `gatewayTypes.ts` — extend the `gateway.start_timeout` event union
with the new optional field.
* `createGatewayEventHandler.ts` — emit the trimmed lines after the
existing `gateway startup timed out` activity entry, classified
`error`.
Tests: regression test in `createGatewayEventHandler.test.ts` checks
that `ModuleNotFoundError` / `FileNotFoundError` lines from the tail
land in `getTurnState().activity` so they show up in the UI immediately.
Validation: `npm run type-check` clean, `npm test --run` 390/390.
* review(copilot): filter blanks before slice and cap stderr tail at 120 chars
Tables rendered through `<Md>` had no separator and no header weight,
so they read as a paragraph with extra whitespace. This adds two tiny,
border-free changes that survive Ink's grapheme-approximate column
widths better than a full outline:
* Bold the header row, keeping the existing amber colour.
* Insert a dim `─`-dashed rule between the header and body rows.
We deliberately stay away from a full outline — column widths are
measured via `stripInlineMarkup(...).length`, which is grapheme-aware
but still off by a cell on East Asian wide characters and emoji-mid-
cell strings. A header rule plus the existing 2-space column gap
gives the visual hierarchy the issue asks for without amplifying that
inaccuracy into a misaligned border.
Validation: `npm run type-check` clean, `npm test --run` 389/389.
- Remove dead _lmstudio_loaded_context attribute from run_agent.py (set
but never read — the loaded context is pushed to context_compressor.update_model
which is the actual consumer)
- Cache empty reasoning options with 60s TTL to avoid per-turn HTTP probe
for non-reasoning LM Studio models. Non-empty results cached permanently.
- Extract _lmstudio_server_root(), _lmstudio_request_headers(), and
_lmstudio_fetch_raw_models() shared helpers in models.py — eliminates
URL-strip + auth-header + HTTP-call duplication across probe_lmstudio_models,
ensure_lmstudio_model_loaded, and lmstudio_model_reasoning_options
- Revert runtime_provider.py base_url precedence change: preserve the
established contract (saved config.base_url > env var > default) for all
api_key providers
- Remove unnecessary config version bump 22→23
- Fix TUI test: relax target_model assertion to avoid module-cache flake
- AUTHOR_MAP: added rugved@lmstudio.ai → rugvedS07
CopilotACPClient communicates via subprocess stdio and returns a plain
SimpleNamespace from _create_chat_completion(). The streaming path tries
to iterate this as a stream, crashing with:
TypeError: 'types.SimpleNamespace' object is not iterable
Mirror the existing ACP exclusion pattern (used for Responses API upgrade)
to disable streaming when provider is copilot-acp or base_url starts with
acp:// or acp+tcp://.
Based on PR #9428 by @ningfangbin and issue #16271 by @Joseph19820124.
Fixes#16271
* ci(nix): auto-fix stale npm hashes on push to main
When a PR merges to main with updated package-lock.json or package.json
in ui-tui/ or web/, the new auto-fix-main job detects stale npmDepsHash
values and pushes a fix commit directly to main.
This eliminates the recurring manual hash-bump PRs (#15420, #15314,
#15272, #15244) by reusing the existing fix-lockfiles --apply pipeline.
The fix commit only touches nix/*.nix files, which are outside the push
path filter (package-lock.json / package.json), so it cannot re-trigger
itself.
Closes#15314
* fix(ci): use GitHub App token for auto-fix-main push
GITHUB_TOKEN commits are invisible to workflow triggers (GitHub's
infinite-loop prevention). The auto-fix-main job pushes directly to
main, so the fix commit never triggered downstream nix.yml verification.
Mint a short-lived token via the repo's GitHub App (daimon-nous, APP_ID
+ APP_PRIVATE_KEY secrets) so the push is treated as a real event and
nix.yml fires to verify the corrected hashes.
Tested via workflow_dispatch dry-run: app token minted successfully,
checkout with app token succeeded, fix job correctly gated.
Resolves review feedback from Bugbot (r3144569551).
* ci(nix): rename lockfile check job for required status check
Rename 'check' → 'nix-lockfile-check' so the status check name is
unambiguous when added as a required check on main.
* fix(ci): harden auto-fix-main against races, loops, and silent failures
Address adversarial review findings:
1. Race condition (#1): Job-level concurrency with cancel-in-progress
collapses back-to-back pushes; ref: main checkout always gets latest
branch state; explicit push target (origin HEAD:main).
2. Loop prevention (#2): File-whitelist check before commit aborts if
any file outside nix/{tui,web}.nix was modified, preventing
accidental self-triggering.
3. Silent infra failures (#8): nix-lockfile-check now fails explicitly
when fix-lockfiles exits without reporting stale status (catches nix
setup failures, network errors, script bugs that bypass continue-on-error).
4. Commit traceability (#11): Auto-fix commits include source SHA and
workflow run URL in the commit body.
5. Explicit push target (#12): git push origin HEAD:main instead of
bare git push.
---------
Co-authored-by: alt-glitch <alt-glitch@users.noreply.github.com>
* fix(nix): make extraPackages actually work — wire into per-user profile
#17030 deprecated extraPackages because it only set the systemd service
PATH, which the terminal backend's login-shell snapshot discards.
Instead of deprecating, fix it: set users.users.${cfg.user}.packages
so NixOS builds a per-user profile at /etc/profiles/per-user/hermes/bin.
This path is included in PATH by /etc/set-environment, which the login
shell sources, so the terminal backend's snapshot picks it up.
One line of actual logic:
users.users.${cfg.user}.packages = cfg.extraPackages;
Verified in a NixOS VM test: su - hermes -c 'which hello' resolves
to /etc/profiles/per-user/hermes/bin/hello.
Reverts the deprecation warning and docs changes from #17030, restores
extraPackages as the recommended way to give the agent extra tools.
Container mode is unaffected — extraPackages was always native-only
(the systemd path line is inside !cfg.container.enable).
* nix: clarify additive merge semantics for extraPackages user profile
---------
Co-authored-by: Siddharth Balyan <daimon@noreply.github.com>
BOOT.md was merged in PR #3733 before the feature was ready — the
built-in hook spawned a bare AIAgent() with no model/runtime kwargs,
which immediately 401s on any provider with a custom endpoint. Three
separate community PRs (#5240, #12514, #14992) tried to paper over it.
Remove the BOOT.md hook entirely and its user-facing docs/tips. Keep
the gateway/builtin_hooks/ package and the HookRegistry._register_builtin_hooks()
hook-point intact as the extension surface for future always-on
gateway hooks.
Closes#5239.
Co-authored-by: teknium1 <teknium@users.noreply.github.com>
* perf(startup): lazy-import OpenAI, Anthropic, Firecrawl, account_usage
Four heavy SDK/module imports are now deferred off the hot startup path.
Net savings on cold module imports:
cli 1200 → 958 ms (-242)
run_agent 1220 → 901 ms (-319)
tools.web_tools 711 → 423 ms (-288)
agent.anthropic_adapter 230 → 15 ms (-215)
agent.auxiliary_client 253 → 68 ms (-185)
Four independent changes in one PR since they all use the same pattern
and share the same risk profile (heavy SDK import → lazy proxy or
function-local import):
1. tools/web_tools.py:
'from firecrawl import Firecrawl' moved into _get_firecrawl_client(),
which is only called when backend='firecrawl'. Users on Exa/Tavily/
Parallel pay zero firecrawl cost.
2. cli.py + gateway/run.py:
'from agent.account_usage import ...' moved into the /limits handlers.
account_usage transitively pulls the OpenAI SDK chain; only needed
when the user runs /limits.
3. agent/anthropic_adapter.py:
'try: import anthropic as _anthropic_sdk' replaced with a cached
'_get_anthropic_sdk()' accessor. The three usage sites
(build_anthropic_client, build_anthropic_bedrock_client,
read_claude_code_credentials_from_keychain) now resolve via the
accessor. All pre-existing test patches of
'agent.anthropic_adapter._anthropic_sdk' keep working because the
accessor respects any value already in module globals.
4. agent/auxiliary_client.py AND run_agent.py:
'from openai import OpenAI' replaced with an '_OpenAIProxy()' module-
level object that looks like the OpenAI class but imports the SDK on
first call/isinstance check. This preserves:
- 15+ in-module OpenAI(...) construction sites in auxiliary_client
and the single site in run_agent's _create_openai_client (Python's
function-scope name lookup finds the proxy, forwards the call);
- 'patch("agent.auxiliary_client.OpenAI", ...)' and
'patch("run_agent.OpenAI", ...)' test patterns used by 28+ test
files (patch replaces the module attribute as usual).
Tried two alternatives first:
- 'from openai._client import OpenAI' — doesn't skip openai/__init__.py
(the audit's hypothesis here was wrong).
- Module-level __getattr__ — works for external access but Python
function-scope name resolution skips __getattr__, so in-module
OpenAI(...) calls NameError.
Note: 'openai' still loads on 'import cli' because
cli.py -> neuter_async_httpx_del() -> openai._base_client, and
run_agent.py -> code_execution_tool.py (module-level
build_execute_code_schema) -> _load_config() -> 'from cli import
CLI_CONFIG'. Deferring those is a separate, larger change — out of scope
for this PR. The savings above all come from avoiding the openai/*,
anthropic/*, and firecrawl/* top-level type-tree imports on paths that
don't need them.
Verified:
- 302/302 tests in tests/agent/{test_anthropic_adapter,
test_bedrock_1m_context, test_minimax_provider, test_anthropic_keychain}
pass. Two pre-existing failures on main unchanged.
- 106/106 tests/agent/test_auxiliary_client.py pass (1 pre-existing fail).
- 97/97 tests/run_agent/test_create_openai_client_kwargs_isolation.py,
test_plugin_context_engine_init.py, test_invalid_context_length_warning.py,
test_api_max_retries_config.py,
tests/hermes_cli/test_gemini_provider.py, test_ollama_cloud_provider.py
pass (1 pre-existing fail).
- Live hermes chat smoke: 2 turns + /model switch + tool calls, zero
errors in the 57-line agent.log window.
- Module-level import of run_agent + auxiliary_client + anthropic_adapter
no longer pulls 'anthropic' or 'firecrawl' at all.
* fix(gateway): restore top-level account_usage import for test-patch surface
CI caught two failures in tests/gateway/test_usage_command.py that I
missed locally:
AttributeError: 'module' object at gateway.run has no attribute 'fetch_account_usage'
The test uses monkeypatch.setattr('gateway.run.fetch_account_usage', ...)
to inject a fake account-fetch call. Moving the import inside the
handler deleted that module-level attribute, breaking the patch surface.
Restoring the top-level import in gateway/run.py gives up the ~230 ms
gateway-boot savings from that one lazy, but:
1. the gateway is a long-running daemon — boot cost is paid once per
install, not per turn;
2. the other four lazy-imports (firecrawl, openai, anthropic, cli's
account_usage) remain in place and still account for the bulk of
the savings reported in the PR body;
3. preserving the patch surface keeps the established
'gateway.run.fetch_account_usage' monkeypatch pattern working
without touching tests.
Verified: tests/gateway/test_usage_command.py — 8 passed, 0 failed.
Full targeted sweep (2336 tests across agent/gateway/hermes_cli/run_agent):
2332 passed, 4 failed — all 4 pre-existing on main.
---------
Co-authored-by: teknium1 <teknium@users.noreply.github.com>
load_config() and read_raw_config() now cache their result keyed on
the config file's (mtime_ns, size). On cache hit they return a deepcopy
of the cached value, skipping yaml.safe_load + deep-merge + normalize +
env-var expansion entirely. save_config() + migrate_config() write via
atomic_yaml_write which produces a fresh inode, so stat() sees a new
mtime_ns and the next load repopulates automatically — no explicit
invalidation hook needed.
Measured per-call cost:
load_config() cold: 13.3 ms
load_config() cached: 0.23 ms (57x faster)
read_raw_config() cached: 0.13 ms
A single gateway turn hits the config 5-15 times (session context,
auxiliary client resolution, memory config, plugin hooks, approval
lookups, per-tool settings). That's 65-200 ms/turn of pure YAML
re-parsing on main. After this change: 1-3 ms/turn.
Also migrates gateway/run.py's 6 direct yaml.safe_load(config.yaml)
call sites through _load_gateway_config, which now shares the
read_raw_config cache when _hermes_home agrees with the canonical
config path. The direct-read fallback is retained for tests that
monkeypatch gateway_run._hermes_home without touching HERMES_HOME.
Safety:
- load_config() returns a deepcopy on every call; the 67+ call sites
that mutate the result (cfg["model"]["default"] = ..., etc.) can't
corrupt the cache.
- save_config() / atomic_yaml_write bump mtime, naturally invalidating
the cache for the next reader.
- Cache is keyed on str(config_path), so HERMES_HOME profile switches
don't collide.
Verified:
- 112 config tests pass (test_config, test_config_env_expansion,
test_config_env_refs, test_config_drift, test_config_validation,
test_aux_config).
- 87 gateway tests pass (test_verbose_command, test_session_info,
test_compress_focus, test_runtime_footer, test_resume_command,
test_reasoning_command, test_approve_deny_commands,
test_run_progress_interrupt).
- Live hermes chat smoke — 2 turns + /model switch + tool calls,
zero errors in agent.log.
Co-authored-by: teknium1 <teknium@users.noreply.github.com>
Previously, check_browser_requirements() only checked for the agent-browser
CLI, not the Chromium binary it drives. When the CLI was present but
Chromium wasn't (common in Docker images predating the playwright install
step), the browser tool was advertised to the agent, every call hung for
the full command timeout (~30s each, ~220s for a chained navigate), and
the agent eventually gave up with no useful error — users saw 'browser
not working' with empty errors.log.
Changes:
- tools/browser_tool.py: add _chromium_installed() checking
PLAYWRIGHT_BROWSERS_PATH + default Playwright cache paths for
chromium-* / chromium_headless_shell-* dirs; wire into
check_browser_requirements() for local mode (cloud providers
unaffected). _run_browser_command fails fast with an actionable
Docker vs. host message instead of hanging. _running_in_docker()
checks /.dockerenv and /proc/1/cgroup.
- hermes_cli/tools_config.py: post_setup for 'Local Browser' now runs
'agent-browser install --with-deps' after npm install to actually
download Chromium. In Docker, points user at the updated image pull
instead of trying to install into a read-only layer. Cloud-provider
post_setup (browserbase) skips Chromium install entirely.
- tests/tools/test_browser_chromium_check.py: new tests covering
search roots, install detection, requirements branches (local/cloud/
camofox), and the fast-fail guard in docker/non-docker contexts.
- tests/tools/test_browser_homebrew_paths.py: 5 existing subprocess-path
tests now mock _chromium_installed=True since they exercise the
post-guard subprocess path.
Co-authored-by: teknium1 <teknium@users.noreply.github.com>
Discord's per-app command-management bucket is ~5 writes / 20 s. A
mass-prune-plus-upsert reconcile (77 orphans + 30 desired = 107 writes
in the reported case) can't finish under the old flat 30 s budget, and
the subsequent reconnect retries inside the rate-limit cooldown also
time out — leaving slash commands broken for ~60 min until the bucket
fully recovers.
Bump the timeout to 600 s so realistic bursts drain, update the warning
message to point at the saturated bucket instead of a hardcoded 30 s.
The 600 s cap still guards against a true hang.
Credit to @Tranquil-Flow for PR #16739 and @davidbordenwi for reporting
#16713 with the bucket-math diagnosis.
Closes#16713.
Co-authored-by: Teknium <teknium@nousresearch.com>
The telegram.reactions key was already wired up (gateway/config.py bridges
it to TELEGRAM_REACTIONS at startup) but was undocumented and missing from
DEFAULT_CONFIG, so users had no way to discover it. Add it with the
existing off-by-default behavior preserved.
No behavior change — runtime default stays False.
Co-authored-by: teknium1 <teknium@users.noreply.github.com>
extraPackages adds packages to the systemd service PATH, but the
terminal backend's login-shell snapshot rebuilds PATH from NixOS system
profiles, so tools added via extraPackages are invisible to terminal
commands, skills, and cron jobs — the entire use case.
Changes:
- Mark the option description as deprecated with explanation
- Emit a NixOS warning when extraPackages is non-empty, including a
ready-to-paste environment.systemPackages replacement
- Update docs: quick-reference table, plugin example, and options
reference all point to environment.systemPackages
The option still functions (non-breaking) so existing configs keep
working while users migrate.
Auxiliary tasks (title_generation, vision, compression, web_extract,
session_search) now pick the correct wire protocol based on the
endpoint, not just on which resolve_provider_client branch built the
client. Fixes 404s on Kimi Coding Plan and any other named provider
whose endpoint speaks Anthropic Messages.
Root cause: the 'api_key' branch of resolve_provider_client (and the
Step 2 fallback chain inside _resolve_auto) always built a plain
OpenAI client regardless of what the endpoint actually spoke. For
provider=kimi-coding + model=kimi-for-coding, that meant:
POST https://api.kimi.com/coding/v1/chat/completions
{ "model": "kimi-for-coding", ... }
→ 404 resource_not_found_error
The /coding route only accepts the Anthropic Messages shape (the main
agent already uses api_mode=anthropic_messages for it). Earlier fixes
(#16819, #22ddac4b1) patched the anonymous-custom, named-custom, and
external-process branches — but the named api_key branch (kimi-coding,
minimax, zai, future /anthropic providers) was the fourth sibling and
never got the same treatment.
Fix: one module-level helper _maybe_wrap_anthropic() that rewraps a
plain OpenAI client in AnthropicAuxiliaryClient when:
- api_mode is explicitly 'anthropic_messages', OR
- the URL ends in '/anthropic', OR
- the host is api.kimi.com + path contains '/coding', OR
- the host is api.anthropic.com.
Wired into _wrap_if_needed (covers all resolve_provider_client
branches that already go through it) and into the Step 2 api_key
fallback chain inside _resolve_auto. Explicit api_mode still wins:
passing api_mode='chat_completions' forces OpenAI wire, and already-
wrapped specialized adapters (Codex, Gemini native, CopilotACP) pass
through unchanged.
E2E verified:
- resolve_provider_client('kimi-coding', 'kimi-for-coding')
→ AnthropicAuxiliaryClient (was plain OpenAI, which 404'd)
- _resolve_auto Step 1 for kimi-coding runtime → AnthropicAuxiliaryClient
- resolve_provider_client('openrouter', ...) → plain OpenAI (no regression)
- api_mode='chat_completions' override → plain OpenAI (explicit wins)
Tests:
- tests/agent/test_auxiliary_transport_autodetect.py (new): 21 tests
covering URL detection, wrap decisions, and integration.
- 204/205 existing auxiliary tests pass (1 pre-existing failure on
main, unrelated to this change).
Co-authored-by: teknium1 <teknium@users.noreply.github.com>
Append a compact 'model · 68% · ~/projects/hermes' footer to the FINAL
message of each turn, disabled by default (display.runtime_footer.enabled).
Answers the Telegram-side parity ask: runtime context that the CLI status
bar already shows is now available in messaging replies when enabled.
Wiring:
- gateway/runtime_footer.py: resolve_footer_config + format_runtime_footer +
build_footer_line. Pure-function renderer; per-platform overrides under
display.platforms.<platform>.runtime_footer.
- gateway/run.py: appends footer to response right after reasoning prepend
so it lands only on the final message (never tool progress or streaming
chunks). When streaming already delivered the body (already_sent), the
footer is sent as a small trailing message instead.
- agent_result now exposes context_length alongside last_prompt_tokens so
the footer can compute the pct; both gateway return paths updated.
- /footer [on|off|status] slash command, wired in CLI (cli.py) and gateway
(gateway/run.py both running-agent bypass and main dispatch). Global
toggle only; per-platform overrides via config.yaml.
Graceful degradation:
- Missing context_length (unknown model) → pct field silently dropped
(no '?%' artifact).
- Empty final_response → no footer appended.
- Unknown field names in config → silently ignored.
Tests: 25-case unit suite (tests/gateway/test_runtime_footer.py) plus E2E
harness covering streaming vs non-streaming branches, per-platform override,
and the exact argument contract gateway/run.py uses.
Co-authored-by: teknium1 <teknium@users.noreply.github.com>
Mechanical cleanup across 43 files — removes 46 unused imports
(F401) and 14 unused local variables (F841) detected by
`ruff check --select F401,F841`. Net: -49 lines.
Also fixes a latent NameError in rl_cli.py where `get_hermes_home()`
was called at module line 32 before its import at line 65 — the
module never imported successfully on main. The ruff audit surfaced
this because it correctly saw the symbol as imported-but-unused
(the call happened before the import ran); the fix moves the import
to the top of the file alongside other stdlib imports.
One `# noqa: F401` kept in hermes_cli/status.py for `subprocess`:
tests monkeypatch `hermes_cli.status.subprocess` as a regression
guard that systemctl isn't called on Termux, so the name must
exist at module scope even though the module body doesn't reference
it. Docstring explains the reason.
Also fixes an invalid `# noqa:` directive in
gateway/platforms/discord.py:308 that lacked a rule code.
Co-authored-by: teknium1 <teknium@users.noreply.github.com>
The contributor's PR (#16750) scoped the fix to run_setup_wizard() and
explicitly punted the two sibling sites. Both have the identical
[ -e /dev/tty ] pattern followed by a < /dev/tty redirect and crash in
Docker the same way:
- scripts/install.sh:732 install_system_packages() -- apt sudo prompt
fallback. sudo ... < /dev/tty dies with the same ENXIO.
- scripts/install.sh:1395 maybe_start_gateway() -- gateway-install gate,
same function path as the wizard reproducer.
Fix both with the same (: </dev/tty) 2>/dev/null probe, and parametrize
the regression test over all three gated functions so any future
regression is caught regardless of which site breaks.
Address the three Copilot inline findings on the regression test:
- Switch _extract_run_setup_wizard() from str.index() with hard-coded
markers (which raises ValueError if `maybe_start_gateway()` is renamed
or the marker leaks into a comment) to an anchored regex on the
function-definition + closing-brace boundaries.
- Match `[ -e /dev/tty ]` with surrounding whitespace, optional quoting,
and the `test -e /dev/tty` form so the regression guard catches every
spelling of the existence-only check, not just the exact substring.
- Replace the literal `(: </dev/tty)` substring assertion with a
higher-level invariant — the gate must be an `if`/`if !` whose test
redirects stdin from /dev/tty — so equivalent open-based probes
(`exec 3</dev/tty` + close, brace-grouped variants, etc.) keep the
test green while the bare existence check stays caught.
Verified guard: both tests still pass on the fix and both fail on
`origin/main` with the documented messages.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
In Docker builds the `/dev/tty` device node is present in the mount
namespace, so `[ -e /dev/tty ]` returns true — but opening it fails
with `ENXIO: No such device or address`. Under the old gate the
"no terminal available" skip never triggered, the setup wizard ran,
and the build aborted a few lines later when bash tried `< /dev/tty`:
/tmp/install.sh: line 1347: /dev/tty: No such device or address
Replace the existence check with `(: </dev/tty) 2>/dev/null`, which
actually attempts to open /dev/tty in a subshell. The probe succeeds
when piped from `curl | bash` on a real terminal (the wizard's intended
use case) and fails cleanly in Docker build / CI contexts so the skip
kicks in before the redirect can crash.
Add a regression test that statically asserts run_setup_wizard does not
gate on the bare existence check and that the open-based probe is in
place.
Fixes#16746.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
delegate_task runs inside the parent turn and is cancelled when the parent is interrupted (new user message, /stop, /new). The child status payload (status=interrupted, exit_reason=interrupted) is already honest, but the tool schema and user-facing docs did not set the expectation, so users reasonably assumed delegated subagents would keep running in the background after interrupting the parent.
Updates:
- tools/delegate_tool.py DELEGATE_TASK_SCHEMA description adds a WHEN NOT TO USE bullet pointing at cronjob / terminal(background=True, notify_on_complete=True) for durable long-running work.
- website/docs/user-guide/features/delegation.md gains a Lifetime and Durability callout above Key Properties.
- website/docs/guides/delegation-patterns.md expands the Use something else list and the Constraints section with the same guidance.
Reported by LizLiz (@lizliz404) via Teknium.
Co-authored-by: teknium1 <teknium@users.noreply.github.com>
The gateway caches one AIAgent per session to preserve prompt-cache hits,
keyed by _agent_config_signature(). The signature previously only
fingerprinted model/credentials/toolsets/ephemeral-prompt — NOT the
compression or context_length config. As a result, users who edited
model.context_length or compression.threshold in config.yaml on a
long-lived gateway saw no effect until they triggered an unrelated
cache eviction (/model switch, /reset, gateway restart).
Add a new cache_keys parameter to _agent_config_signature and a
_CACHE_BUSTING_CONFIG_KEYS registry listing config values the agent
bakes in at construction time. Call sites read the current config and
pass it through — next gateway message with an edited config
rebuilds the agent.
Keys registered:
- model.context_length
- compression.enabled
- compression.threshold
- compression.target_ratio
- compression.protect_last_n
Reported by @OP (Apr 26 feedback bundle).
## Changes
- gateway/run.py: new _CACHE_BUSTING_CONFIG_KEYS tuple,
_extract_cache_busting_config classmethod, cache_keys kwarg on
_agent_config_signature, call site passes the extracted dict
- tests/gateway/test_agent_cache.py: 11 new tests
(5 on _agent_config_signature behavior, 6 on _extract_cache_busting_config)
Co-authored-by: teknium1 <teknium@users.noreply.github.com>
Network errors through proxies (e.g. sing-box) can leave httpx
connections in a half-closed state occupying pool slots. After enough
reconnect cycles the 256-connection default fills up entirely, causing
Pool timeout: All connections in the connection pool are occupied.
Fix: cycle only the getUpdates request object (_request[0]) via
shut-down + re-initialize before restarting polling. This drains stale
connections without touching the general request (_request[1]) that
concurrent send_message / edit_message calls rely on.
The drain is applied to both _handle_polling_network_error and
_handle_polling_conflict reconnect paths via a shared
_drain_polling_connections() helper. Failures in the drain are
swallowed so reconnect always proceeds.
Based on #16466 by @Mirac1eSky.
Auxiliary callers that configure reasoning via
auxiliary.<task>.extra_body.reasoning were having that config silently
dropped by the Codex Responses adapter — it only forwarded
messages/model/tools through to responses.stream(), never translating
chat.completions-shaped reasoning hints into the Responses API's
top-level reasoning + include fields.
Mirror the main-agent translation from agent/transports/codex.py:
- extra_body.reasoning.effort → resp_kwargs.reasoning.{effort, summary:"auto"}
- 'minimal' → 'low' clamp (Codex backend rejects 'minimal')
- Always include ['reasoning.encrypted_content'] when reasoning is enabled
- {'enabled': False} → omit reasoning and include entirely
- Non-dict reasoning values are ignored defensively
Reported by @OP (Apr 26 feedback bundle).
## Changes
- agent/auxiliary_client.py: _CodexCompletionsAdapter.create() now reads
and translates extra_body.reasoning before calling responses.stream()
- tests/agent/test_auxiliary_client.py: 9 new tests covering all effort
levels, the minimal→low clamp, the disabled path, the no-op paths,
and defensive handling of wrong-shape inputs
Co-authored-by: teknium1 <teknium@users.noreply.github.com>
The gateway session-hygiene pre-compression safety valve had a hardcoded
400-message threshold. On long-lived sessions with short turns this was
either too high (users with aggressive compression preferences) or too
low (users with very large context models who want to keep more history
in-flight).
Add compression.hygiene_hard_message_limit (default 400) so it can be
tuned without forking the gateway.
Reported by @OP (Apr 26 feedback bundle).
## Changes
- hermes_cli/config.py: new DEFAULT_CONFIG key with 400 default
- gateway/run.py: read compression.hygiene_hard_message_limit at
hygiene-time, fall back to 400 if missing/invalid
- tests/gateway/test_session_hygiene.py: two tests — override fires at
the configured limit, default does not fire below 400
Co-authored-by: teknium1 <teknium@users.noreply.github.com>
When openai-codex tokens expire or the ChatGPT account hits a 429
window, the pool entry gets marked STATUS_EXHAUSTED with
last_error_reset_at many hours in the future. If the user then runs
`hermes model` / `hermes auth openai-codex` to reauth, fresh tokens
land in ~/.hermes/auth.json but the pool entry stayed frozen behind
its reset_at — every request kept failing with 'credential pool: no
available entries (all exhausted or empty)' until the original window
elapsed.
_available_entries() already had auth.json/credentials-file resync
branches for anthropic/claude_code and nous/device_code; openai-codex
was missing. Added _sync_codex_entry_from_auth_store() mirroring the
nous version (reads state["tokens"][{access,refresh}_token] +
state["last_refresh"]) and wired it into the exhausted-entry resync
loop.
Also softens the 'codex CLI not found' doctor warning — native
device-code OAuth does not require the Codex binary, only
importing existing Codex CLI tokens does. Downgraded to an info line.
Reported on Discord by p1aceho1der: Codex stalled indefinitely after
a rate-limit reset, reauth didn't help, and doctor falsely warned
that the codex CLI was required.
Co-authored-by: teknium1 <teknium@users.noreply.github.com>
Gemini 3 Flash documents low/medium/high as the accepted thinkingLevel
values. The salvaged bridge was forwarding Hermes' "minimal" effort to
Flash verbatim, which is not a documented Gemini level and risks a 400
from the native adapter.
Clamp minimal->low on Flash (matching how Pro already clamps minimal+low
down), and funnel anything outside {low, medium, high} into medium to
keep the request valid by construction. No behaviour change for the
documented effort levels.
Telegram has no native table syntax. The gateway auto-rewrites pipe
tables into row-group bullets (see previous commit), but letting models
know up front means they emit the clean form directly instead of
relying on post-processing to synthesize headings.
Also helps users whose MEMORY.md formatting policies were being
overridden — the platform hint now carries the guidance.
The rate-limit branch added by the original PR did sleep+continue with
no attempt to record the last error, so persistent iLink -2 responses
exhausted the retry loop and hit 'assert last_error is not None',
raising AssertionError instead of a descriptive RuntimeError.
Record last_error = RuntimeError(...) before continuing, and break out
of the loop on the final attempt instead of sleeping uselessly.
- Change MAX_MESSAGE_LENGTH from 4000 to 2000 to match Weixin iLink API limit
- Add RATE_LIMIT_ERRCODE = -2 handling with 3x backoff retry
- Increase default send_chunk_delay_seconds from 0.35 to 1.5 to avoid rate limits
- Increase default send_chunk_retries from 2 to 4 for better reliability
- Use _split_text() in send() to chunk long messages before delivery
Fixes#16411
Add tests/test_cli_manual_compress.py verifying _manual_compress passes
None (not the cached system prompt) to _compress_context, forwards the
/compress <topic> focus string, rotates CLI session_id to the new child
session, and clears the pending title.
Co-authored-by: revar <revar@users.noreply.github.com>
_manual_compress() passed self.agent._cached_system_prompt to
_compress_context() as the system_message argument. _compress_context
calls _build_system_prompt(system_message), which appends system_message
to prompt_parts that already contain the agent identity block — causing
the identity to appear twice in the new session's system prompt
(20,957 -> 42,303 chars, +102% as reported in issue #15281).
Fix: pass None instead of _cached_system_prompt. _build_system_prompt(None)
rebuilds the system prompt correctly from scratch without appending a
pre-built prompt on top of the identity layers.
Fixes#15281
Follow-up to PR #16802 (BeliefanX). The original fix read
`agent_history[-1].get("timestamp")` for the tool-tail freshness gate,
but `gateway/run.py` strips the `timestamp` field off all tool/tool_call
rows when building `agent_history` from the raw transcript (see
`clean_msg = {k: v for k, v in msg.items() if k != "timestamp"}`). At
runtime the tool-tail branch always saw `None` and silently took the
legacy-fresh path — the stale-guard never fired for the tool-tail case
it was supposed to cover.
Changes:
- Read the freshness signal from the RAW `history` list (via new
`_last_transcript_timestamp()` helper) BEFORE the strip. Both the
resume_pending branch and the tool-tail branch use this single signal,
replacing the two divergent ones.
- Default window bumped 15 min → 1 hour via new
`_AUTO_CONTINUE_FRESHNESS_SECS_DEFAULT`. The 15-minute default was
shorter than the default `gateway_timeout` of 30 min, so a legitimate
long-running turn interrupted near its timeout boundary and resumed
shortly after would have been misclassified as stale.
- Configurable via `config.yaml` `agent.gateway_auto_continue_freshness`
(bridged to `HERMES_AUTO_CONTINUE_FRESHNESS` at gateway startup — same
pattern as `gateway_timeout`). Set to 0 to disable the gate.
- `_coerce_gateway_timestamp` now explicitly rejects bool (which is a
subclass of int and would otherwise coerce to 0.0/1.0).
- Tests rewritten to exercise the real production data shape: raw
`history` → `_build_agent_history` strip → freshness decision. A
regression guard (`test_stale_tool_tail_with_production_data_shape`)
asserts `agent_history` tool rows carry NO timestamp, protecting
against someone "fixing" the original bug by re-adding the stripped
field (which would break the OpenAI tool-result message contract).
Add BeliefanX to scripts/release.py AUTHOR_MAP.
E2E verified: config.yaml → env var bridge → helper returns configured
value; default 1h window; malformed/empty env var falls back to default;
ISO-Z timestamps parse; ms-epoch coerced; bool rejected.
Extract the islink/realpath guard from the 16743 fix into a single
atomic_replace() helper in utils.py, then migrate every os.replace()
call site in the codebase to use it.
The original PR #16777 correctly identified and fixed the bug, but
only patched 9 of ~24 call sites. The same bug class (managed
deployments that symlink state files silently losing the link on
every write) still existed at auth.json, sessions file, gateway
config, env_loader, webhook subscriptions, debug store, model
catalog, pairing, google OAuth, nous rate guard, and more.
Rather than add another 10+ copies of the same three-line guard,
consolidate into atomic_replace(tmp, target) which:
- resolves symlinks via os.path.realpath before os.replace
- returns the resolved real path so callers can re-apply permissions
- is a drop-in replacement for os.replace at the use sites
Changes:
- utils.py: new atomic_replace() helper + atomic_json_write /
atomic_yaml_write now call it instead of inlining the guard
- 16 files: all os.replace() call sites migrated to atomic_replace()
- agent/{google_oauth, nous_rate_guard, shell_hooks}.py
- cron/jobs.py
- gateway/{pairing, session, platforms/telegram}.py
- hermes_cli/{auth, config, debug, env_loader, model_catalog, webhook}.py
- tools/{memory_tool, skill_manager_tool, skills_sync}.py
Tests: tests/test_atomic_replace_symlinks.py pins the invariant for
atomic_replace + atomic_json_write + atomic_yaml_write, covers plain
files, first-time creates, broken symlinks, and permission preservation.
Refs #16743
Builds on #16777 by @vominh1919.
os.replace(tmp, path) replaces the symlink itself with a regular file,
breaking users who symlink config.yaml, SOUL.md, or .env from ~/.hermes/
to a dotfiles repo or managed profile package.
Fix: resolve symlinks via os.path.realpath() before os.replace(), so the
real file is overwritten in-place while the symlink survives.
Fixed in 7 files covering all os.replace call sites:
- utils.py (atomic_json_write, atomic_yaml_write — fixes save_config)
- hermes_cli/config.py (env sanitizer, save_env_value, remove_env_value)
- tools/skill_manager_tool.py (_atomic_write_text — SOUL.md writes)
- tools/memory_tool.py (memory file writes)
- tools/skills_sync.py (manifest writes)
- cron/jobs.py (job state + output file writes)
- agent/shell_hooks.py (hook file writes)
FixesNousResearch/hermes-agent#16743
Real OpenClaw configs key agents.defaults.models by full provider/model
API ID with an 'alias' field on the value (e.g.
{'anthropic/claude-opus-4-6': {'alias': 'Claude Opus 4.6'}}). Add
regression tests for issue #16745 covering:
- reverse-lookup of alias against real schema (keyed by API ID)
- alias resolution when model is a bare string vs {'primary': ...}
- passthrough when the value is already a provider/model API ID
- passthrough when the alias has no catalog match
- string-valued catalog entries (belt-and-suspenders)
- no catalog at all
`hermes claw migrate` copied OpenClaw's model setting verbatim, which
could be a display alias (e.g. "Claude Opus 4.6") instead of the actual
API ID (e.g. "claude-opus-4-6"). Hermes then sent the alias to the API,
causing HTTP 404 model not found.
Fix: look up the model string in agents.defaults.models (plural) alias
catalog. If found, use the resolved "id" field, prepending the provider
prefix if needed. If not found (already an API ID), pass through unchanged.
FixesNousResearch/hermes-agent#16745
DeepSeek API returns HTTP 400 with 'Insufficient Balance' message when
account funds are depleted. This pattern was not in _BILLING_PATTERNS,
causing the error to be misclassified instead of triggering billing
exhaustion handling (e.g., fallback to alternate provider).
Suggested by teknium1 in PR review of #15586.
Adds tools.schema_sanitizer.strip_nullable_unions as the single
implementation for collapsing anyOf/oneOf nullable unions. Both the
MCP input-schema normalizer and the Anthropic tool-schema guard now
delegate to it instead of re-implementing the same walk three times.
The global sanitizer also gains a final pass so any tool that slips
past the two earlier hooks (plugin tools, non-MCP custom tools with
Pydantic-shaped schemas) still gets safe input_schemas on Anthropic.
- tools/schema_sanitizer.py:
* New public strip_nullable_unions(schema, keep_nullable_hint=True).
* _sanitize_single_tool() calls it as a final pass (hint preserved
so coerce_tool_args can still map string "null" to None).
- tools/mcp_tool.py: _normalize_mcp_input_schema delegates.
- agent/anthropic_adapter.py: _normalize_tool_input_schema delegates
with keep_nullable_hint=False (Anthropic does not recognize nullable).
No behavioral change for the fix itself; tests (73/73 targeted +
E2E across MCP→sanitizer→Anthropic paths) pass.
25 new tests (all Bedrock API calls mocked, no real AWS creds needed):
tests/hermes_cli/test_bedrock_model_picker.py (20 tests):
- provider_model_ids("bedrock") uses live discovery, returns regional
model IDs, falls back gracefully on empty/exception, resolves all
bedrock aliases (aws, aws-bedrock, amazon-bedrock) to live discovery
- list_authenticated_providers() section 2: bedrock appears with AWS
creds, model list from discover_bedrock_models(), total_models
matches, is_current flag works, absent creds hides bedrock, discovery
failure does not crash, no duplicate entries
- Region routing: botocore profile eu-central-1 yields eu.* model IDs
end-to-end; env var takes priority over botocore profile
- providers.py overlay: exists with correct transport/auth_type, label
is non-empty, all aliases normalize to bedrock
tests/agent/test_bedrock_adapter.py (5 tests):
- resolve_bedrock_region() botocore profile fallback, botocore failure
fallback, us-east-1 hard fallback (with botocore mocked)
provider_model_ids("bedrock") fell through to a static _PROVIDER_MODELS
table containing only hardcoded us.* model IDs. Users configured for
non-US AWS regions (eu-central-1, ap-northeast-1, etc.) saw wrong or no
models in /model and autocomplete.
Root causes fixed:
1. models.py: provider_model_ids() now calls discover_bedrock_models()
keyed by the resolved region before falling back to the static table.
A new bedrock_model_ids_or_none() helper in bedrock_adapter.py
consolidates the discover -> extract IDs -> fallback pattern used by
all three call sites.
2. providers.py: registers bedrock in HERMES_OVERLAYS with
transport=bedrock_converse and auth_type=aws_sdk so
get_provider("bedrock") and resolve_provider_full("bedrock") work.
3. model_switch.py: list_authenticated_providers() sections 2 and 3
detect AWS credentials via has_aws_credentials() for aws_sdk
overlays and use live discovery for the model list.
4. bedrock_adapter.py: resolve_bedrock_region() reads the configured
region from botocore.session before falling back to us-east-1,
covering users who set their region in ~/.aws/config via a named
profile rather than env vars.
5. tui_gateway/server.py: passes provider= to get_model_context_length()
so context window lookups work correctly for the Bedrock provider.
* fix(anthropic): remove Claude Code fingerprinting from OAuth Messages API path
OAuth requests now identify as Hermes on the wire. Removed:
- "You are Claude Code, Anthropic's official CLI for Claude." system
prompt prepend
- Hermes Agent → Claude Code / Nous Research → Anthropic
system-prompt substitutions
- mcp_ tool-name prefix on outgoing tool schemas + message history
- Matching mcp_ strip on inbound tool_use blocks (strip_tool_prefix path
removed from AnthropicTransport.normalize_response, + all 5 call
sites in run_agent.py and auxiliary_client.py)
- user-agent: claude-cli/<v> (external, cli) and x-app: cli headers on
the Messages API client
Added:
- OAuth path strips context-1m-2025-08-07 — Anthropic rejects OAuth
requests carrying it with HTTP 400 'This authentication style is
incompatible with the long context beta header.'
Kept (auth plumbing, not identity spoofing):
- _is_oauth_token classifier and is_oauth flag threading
- Bearer vs x-api-key auth routing
- _OAUTH_ONLY_BETAS (claude-code-20250219, oauth-2025-04-20) — backend
requires these on the OAuth-gated Messages endpoint
- _OAUTH_CLIENT_ID (Claude Code's) — Anthropic doesn't issue OAuth
creds to third parties; this is the only way the login flow works
- claude-cli/<v> User-Agent on the OAuth token exchange + refresh
endpoints at platform.claude.com/v1/oauth/token — bare requests get
Cloudflare 1010 blocked
Verified live against api.anthropic.com with a fresh sk-ant-oat01-*
token:
- claude-haiku-4-5 simple message: HTTP 200, 'OK' response
- claude-haiku-4-5 tool call: HTTP 200, stop_reason=tool_use, tool
named 'terminal' (no mcp_ prefix) round-tripped correctly
- Outgoing wire: no user-agent, no x-app, real Hermes identity in
system prompt, real tool name in schema
Closes/supersedes #16820 (mcp_ PascalCase normalization patch — no longer
needed since the mcp_ round-trip is gone).
* fix(anthropic): resolve_anthropic_token() reads credential pool first
Close the gap where ~/.hermes/auth.json → credential_pool.anthropic
(where hermes login + dashboard PKCE flow write OAuth tokens) was not
in resolve_anthropic_token()'s source list.
Before: users who authed via hermes login got the token written into
the pool, but legacy fallback code paths (auxiliary_client, models
catalog fetch, explicit-runtime path) that call resolve_anthropic_token()
saw None and raised 'No Anthropic credentials found' — even though the
token was sitting in auth.json.
New priority 1: pool.select() with env-sourced entries skipped. Skipping
env:* entries preserves the existing env-var priority logic further
down the chain (static env OAuth → refreshable Claude Code upgrade via
_prefer_refreshable_claude_code_token).
Surfaced while writing the hermes-agent-dev skill playbook for
'finding a live OAuth token for an E2E test'.
---------
Co-authored-by: teknium1 <teknium@users.noreply.github.com>
Adds a pre-call sanitizer that detects assistant messages containing only
reasoning (reasoning / reasoning_content, no visible content, no
tool_calls) and drops them from the API copy. Adjacent user messages
left behind are merged so role alternation is preserved for the
provider.
Mirrors Claude Code's approach in src/utils/messages.ts
(filterOrphanedThinkingOnlyMessages + mergeAdjacentUserMessages). We
drop the whole turn rather than fabricate stub text (the '.' /
'(continued)' pattern from contributor PRs #11098, #13010, #16842 that
were rejected because they put words in the model's mouth).
The stored conversation history (self.messages) is never mutated — only
the per-call api_messages copy. Users still see the reasoning block in
the CLI/gateway transcript; only the wire copy is cleaned. Session
persistence keeps the full trace.
Two call sites covered:
- Main agent loop, after _sanitize_api_messages (catches every turn).
- Iteration-limit-summary fallback path.
Tests: tests/run_agent/test_thinking_only_sanitizer.py — 25 cases
covering detection (string/list content, whitespace-only, tool_calls,
reasoning_details list form), drop behavior, adjacent-user merge
(string+string, list+list, mixed), non-mutation of input dicts, and
system-message handling.
E2E live-tested against 5 providers with a poisoned history (empty
assistant message + reasoning_content): OpenRouter→Anthropic/OpenAI/
DeepSeek-R1/Qwen, native Gemini. All 5 accepted the cleaned request.
Happy-path regression (5/5) confirms the sanitizer is a noop when no
thinking-only turn exists.
Related: #16823 (wontfix — stub-text approach rejected).
Co-authored-by: teknium1 <teknium@users.noreply.github.com>
Registers tencent-tokenhub (https://tokenhub.tencentmaas.com/v1) as a
new API-key provider with model tencent/hy3-preview (256K context).
- PROVIDER_REGISTRY entry + TOKENHUB_API_KEY / TOKENHUB_BASE_URL env vars
- Aliases: tencent, tokenhub, tencent-cloud, tencentmaas
- openai_chat transport with is_tokenhub branch for top-level
reasoning_effort (Hy3 is a reasoning model)
- tencent/hy3-preview:free added to OpenRouter curated list
- 60+ tests (provider registry, aliases, runtime resolution,
credentials, model catalog, URL mapping, context length)
- Docs: integrations/providers.md, environment-variables.md,
model-catalog.json
Author: simonweng <simonweng@tencent.com>
Salvaged from PR #16860 onto current main (resolved conflicts with
#16935 Azure Anthropic env-var hint tests and the --provider choices=
list removal in chat_parser).
Three related fixes around custom env-var-name hints for provider entries.
1. Azure Anthropic path: previously hardcoded to look up AZURE_ANTHROPIC_KEY
then ANTHROPIC_API_KEY with no way to override. If a user wrote
model:
provider: anthropic
base_url: https://my-resource.services.ai.azure.com/anthropic
key_env: MY_CUSTOM_KEY
the key_env hint was silently ignored and the resolver raised
'No Azure Anthropic API key found' even when MY_CUSTOM_KEY was set
in the environment. The runtime now checks, in order:
(1) os.getenv(model_cfg.key_env)
(2) os.getenv(model_cfg.api_key_env) # docs alias
(3) model_cfg.api_key # inline value
(4) AZURE_ANTHROPIC_KEY # historical default
(5) ANTHROPIC_API_KEY # historical default
Error message updated to mention key_env as an option.
2. Provider entry normalizer (_normalize_custom_provider_entry): accept
'api_key_env' as a snake_case alias for 'key_env', and 'apiKeyEnv' as a
camelCase alias. Adds both to the _KNOWN_KEYS set so the 'unknown
config keys ignored' warning doesn't fire on valid configs.
3. _VALID_CUSTOM_PROVIDER_FIELDS: add 'key_env'. That set documents
supported custom_providers entry fields; it was drifting from reality
since key_env has been read at runtime in auxiliary_client.py,
runtime_provider.py, and main.py for a while.
Docs: website/docs/guides/azure-foundry.md now uses the canonical key_env
field and notes that api_key_env / keyEnv / apiKeyEnv are accepted as
aliases.
Validation: 12 new tests in test_runtime_provider_resolution.py covering
all 5 Azure Anthropic resolution paths + 4 normalizer-alias tests. Pass
rate across related suites (165 + 46 tests): 100%.
Co-authored-by: teknium1 <teknium@users.noreply.github.com>
The mention_user_id injection from #38a6bada9 unconditionally attached an
@user:server mention pill + MSC3952 m.mentions.user_ids payload to every
outbound reply and every tool-progress status update. The stated intent
was push notifications in muted rooms, but shipped as always-on in every
room, DM or group, muted or not — so every reply pinged the user.
- gateway/platforms/base.py: stop injecting mention_user_id into send
metadata on every reply; restore the original _thread_metadata passthrough.
- gateway/run.py: drop mention_user_id from status-thread metadata.
- gateway/platforms/matrix.py: drop the mention-pill append block in
_send_text that consumed the metadata. Keep the reaction-based exec
approval half of #38a6bada9 and the inbound/outbound m.mentions
handling (unrelated to the per-reply ping).
Reported by Elkim [NOUS] on Discord.
Co-authored-by: teknium1 <teknium@users.noreply.github.com>
* feat(claw-migrate): harden OpenClaw import with plan-first apply, redaction, and pre-migration backup
Adopts four design patterns from OpenClaw's reciprocal migrate-hermes
importer so both migration paths have the same safety posture.
- **Refuse-on-conflict apply.** 'hermes claw migrate' now refuses to
execute when the plan has any conflict items, unless --overwrite is
set. Previously the user could say 'yes, proceed' and end up with a
silent partial migration that skipped every conflicting item.
- **Engine-level secret redaction.** The report.json and summary.md
written to disk (and --json stdout) run through a redactor that
matches OpenClaw's key-name markers and value-shape patterns
(sk-*, ghp_*, xox*-, AIza*, Bearer *). Prevents accidental API key
leakage in bug reports and support channels.
- **Pre-migration tarball snapshot.** Apply creates one timestamped
restore-point archive of ~/.hermes/ at ~/.hermes/migration/pre-migration-backups/
before any mutation, excluding regenerable directories
(sessions, logs, cache). Opt out with --no-backup.
- **Blocked-by-earlier-conflict sequencing.** If a config.yaml write
hits conflict/error mid-apply, subsequent config-mutating options
are marked skipped with reason 'blocked by earlier apply conflict'
rather than attempting partial writes.
- **Structured warnings[] and next_steps[] on the report** — actionable
guidance surfaces in both JSON output and summary.md.
- **--json output mode** — emits the redacted report on stdout for CI.
Also flips --preset full to NOT auto-enable --migrate-secrets. Users
now have to opt in to secret import explicitly, mirroring OpenClaw's
two-phase posture.
Status/kind/action constants are defined (STATUS_MIGRATED etc) with
values that match the existing strings the script emits, so the
report schema is backward-compatible. ItemResult gains a 'sensitive'
bool field that redaction and consumers can key off.
Validation: 26 new unit tests + 1 updated test in tests/skills/
test_openclaw_migration_hardening.py and test_claw.py cover redaction
(key markers, value patterns, recursion, on-disk), warnings/next_steps,
blocked-by-earlier sequencing, --json mode, and the preset-flip.
Manual E2E against a fake $HERMES_HOME with real-shaped secrets
confirmed: (1) secrets never appear in stdout or on disk,
(2) _cmd_migrate refuses apply when plan has conflicts,
(3) --overwrite proceeds past the guard and the backup tarball is
created, (4) --no-backup skips the archive.
Related docs: website/docs/guides/migrate-from-openclaw.md and
website/docs/reference/cli-commands.md updated to reflect the
preset-flip and new --no-backup flag.
* refactor(claw-migrate): reuse hermes backup system for pre-migration snapshot
Drops the inline tarball in hermes_cli/claw.py in favor of
hermes_cli.backup.create_pre_migration_backup(), which shares an
implementation with create_pre_update_backup via a new
_write_full_zip_backup helper. Benefits:
- Consistent exclusion rules with hermes backup (_EXCLUDED_DIRS,
_EXCLUDED_SUFFIXES, _EXCLUDED_NAMES — single source of truth).
- SQLite safe-copy via _safe_copy_db (state.db restores cleanly).
- Zip format restorable with 'hermes import <archive>'.
- Lives under ~/.hermes/backups/pre-migration-*.zip alongside
pre-update-*.zip — one place for all snapshot archives.
- Auto-prune rotation with separate keep counters (pre-migration
keeps 5, pre-update keeps 5, they don't touch each other's files).
7 new tests in tests/hermes_cli/test_backup.py lock the contract:
directory location, shared exclusion rules, _validate_backup_zip
acceptance (i.e. restorable with 'hermes import'), non-recursive
into prior backups, rotation, missing-home handling, and the
invariant that pre-migration rotation never touches pre-update
backups.
Help text and docs updated — the restore hint now says
'hermes import <name>' instead of 'tar -xzf <archive> -C ~/'.
* chore(claw-migrate): use backup._format_size and drop duplicate output line
Minor polish using another existing primitive from hermes_cli.backup:
- Show backup archive size with _format_size (e.g. '(245 B)' or '(2.4 MB)')
matching the format hermes backup already uses.
- Drop the duplicate 'Pre-migration backup saved' line after Migration
Results — the earlier 'Pre-migration backup: <path> (<size>)' line
already surfaces the path before apply runs.
---------
Co-authored-by: teknium1 <teknium@users.noreply.github.com>
Follow-up to the static list refresh: replace the hardcoded xAI entries
with _xai_curated_models(), mirroring the _codex_curated_models()
pattern from PR #7844. The helper reads $HERMES_HOME/models_dev_cache.json
at import time (no network call) and falls back to a small static list
when the cache is missing or malformed.
Why: _PROVIDER_MODELS["xai"] has drifted once already (issue #16699) and
will drift again next time xAI renames a model. Hermes already maintains
the models.dev cache and uses it for context-length lookups; pointing
_PROVIDER_MODELS at the same source means the /model picker self-heals on
the next cache refresh instead of requiring a PR.
Behavior:
- With cache populated (normal user): shows every current xAI model ID,
picks up renames automatically on next refresh.
- Without cache (fresh install, offline): falls back to a static snapshot
of the 9 current flagship IDs.
- Malformed cache / unexpected shape: same static fallback, no crash.
Import time verified <20ms — disk read only, no HTTP.
Addresses the structural piece of #16699 ("consider a single
_provider_models(provider) resolver") for xAI. Other per-provider lists
can adopt the same pattern as drift is observed.
_PROVIDER_MODELS["xai"] was pointing at model IDs the xAI direct API
no longer accepts:
- grok-4.20-reasoning
- grok-4-1-fast-reasoning
Replaced with the actual current xAI catalog IDs from models.dev
($HERMES_HOME/models_dev_cache.json, mirror of https://models.dev/api.json):
grok-4.20-0309-reasoning
grok-4.20-0309-non-reasoning
grok-4.20-multi-agent-0309
grok-4-1-fast
grok-4-1-fast-non-reasoning
grok-4-fast
grok-4-fast-non-reasoning
grok-4
grok-code-fast-1
The xAI-direct API (https://api.x.ai/v1) serves the dated IDs shown
above; the bare aliases (grok-4.20, grok-4.1-fast, etc.) are
OpenRouter/Vercel-gateway normalizations and are not accepted on
xAI-direct. Those gateways remain unaffected.
Fixes#16699
Follow-up to PR #16819 applying the same treatment to the two sibling
fallback sites in resolve_provider_client() that carry the identical bug
class as the anonymous-custom branch:
- Named custom provider (providers: / custom_providers: config entries):
apply _to_openai_base_url() on the OpenAI-wire path (chat_completions /
codex_responses), leave custom_base untouched on the anthropic_messages
path where the /anthropic surface is intentional. Prefer
main_runtime.get('model') over _read_main_model() so the entry model
still wins first. The ImportError fallback for anthropic_messages now
redoes query-param extraction against the rewritten URL so the final
OpenAI client hits /v1.
- external_process branch (copilot-acp): same main_runtime.get('model')
fallback before _read_main_model() so auxiliary tasks on this provider
track live /model switches instead of stale config.yaml.
Keeps the fix consistent across all three custom-endpoint fallback sites
in resolve_provider_client().
Three related issues prevented user-defined providers in `providers:` and
`model_aliases:` from being reachable through standard CLI flags. Requests
silently routed to the configured `model.base_url` instead of the user-
intended endpoint.
* hermes_cli/model_switch.py — root cause of the silent misrouting:
`_ensure_direct_aliases()` rebound `DIRECT_ALIASES` to a freshly-loaded
dict, leaving every `from hermes_cli.model_switch import DIRECT_ALIASES`
caller stuck on the stale empty original. Switched to `.update()` so
module attribute references stay valid.
* hermes_cli/main.py — chat subcommand `--provider` had `choices=[...]`
hardcoded to built-in providers, rejecting valid keys from user
`providers:` config. Dropped the choices list; runtime resolution
validates correctly downstream.
* hermes_cli/oneshot.py — `-m <alias>` only resolved the model name; the
alias's base_url was never propagated. Now consults `DIRECT_ALIASES`
before falling through to `detect_provider_for_model`, and threads the
alias's base_url to `resolve_runtime_provider(explicit_base_url=...)`.
* hermes_cli/runtime_provider.py — `_resolve_named_custom_runtime` now
honors `(provider="custom", explicit_base_url=...)` so a base_url
propagated from a direct-alias resolution actually builds a runtime
instead of falling through to provider-registry handlers that don't
know about ad-hoc local endpoints.
Verified: `hermes chat --provider <user-key> -m <model> -q "..."` and
`hermes -m <user-alias> -z "..."` both route to the user-intended
endpoint, observable via the target server's request log.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Follow-up to #15328's vision-unsupported retry branch in run_agent.py.
_strip_images_from_messages() previously deleted any message whose content
was entirely images. That's fine for synthetic user messages injected for
attachment delivery, but it breaks providers for tool-role messages — the
paired tool_call_id on the preceding assistant message ends up unmatched,
which OpenAI-compatible APIs reject with HTTP 400.
Fix: tool-role messages whose content becomes empty are replaced with a
plaintext placeholder that preserves the tool_call_id linkage. Only
non-tool messages are dropped. Added 10 tests covering the role-alternation
invariants + image-type coverage.
Image-rejection detector: expanded phrase list (image content not
supported / multimodal input / vision input / model does not support
image) and gated on 4xx status so transient 5xx errors never get
misinterpreted as 'server said no to images'. Detection is documented as
best-effort English phrase matching.
AUTHOR_MAP: mapped 3820588+ddupont808@users.noreply.github.com to
ddupont808 so release notes attribute the salvage correctly.
Tool handlers (e.g. computer_use capture) return a _multimodal envelope
dict when a screenshot is attached. The tool-message builder was passing
this raw dict as the `content` field of role:tool messages, which is an
illegal format — OpenAI-compatible APIs expect a string or a content-parts
list, not a plain Python dict, and would reject it with a 400/422 error.
Fix: unwrap _multimodal results to their `content` list
([{type:text,...},{type:image_url,...}]) in both the parallel and
sequential tool-call paths. The Anthropic adapter already handles content
lists natively; vision-capable OpenAI-compatible servers (mlx-vlm,
GPT-4o, etc.) accept image_url parts in tool messages directly.
Also add a _vision_supported adaptive fallback: on first image-rejection
error ("Only 'text' content type is supported." etc.) the agent strips all
image parts from the message history and retries with text only, so
text-only endpoints degrade gracefully without crashing the session.
Extends the cua-driver computer-use backend to drive backgrounded macOS
windows without stealing keyboard or mouse focus from the foreground app.
All changes target the cua-driver MCP backend and the shared dispatcher.
## cua_backend.py
**Window-aware capture**: capture() now calls list_windows + get_window_state
instead of the removed capture tool. Prefers structuredContent.windows
(MCP 2024-11-05+ cua-driver) for zero-parse window enumeration; falls back
to regex-parsed text for older builds. Stores the selected (pid, window_id)
as sticky context so subsequent action calls do not need a redundant round-trip.
**Action routing**: click/scroll/type_text/key all carry the sticky pid
(and window_id for element-indexed clicks). type_text routes through
type_text_chars (individual key events) rather than AX attribute write --
WebKit AXTextFields reject attribute writes from backgrounded processes.
**Key parsing**: _parse_key_combo splits cmd+s-style strings into
(key, [modifiers]) and routes to hotkey (modifier present) or
press_key (bare key) -- cua-driver actual tool names.
**set_value method**: new set_value(value, element) calls the cua-driver
set_value MCP tool. For AXPopUpButton / HTML select in a backgrounded Safari,
AXPress opens the native macOS popup which closes immediately when the app is
non-frontmost; set_value AX-presses the matching child option directly
(no menu required, no focus steal).
**focus_app**: reimplemented as a pure window-selector (enumerates
list_windows, sets sticky pid/window_id) without ever raising the window
or stealing focus.
**list_apps**: fixed tool name from listApps to list_apps; handles plain-text
response via regex when structured data is absent.
**Structured-content extraction**: _extract_tool_result now surfaces
structuredContent from MCP results, enabling the list_windows window array
without text parsing.
**Helpers**: _parse_windows_from_text, _parse_elements_from_tree,
_split_tree_text, _parse_key_combo extracted as module-level functions.
## schema.py
Added set_value to the action enum with a description explaining when to
prefer it over click (select/popup elements, sliders, no focus steal).
Added value field for set_value payloads.
## tool.py
Routed set_value action through _dispatch to backend.set_value.
Added set_value to _DESTRUCTIVE_ACTIONS (approval-gated).
Fixed MIME-type detection in _capture_response: cua-driver may return
JPEG; detect from base64 magic bytes (/9j/ -> image/jpeg, else image/png)
rather than hardcoding image/png.
## agent/display.py + run_agent.py
Guard _detect_tool_failure and result-preview logic against non-string
function_result values: multimodal tool results (dicts with _multimodal=True)
are not string-sliceable; treat them as successes and fall back to str()
for length/preview.
Background macOS desktop control via cua-driver MCP — does NOT steal the
user's cursor or keyboard focus, works with any tool-capable model.
Replaces the Anthropic-native `computer_20251124` approach from the
abandoned #4562 with a generic OpenAI function-calling schema plus SOM
(set-of-mark) captures so Claude, GPT, Gemini, and open models can all
drive the desktop via numbered element indices.
- `tools/computer_use/` package — swappable ComputerUseBackend ABC +
CuaDriverBackend (stdio MCP client to trycua/cua's cua-driver binary).
- Universal `computer_use` tool with one schema for all providers.
Actions: capture (som/vision/ax), click, double_click, right_click,
middle_click, drag, scroll, type, key, wait, list_apps, focus_app.
- Multimodal tool-result envelope (`_multimodal=True`, OpenAI-style
`content: [text, image_url]` parts) that flows through
handle_function_call into the tool message. Anthropic adapter converts
into native `tool_result` image blocks; OpenAI-compatible providers
get the parts list directly.
- Image eviction in convert_messages_to_anthropic: only the 3 most
recent screenshots carry real image data; older ones become text
placeholders to cap per-turn token cost.
- Context compressor image pruning: old multimodal tool results have
their image parts stripped instead of being skipped.
- Image-aware token estimation: each image counts as a flat 1500 tokens
instead of its base64 char length (~1MB would have registered as
~250K tokens before).
- COMPUTER_USE_GUIDANCE system-prompt block — injected when the toolset
is active.
- Session DB persistence strips base64 from multimodal tool messages.
- Trajectory saver normalises multimodal messages to text-only.
- `hermes tools` post-setup installs cua-driver via the upstream script
and prints permission-grant instructions.
- CLI approval callback wired so destructive computer_use actions go
through the same prompt_toolkit approval dialog as terminal commands.
- Hard safety guards at the tool level: blocked type patterns
(curl|bash, sudo rm -rf, fork bomb), blocked key combos (empty trash,
force delete, lock screen, log out).
- Skill `apple/macos-computer-use/SKILL.md` — universal (model-agnostic)
workflow guide.
- Docs: `user-guide/features/computer-use.md` plus reference catalog
entries.
44 new tests in tests/tools/test_computer_use.py covering schema
shape (universal, not Anthropic-native), dispatch routing, safety
guards, multimodal envelope, Anthropic adapter conversion, screenshot
eviction, context compressor pruning, image-aware token estimation,
run_agent helpers, and universality guarantees.
469/469 pass across tests/tools/test_computer_use.py + the affected
agent/ test suites.
- `model_tools.py` provider-gating: the tool is available to every
provider. Providers without multi-part tool message support will see
text-only tool results (graceful degradation via `text_summary`).
- Anthropic server-side `clear_tool_uses_20250919` — deferred;
client-side eviction + compressor pruning cover the same cost ceiling
without a beta header.
- macOS only. cua-driver uses private SkyLight SPIs
(SLEventPostToPid, SLPSPostEventRecordTo,
_AXObserverAddNotificationAndCheckRemote) that can break on any macOS
update. Pin with HERMES_CUA_DRIVER_VERSION.
- Requires Accessibility + Screen Recording permissions — the post-setup
prints the Settings path.
Supersedes PR #4562 (pyautogui/Quartz foreground backend, Anthropic-
native schema). Credit @0xbyt4 for the original #3816 groundwork whose
context/eviction/token design is preserved here in generic form.
Opt-in Langfuse tracing for Hermes conversations — LLM calls, tool
usage, usage/cost breakdown per span. Hooks into pre/post_api_request,
pre/post_llm_call, pre/post_tool_call. SDK is optional; missing SDK or
credentials renders the plugin inert.
Salvaged from PR #16845 by @kshitijk4poor, who wrote the plugin
(~875 LOC, 6 hooks, Langfuse usage-details/cost-details normalization,
read_file payload summarization).
Salvage scope (why this isn't PR #16845 as-authored):
- Lives at plugins/observability/langfuse/ (standalone kind, opt-in via
plugins.enabled) instead of a new parallel optional-plugins/
directory. Standalone bundled plugins are already opt-in — only their
plugin.yaml is scanned at startup; the Python module is not imported
unless the user enables it. The premise of optional-plugins/ (avoid
import cost for users who don't want it) is already solved by the
existing plugin system.
- Dropped the triple activation gate (plugins.enabled +
plugins.langfuse.enabled + HERMES_LANGFUSE_ENABLED). The Hermes plugin
system's own enable/disable is authoritative; runtime credentials
gate whether the hook actually traces.
- Rewrote _is_enabled() → cached _get_langfuse() with an _INIT_FAILED
sentinel. The original called hermes_cli.config.load_config() from
every hook invocation (full yaml parse + deep merge + env expansion
on every pre/post_tool_call, potentially 100+ times per turn). The
cached version reads env once and returns the cached client or None
on every subsequent call with zero further work.
- hermes tools → Langfuse Observability post-setup adds
observability/langfuse to plugins.enabled directly (via
_save_enabled_set) instead of going through an install-copy flow.
Enable:
hermes tools # interactive
hermes plugins enable observability/langfuse # manual
Required env (set by `hermes tools` or in ~/.hermes/.env):
HERMES_LANGFUSE_PUBLIC_KEY
HERMES_LANGFUSE_SECRET_KEY
HERMES_LANGFUSE_BASE_URL # optional
Co-authored-by: kshitijk4poor <kshitijk4poor@gmail.com>
Narrow plaintext shortcut that rewrites a tiny set of admin phrases
("restart gateway", "restart the gateway", "restart hermes") into the
/restart slash command, but only in DMs. Scope is intentionally tight:
- DM text messages only — group chats keep natural-language semantics
- Exact restart-style phrases only
- Skips anything already starting with "/"
Without this, the LLM can receive "restart gateway" as a user turn and
try to satisfy it via the terminal tool (systemctl restart ...). That
kills the gateway while the originating agent is still running, which
leaves systemd in "draining" state waiting on a process it's about to
kill. Routing the phrase to the slash-command dispatcher bypasses the
agent loop and uses the existing restart machinery (request_restart).
Called once, at the adapter level in BasePlatformAdapter.handle_message,
so every platform gets it for free and pending-message reinjection is
covered by the same call site.
Adds 2 Telegram-parametrized e2e tests: DM routes to request_restart,
group chats fall through to the normal agent path.
Runtime already supports list-form fallback_model (run_agent.py:1459
iterates fallback_chain; fallback_cmd.py migrates legacy single-dict
configs to list format). The config validator and save_config comment
gate still assumed single-dict form and flagged list-form configs as
errors. Fix both:
- validate_config_structure: when fallback_model is a list, validate
each entry has provider+model; keep the existing single-dict path.
- save_config: suppress the "add fallback_model" comment when any list
entry is well-formed.
Adds 4 list-form validator tests.
PR #16858's session-scoped interactive sudo password cache falls back to
a thread-identity scope when no HERMES_SESSION_KEY is bound. ACP never
set that contextvar, so two ACP sessions landing on the same reused
ThreadPoolExecutor thread still shared the cache — the exact scenario
the PR headlined.
acp_adapter/server.py now:
- binds HERMES_SESSION_KEY=<session_id> via gateway.session_context
inside _run_agent() (and clears on exit)
- wraps the loop.run_in_executor(_executor, _run_agent) call in a fresh
contextvars.copy_context() so concurrent ACP sessions don't stomp on
each other's ContextVar writes (executor pool threads would otherwise
share a context).
Adds tests/acp/test_approval_isolation.py::
test_sudo_password_cache_isolated_across_acp_sessions_on_same_pool_thread
which drives two back-to-back sessions through a 1-worker ThreadPoolExecutor
and asserts B does not observe A's cached password.
Follow-up on top of the cherry-picked contributor commit for #16751:
1. Delete triggers: the original PR switched FTS5 from external to inline
content mode and concatenated content || tool_name || tool_calls in
the insert/update triggers, but left the delete triggers passing
old.content to the FTS5 delete-command. FTS5 inline delete requires
the content to match what was stored, so every DELETE on messages
raised 'SQL logic error'. Replaced with plain DELETE FROM ... WHERE
rowid = old.id on all four delete paths (normal + trigram, delete +
update-delete).
2. v11 migration: existing DBs have the old external-content FTS tables
and triggers. Because CREATE VIRTUAL TABLE IF NOT EXISTS / CREATE
TRIGGER IF NOT EXISTS skip when the objects already exist, upgraders
would have kept the broken behavior forever. Bumped SCHEMA_VERSION
to 11 and added a migration that drops both FTS tables + all 6 old
triggers, recreates them via FTS_SQL / FTS_TRIGRAM_SQL, and backfills
from messages using the same concatenation expression.
3. Regression tests: 6 new tests cover INSERT / UPDATE / DELETE paths
for tool_name + tool_calls indexing plus the full v10 -> v11 upgrade
path on a hand-built legacy DB.
The FTS5 virtual tables (messages_fts, messages_fts_trigram) previously
only indexed the content column via external content mode. Tool calls
and tool names stored in the tool_calls (JSON) and tool_name columns
were invisible to FTS5 search.
Root cause: FTS5 triggers only INSERTed new.content into the index.
Changes:
- Switch FTS5 tables from external content (content=messages) to inline
mode so that trigger-inserted content is both indexed and stored
- Update all 6 FTS5 triggers to concatenate content, tool_name, and
tool_calls when indexing new messages
- Extend the short-CJK LIKE fallback to also search tool_name and
tool_calls columns
Closes: #16751
FTS5 default tokenizer splits 'sp_new1' into tokens 'sp' and 'new1'.
Without quoting, a search for 'sp_new' becomes an AND query
('sp AND new') that fails to match rows indexed as 'sp_new1'.
Fix: add underscore to the character class in Step 5 regex
([.-] -> [._-]) so underscored terms are wrapped in double quotes.
Also adds test_sanitize_fts5_quotes_underscored_terms.
Both keys are documented in cli-config.yaml.example and read at runtime by
hermes_cli/timeouts.py (get_provider_request_timeout and get_provider_stale_timeout),
but the provider-entry validator in config.py flagged them as unknown, producing
noisy warnings on every CLI invocation for users who followed the documented config.
Fixes#16779
- App.tsx doc comment: replace stale ChatPageHost reference with
'persistent chat host block rendered inline near the bottom of this
file' so readers can find the actual code.
- App.tsx persistent host: show a small spinner on /chat while plugin
manifests are loading instead of a blank content area. Direct
/chat deep-links used to paint empty for up to ~2s in the worst
case (plugin-registration safety timeout) because both the route
sink (null) and the persistent host (!pluginsLoading gate) render
nothing during that window. Non-chat routes stay empty as before.
- ChatPage.tsx: rename setter to match the 'raw' state — useState
now destructures as [mobilePanelOpenRaw, setMobilePanelOpenRaw],
and all four call sites (closeMobilePanel, matchMedia listener,
open-button onClick, plus destructure) updated accordingly. No
behavior change; matches the 'raw vs derived' convention the
original comment set up.
The dashboard's Chat tab (hermes dashboard --tui) lost its session
whenever the user navigated to another tab and came back. React Router
unmounted ChatPage on path change, which ran the cleanup function,
closed the PTY WebSocket, and terminated the underlying TUI child -
so the next mount generated a fresh channel id, spawned a new PTY, and
started a brand-new conversation.
Rather than rebuild the destroyed state (session id capture + resume
via HERMES_TUI_RESUME would reload history from disk but drop in-flight
tool state, scrollback, and picker position), keep the component tree
alive.
* Pull ChatPage out of Routes into a sibling always-mounted host that
toggles visibility via display:none keyed off the current route. A
tiny ChatRouteSink still claims /chat so the catch-all redirect
does not fire.
* xterm instance, WebSocket, PTY child, and TUI/agent state all
survive; returning to /chat shows the exact conversation the user
left.
* Respect plugin `/chat` overrides: if a plugin manifest declares
`tab.override: "/chat"`, the Routes tree already swaps the element
for <PluginPage /> — we additionally suppress the persistent host
so the two don't paint on top of each other. Preserves the
pre-persistence contract that a plugin owning /chat replaces the
built-in chat UI entirely.
* Wait for usePlugins() to finish loading before mounting the
persistent host. Manifests arrive asynchronously from
/api/dashboard/plugins, so without the `!pluginsLoading` gate the
host would mount with manifests=[], spawn a PTY, and then unmount
mid-session when the manifest list resolves and reveals a /chat
override. Typical delay is <50ms; worst case is the 2s plugin-
registration safety timeout. Cheaper than killing someone's
conversation underneath them.
* Gate page-header slot (`setEnd`), the mobile sheet's portalled
render, and body-scroll lock on a new `isActive` prop so the hidden
ChatPage doesn't fight the active page for shared state. The
scroll-lock effect keys on the *derived* `mobilePanelOpen` (which is
`isActive && mobilePanelOpenRaw`) rather than the raw state — that
way tab-switch flips the dep false, fires the cleanup, and releases
`document.body.style.overflow`. Keying on the raw state would leave
body.overflow="hidden" stuck on /sessions and every other tab until
the user navigated back to /chat and explicitly closed the sheet.
* When isActive flips false to true, force a double-rAF fit:
display:none collapses the host box and ResizeObserver does not fire
on display changes, so xterm would otherwise stay at a stale or 1x1
grid. Also early-return from syncTerminalMetrics when the host has
zero area, since fit() on a zero-sized element produces a 1x1
terminal.
* Focus handling on tab return: only steal focus into the terminal if
focus wasn't already parked somewhere inside ChatPage (e.g. the
sidebar model picker, a tool-call entry). Yanking focus away from
whatever the user last clicked is surprising and a screen-reader
foot-gun; the typical "first activation" case still focuses the
terminal because document.activeElement is <body> at that point.
Trade-off worth flagging, deliberately not mitigated in this change:
while hidden, ChatPage still holds a PTY child + WebSocket + xterm
instance for the dashboard's full lifetime. The WS keeps delivering
bytes and xterm keeps parsing them into a display:none host (cheap —
no paint work, but not free). Reasonable costs to pay for the session
preservation; if they become a problem we can pause `term.write` when
!isActive or idle-disconnect after N minutes hidden.
Lint clean on touched files. tsc -b && vite build pass.
Follow-up to the salvaged PR #16867 that added the read path for
agent.disabled_toolsets in _get_platform_tools():
- Document the new config key under a "Global Toolset Disable" section
in website/docs/user-guide/configuration.md, including the precedence
note (global disable overrides per-platform platform_toolsets).
- Map nazirulhafiy@gmail.com -> nazirulhafiy in scripts/release.py
AUTHOR_MAP so release-notes CI attributes the cherry-picked commit.
Previously, agent.disabled_toolsets in config.yaml only worked for CLI
mode (run_agent.py --disabled_toolsets). The gateway always passed
enabled_toolsets to AIAgent, and get_tool_definitions() ignored
disabled_toolsets when enabled_toolsets was set.
Fix: _get_platform_tools() now reads agent.disabled_toolsets from config
and excludes those toolsets from the returned set. This runs last so it
overrides everything above.
Added 3 tests covering cross-platform suppression, explicit platform
config override, and empty/missing config no-op behavior.
Streaming-only providers (glm, MiniMax, gpt-5.x via aigw, Anthropic via
openai-compat shims) emit reasoning through delta.reasoning_content
chunks that get accumulated into the local reasoning_text string — but
never land on the assistant message object as a top-level attribute. The
prior guard at _build_assistant_message only wrote reasoning_content
when the SDK exposed hasattr(msg, 'reasoning_content'), so these
providers persisted the chain-of-thought under the internal 'reasoning'
key and omitted the protocol-standard field.
The poison was silent until the user later switched to a DeepSeek-v4 or
Kimi thinking model, at which point replay failed with HTTP 400:
'The reasoning_content in the thinking mode must be passed back to the
API.' One reported session store accumulated 4,031 poisoned messages
across 1,101 files (#16844).
Fix: add an additive fallback that promotes the already-sanitized
reasoning_text to reasoning_content when no earlier branch wrote it AND
reasoning text was actually captured. Layered on top of the existing
SDK-attr branch and DeepSeek ''-pad (#15250) rather than replacing them,
so every existing behavior is preserved:
- SDK-exposed reasoning_content (OpenAI/Moonshot/DeepSeek SDK) still
wins.
- DeepSeek tool-call ''-pad still fires when the SDK exposes the attr
but the value is None.
- Non-thinking turns with no reasoning leave the field absent, so
_copy_reasoning_content_for_api's cross-provider leak guard (#15748),
promote-from-'reasoning' tier, and thinking-pad tier remain live at
replay time.
- No empty '' gets eagerly written on every assistant turn (which would
have bypassed the read-side ladder and triggered empty thinking-block
insertion in the Anthropic adapter).
Tests: three new TestBuildAssistantMessage cases covering the streaming
promotion path, SDK precedence, and field-absent-when-no-reasoning
invariant.
Credit @Sanjays2402 for the original diagnosis and patch in #16884;
this is a scoped rework that preserves the existing read-side
compensation code as defense in depth.
Refs #16844, #16884, #15250, #15353, #15748.
Address Copilot review on #16868:
1. Tighten pool iteration. ``validate_copilot_token`` only rejects empty
strings and classic PATs (``ghp_*``); a malformed/unsupported ``gho_*``
token at ``credential_pool.copilot[0]`` would pass the gate and short-
circuit the loop, hiding a later valid entry. Switch to calling
``exchange_copilot_token`` directly: only entries that actually exchange
into a live Copilot API token are returned. Bad/expired entries fall
through to the next, and an exhausted pool returns ``""`` so the picker
falls back to the curated list (existing behaviour).
2. Reword the docstring + test module docstring to describe the pool seed
path accurately — ``hermes auth add copilot`` adds an api-key-typed
credential whose ``access_token`` field stores the pasted token, and
``_seed_from_env`` mirrors ``COPILOT_GITHUB_TOKEN`` from
``~/.hermes/.env`` into the pool. The previous wording implied
``auth add copilot`` itself ran the device-code flow, which it does
not (the device-code flow lives in ``hermes model``).
Two new tests cover the iteration change:
- ``test_skips_pool_entry_that_fails_to_exchange`` — pool[0] raises,
pool[1] succeeds, picker uses pool[1].
- ``test_all_pool_entries_fail_exchange_returns_empty`` — every entry
raises, return ``""``.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Users whose only Copilot credential is the OAuth `access_token` saved by
`hermes auth add copilot` (device-code flow) saw the `/model` picker drop
back to a stale hardcoded list. Reason: `_resolve_copilot_catalog_api_key`
only consulted env vars (`COPILOT_GITHUB_TOKEN` / `GH_TOKEN` /
`GITHUB_TOKEN`) and the `gh auth token` CLI fallback, never the credential
pool that Hermes's own login flow writes into `auth.json`. With no token,
the live catalog fetch silently 401s and the picker hides current models
(claude-opus-4.7, claude-sonnet-4.6, gpt-5.5, grok-code-fast-1) — even
though `/model <id>` works fine because runtime inference reads the pool
through a different code path.
Mirror the Codex catalog resolver pattern: env-var first (unchanged), then
walk `read_credential_pool("copilot")` for the first entry with a
supported `access_token` (`gho_*` / `github_pat_*` / `ghu_*`). Run it
through `get_copilot_api_token()` so the catalog request uses the same
exchanged token the runtime path uses. Classic PATs (`ghp_*`) are still
rejected up-front via `validate_copilot_token` since the Copilot API
doesn't accept them.
Strictly additive: env still wins, and a missing/locked auth.json (or any
exception during pool read) still returns "" so the caller falls through
to the curated catalog.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
model_tools.py ran discover_mcp_tools() as a module-level side effect.
discover_mcp_tools() uses a blocking 120s wait internally (via
_run_on_mcp_loop -> future.result(timeout=120)).
The gateway lazy-imports run_agent -> model_tools on the first user
message, which happens inside the asyncio event loop thread. A slow or
unreachable MCP server therefore froze Discord shard heartbeats and
Telegram polling for up to 120s on the first message after gateway
start.
Fix: remove the module-level call. Every entry point now runs
discovery explicitly at its own startup, using the context-appropriate
blocking/non-blocking pattern:
- gateway/run.py: loop.run_in_executor(None, discover_mcp_tools)
before platforms start accepting traffic
- hermes_cli/main.py: inline (no event loop at CLI startup)
- tui_gateway/entry.py: inline (sync stdin loop, no event loop)
- acp_adapter/entry.py: inline before asyncio.run()
Closes#16856.
_handle_set_home_command wrote FEISHU_HOME_CHANNEL / DISCORD_HOME_CHANNEL /
etc. as top-level keys into config.yaml, but load_gateway_config() only
reads home channels from env vars. After every gateway restart the home
channel was lost — on every platform, not just Feishu.
Fix: switch /sethome to save_env_value(), which atomically writes to
~/.hermes/.env and updates the current process env in one shot. The
handler builds the env key from platform_name.upper(), so one line
change repairs /sethome for every platform that has a HOME_CHANNEL
env var.
Also widen _EXTRA_ENV_KEYS in hermes_cli/config.py so HOME_CHANNEL and
HOME_CHANNEL_NAME for every platform are treated as managed env vars:
SIGNAL, SLACK, SMS, DINGTALK, BLUEBUBBLES, FEISHU, WECOM, YUANBAO, plus
the missing *_NAME variants for DISCORD/TELEGRAM/MATTERMOST.
Closes#16806
Co-authored-by: teknium1 <screenmachine@gmail.com>
`/new` after `/model <custom-provider>:<model>` silently reverted to a
native provider whose static catalog happened to contain the same model
name (e.g. `deepseek-v4-pro` → native `deepseek` → 401).
Root cause at the `/model` writeback site: `HERMES_INFERENCE_PROVIDER`
was set unconditionally but `HERMES_TUI_PROVIDER` was only mirrored when
it was already set. On sessions launched without `--provider`,
`HERMES_TUI_PROVIDER` stayed unset, so `_resolve_startup_runtime()` on
`/new` skipped the explicit-provider early return and fell through to
`detect_static_provider_for_model()`.
Fix: set `HERMES_TUI_PROVIDER` unconditionally alongside
`HERMES_INFERENCE_PROVIDER` when `/model` lands. Keeps #15755's
invariant intact — `HERMES_TUI_PROVIDER` remains the canonical
"explicit this process" carrier, `HERMES_INFERENCE_PROVIDER` remains
ambient and does not short-circuit startup resolution.
Bug report and diagnosis: @Bartok9 in #16857 / #16873.
Fixes#16857
Replace the Linux/macOS pgrep regex ("hermes.*dashboard") with a ps
scan + the same explicit patterns list already used on the Windows
branch and in hermes_cli.gateway._scan_gateway_pids:
hermes dashboard
hermes_cli.main dashboard
hermes_cli/main.py dashboard
The old greedy regex would match any cmdline containing both words —
e.g. a chat session whose argv mentions "dashboard" or an unrelated
grafana/dashboard-server process. Added regression tests for both.
Follow-up tightening on #16881.
The dashboard is a long-lived server process users start and forget.
When hermes update replaces files on disk, the running process holds
the old Python backend in memory while the JS bundle gets updated,
producing a silent frontend/backend mismatch (e.g. v0.11.0 changed
the session token header -- old backends reject every API call).
Scan for running dashboard processes after a successful update (both
git and ZIP paths) and print a warning with their PIDs and restart
instructions. Mirrors the existing pattern for gateway processes.
Fixes#16872
When delegation.provider is configured (e.g. minimax-cn), subagents
inherited the parent's acp_command unconditionally. This caused
run_agent.py to initialize CopilotACPClient, which bypassed the
override credentials entirely and used its own default model
(provider=copilot-acp model=qwen3.5-397b-a17b) instead of the
configured delegation.provider and delegation.model.
Fix: when override_provider is set but override_acp_command is not,
clear effective_acp_command and effective_acp_args so the child agent
uses direct API calls with the configured provider credentials.
The existing override_acp_command path is unchanged — explicit ACP
transport overrides still force provider=copilot-acp as before.
Fixes#16816
PR #16888 swaps the opencode-zen/go resolver so that api_mode is always
re-derived from the effective model before the persisted api_mode is
consulted. That's the point of the fix — a stale anthropic_messages
from a previous minimax default must not survive a /model switch to a
chat_completions target (or vice versa) and strip /v1 from base_url.
The prior test asserted the opposite precedence — that a persisted
api_mode won over model-derived mode — and was added in #4508 to lock
in escape-hatch behavior. Under the new precedence that escape hatch
no longer exists for opencode (only for providers that genuinely
support both modes at a single endpoint — and for opencode the model
name is the unambiguous signal). Rename + invert the assertion to
document the intentional behavior change.
Refs #16878.
opencode-zen and opencode-go each serve both anthropic_messages
(e.g. minimax-m2.7) and chat_completions (e.g. deepseek-v4-flash)
models behind a single base_url. The api_mode resolver in
hermes_cli/runtime_provider.py honoured the persisted
model_cfg.api_mode (set by the previous default model) before checking
the opencode model registry, so /model deepseek-v4-flash from a session
whose default was minimax-m2.7 inherited 'anthropic_messages', stripped
'/v1' from base_url (the Anthropic SDK adds its own /v1/messages), and
404'd.
Promote the opencode detection branch above the configured_mode check
in both api_mode resolution paths:
- _resolve_runtime_from_pool_entry (pool-backed providers)
- _resolve_api_key_runtime (api-key providers, fallback path)
Both branches now call opencode_model_api_mode(provider, effective_model)
unconditionally for opencode-zen/go before considering any persisted
api_mode, so the mode always reflects the model the user just switched
to.
Existing tests pass (12/12 in tests/hermes_cli/test_model_switch_opencode_anthropic.py).
Fixes#16878
Switch _PRIORITY_PROCESSING_MODELS and _ANTHROPIC_FAST_MODE_MODELS from
hardcoded frozensets to prefix-based matching. Any gpt-*, o1*, o3*, o4*
(OpenAI) and any claude-* (Anthropic) now exposes /fast.
Fixes the case where gpt-5.5 and other post-catalog models silently
skipped Priority Processing because they weren't in the frozenset.
Future OpenAI/Anthropic releases will work without a catalog bump.
Safety:
- Codex-series (*codex*) still excluded — they route through the Codex
Responses API which doesn't take service_tier.
- Anthropic adapter already gates speed=fast on native endpoints only
(_is_third_party_anthropic_endpoint), so claude-sonnet-4.6 on
OpenRouter/Bedrock/opencode-zen won't leak the unknown beta.
- service_tier=priority is silently dropped by non-OpenAI proxies, so
false positives are harmless.
Drop the duplicate _load_openclaw_config_early() added in the salvaged
commit — load_openclaw_config() (line 979) has the identical body and
is a plain instance method that only needs self.source_root, which is
already set before __init__ needs it.
OpenClaw users who started before the rebrand (when the project was
clawd/clawdbot) often have a custom workspace directory configured via
agents.defaults.workspace in openclaw.json (e.g. ~/clawd/ instead of
~/.openclaw/workspace/).
The migration tool only checked hardcoded relative paths (workspace/,
workspace-main/, workspace-assistant/) inside the source root, so files
like MEMORY.md, skills, and daily memory in custom workspaces were
silently skipped.
This change:
- Reads agents.defaults.workspace from openclaw.json at init time
- Uses it as a final fallback in source_candidate() when files aren't
found in the standard locations
- Standard workspace paths are still preferred (custom is fallback only)
- Custom workspace is only used when it's outside the source_root tree
(avoids double-matching when workspace/ is the default)
Adds two tests:
- Custom workspace files are discovered and migrated
- Standard workspace location is preferred over custom
Flips security.redact_secrets from true to false in DEFAULT_CONFIG, and
the HERMES_REDACT_SECRETS env-var fallback in agent/redact.py now
requires explicit opt-in ("1"/"true"/"yes"/"on") to enable.
New installs and users without a security.redact_secrets key get pass-
through tool output. Existing users whose config.yaml explicitly sets
redact_secrets: true keep redaction on — the config-yaml -> env-var
bridges in hermes_cli/main.py and gateway/run.py still honor their
setting.
Also updates the inline config comments, website docs, and the
hermes-agent skill so /hermes config set security.redact_secrets true
is now the documented way to turn it on.
MatrixAdapter._is_self_sender returns True defensively when _user_id is empty
(whoami not yet resolved) to prevent echo loops — see #15763. The reaction
approval test must therefore initialize a user_id so _on_reaction does not
drop the inbound test event before reaching the approval handler.
Self-contained docker-compose harness that exercises the new bootstrap
branch against a real Continuwuity homeserver. Three tests:
1. fresh bot → bootstrap fires, /keys/query returns master + ssk
with UNPADDED base64 keyids, current device is signed by the
new SSK
2. second startup with same crypto store → bootstrap is skipped
3. MATRIX_RECOVERY_KEY set → existing verify_with_recovery_key path
takes precedence, no new bootstrap
Run via:
docker compose -f tests/e2e/matrix_xsign_bootstrap/docker-compose.yml up -d
python tests/e2e/matrix_xsign_bootstrap/test_bootstrap.py
docker compose -f tests/e2e/matrix_xsign_bootstrap/docker-compose.yml down -v
The test mirrors the bootstrap snippet from matrix.py inline so it can
run without importing the full hermes gateway and its deps. Skipped
automatically when mautrix isn't installed or the homeserver is
unreachable.
All three pass against ghcr.io/continuwuity/continuwuity:latest
(Continuwuity 0.5.7). The unpadded-keyid assertion is the load-bearing
one — it's exactly the property the PR's bootstrap path provides that
the hand-rolled `base64.b64encode().decode()` scripts get wrong.
Without this, every Matrix bot started under hermes-agent shows the
"Encrypted by a device not verified by its owner" badge in Element
indefinitely, because the cross-signing chain (master → SSK → device)
was never published. Operators currently have to write their own
bootstrap script and remember to run it once per bot — and it's easy
to get wrong (the obvious base64.b64encode().decode() produces padded
keyids that matrix-rust-sdk silently rejects in /keys/query, so even
correctly-signed keys fail to load identity in Element).
mautrix already has the right primitive: generate_recovery_key() does
the full flow — generate seeds, upload privates to SSSS, publish
publics to the homeserver, sign the current device with the new SSK,
and return the human-readable recovery key. We invoke it once on
startup if the bot has no existing cross-signing identity, and log
the recovery key with a clear instruction to save it for future
restarts via MATRIX_RECOVERY_KEY (which the existing recovery-key
path already consumes).
Skipped when MATRIX_RECOVERY_KEY is set (existing path takes over)
or when the bot already has cross-signing keys on the homeserver
(get_own_cross_signing_public_keys returns non-None).
Bootstrap failure is non-fatal — logged with hint about UIA; the bot
continues without cross-signing and Element will show the warning
that prompted this PR. That matches the existing soft-fail pattern
for verify_with_recovery_key.
Tested against Continuwuity 0.5.7 (no UIA required). Synapse with
UIA enabled will need a follow-up PR to thread MATRIX_PASSWORD
through to /keys/device_signing/upload.
Five ``except Exception as exc:`` blocks in the Matrix adapter logged
only ``str(exc)`` without ``exc_info=True``:
- _reverify_keys_after_upload → post-upload key verification failure
- _upload_keys_if_needed → initial device-key query failure
- _upload_keys_if_needed → re-upload device keys failure
- _upload_keys_if_needed → initial device key upload failure
- connect → whoami / access-token validation failure
The E2EE key paths here are security-critical: a silent traceback-
less failure during device-key verification or upload makes it
hard for operators to tell whether their Matrix bot is failing
because of a stale token, a federation timeout, or an olm state
mismatch — all three fail with different tracebacks, which
``str(exc)`` alone flattens.
The contributing guide asks for ``exc_info=True`` on error logs.
Append it to each of the five call sites. Pure logging enrichment.
- Wrap _sync_loop sync() call with asyncio.wait_for(timeout=45s) to guard
against TCP-level hangs that the Matrix long-poll timeout cannot catch
- Add logger.debug at the top of _on_room_message so LOG_LEVEL=DEBUG
confirms whether callbacks fire at all (diagnoses #5819, #7914, #12614)
- Add logger.debug when MATRIX_REQUIRE_MENTION silently drops a message,
pointing users to the env var to disable the filter
Adapted for current mautrix-python adapter (PR was written against the
legacy matrix-nio adapter).
Closes#5819
* Port from Kilo-Org/kilocode#9448: roll up subagent costs into parent session total
Child subagents built by delegate_task() each track their own
session_estimated_cost_usd, but the parent agent's total never folded
those numbers in. On runs where the parent mostly delegates and the
children do the expensive work, the footer/UI was reporting a fraction
of the actual spend — sometimes $0.00 when the parent itself made no
billed calls.
Fix:
- Capture each child's session_estimated_cost_usd into _child_cost_usd
on the result entry (before child.close() drops the counter).
- After the existing subagent_stop hook loop, sum the children's costs
and add the total to parent.session_estimated_cost_usd.
- Promote session_cost_source from 'none' -> 'subagent' when the parent
had no direct spend but children did, so the UI doesn't label the
total as having unknown provenance. Real sources (openrouter,
anthropic, etc.) are preserved.
Nested orchestrator -> worker trees roll up naturally: each layer's own
delegate_task() folds its direct children in, and when the orchestrator
itself returns, its parent folds the orchestrator's now-inflated total
on top.
Internal fields (_child_cost_usd, _child_role) are stripped from the
results dict before it's serialised back to the model — same contract
as _child_role already followed.
Tests: TestSubagentCostRollup (5 cases) covers single-child, batch,
zero-cost-children, preserved-source, and legacy-fixture paths.
Source: https://github.com/Kilo-Org/kilocode/pull/9448
* fix(web): scope dashboard config Reset button to the current tab
Reported by @ykmfb001 via X: clicking 'Restore Defaults' (恢复默认值) on
the Auxiliary page wiped the entire config.yaml to defaults, not just
the auxiliary section. The button sits next to the category tabs and
users reasonably assumed 'reset this tab', not 'reset everything'.
Changes:
- handleReset now scopes to the fields in the current view:
active category's fields (form mode) or search-matched fields
(search mode). Only those keys are copied from defaults; the rest
of the config is left alone.
- Added a window.confirm() with the scope name before applying.
- Button is hidden in YAML mode (scoping doesn't apply there).
- Tooltip/aria-label now name the scope, e.g. 'Reset Auxiliary to
defaults'.
- i18n: new resetScopeTooltip / confirmResetScope / resetScopeToast
strings in en + zh; resetDefaults key preserved for compat.
On AWS Bedrock (and Azure AI Foundry), Claude Opus 4.6/4.7 and Sonnet 4.6
are capped at 200K context unless the request carries the
`context-1m-2025-08-07` beta header. On native Anthropic (api.anthropic.com)
1M went GA so the header is a harmless no-op, but Bedrock/Azure still gate
it as beta as of 2026-04.
Hermes was advertising 1M in model_metadata.py (`claude-opus-4-7: 1000000`)
while silently sending a request without the beta — so Bedrock users saw
a 200K ceiling with no error message, and no config knob unblocked it.
Claude Code sends this header by default, which is why the same Bedrock
credentials worked there.
- Add `context-1m-2025-08-07` to `_COMMON_BETAS` (alongside interleaved
thinking and fine-grained tool streaming).
- Strip it in `_common_betas_for_base_url` for MiniMax bearer-auth
endpoints — they host their own models, not Claude, so Anthropic beta
headers are irrelevant and could risk rejection.
- Attach `_COMMON_BETAS` as `default_headers` on the AnthropicBedrock
client. Previously that constructor passed no betas at all, so native
Anthropic had the 1M unlock via default_headers but Bedrock didn't.
- Fast-mode per-request `extra_headers` already rebuilds from
`_common_betas_for_base_url`, so it picks up the 1M beta automatically.
Reported by user 'Rodmar' on Discord: Bedrock Opus 4.7 stuck at 200K while
same credentials worked in Claude Code.
Anyone who ran hermes between Apr 15 (42aeb4ec) and Apr 22 (a7d78d3b)
has schema_version=7 from the pre-renumber api_call_count migration.
When a7d78d3b inserted reasoning_content as the new v7 and pushed
api_call_count to v8, the 'if current_version < 7' gate was already
false for those users, so reasoning_content was never created —
sqlite3.OperationalError: no such column: reasoning_content on any
/continue or /resume touching assistant replays.
Replaces the version-gated ADD COLUMN chain with _reconcile_columns():
on every startup, parse SCHEMA_SQL via an in-memory SQLite and diff
against PRAGMA table_info; ALTER TABLE ADD COLUMN for anything missing.
Follows the Beets / sqlite-utils pattern — SCHEMA_SQL becomes the single
source of truth for declared columns. Self-healing and idempotent.
v10 trigram FTS backfill is retained in a version-gated block — that
migration isn't a column add, it inserts existing message rows into
the new FTS virtual table, so reconciliation can't express it.
schema_version is also kept for future row-data migrations.
Salvaged from #14097 (@kshitijk4poor) onto current main; v10 trigram
preservation and the v9 codex_message_items column (stale-missed by
the original branch) are covered automatically by reconciliation.
Tests:
- Regression: DB at old v7 with api_call_count but no reasoning_content
gets the column on open
- Idempotency: reopening the same DB is a no-op
- Structural invariant: every SCHEMA_SQL column is in the live DB
- Existing v2 migration test still passes
- E2E verified against fresh / v1 / old-v7 / v9 DBs, plus v10 trigram
backfill preserved
Port https://github.com/blader/humanizer (MIT, v2.5.1, 16k stars) into
the built-in skills under skills/creative/humanizer/. Based on Wikipedia's
'Signs of AI writing' guide (WikiProject AI Cleanup) — detects 29 AI-writing
patterns and rewrites them to sound human.
Hermes-native adaptations:
- Description (<60 chars) explains what it's for: 'Humanize text: strip
AI-isms and add real voice.'
- 'When to use this skill' section — trigger phrases (humanize, de-AI,
de-slop, un-ChatGPT, rewrite to not sound like an LLM) plus guidance to
apply it to the agent's own output (release notes, PR descriptions, docs).
- 'How to use it in Hermes' — maps the three real input paths (inline,
file via read_file/patch/write_file, voice-calibration sample) onto the
tools the agent actually has. Drops Claude Code's allowed-tools block.
- Converted frontmatter to Hermes format (metadata.hermes.tags, category,
homepage, related_skills).
Attribution preserved:
- Original author Siqi Chen (@blader) credited in frontmatter and body.
- Full MIT LICENSE copied verbatim alongside SKILL.md.
- Wikipedia / WikiProject AI Cleanup credited.
- 29 patterns, personality/soul section, and full worked example kept
verbatim from the source (29,914 chars).
Validated end-to-end against a clean HERMES_HOME:
- sync_skills() copies skills/creative/humanizer/ including LICENSE.
- skills_list(category='creative') returns the 48-char description.
- skill_view(name='humanizer') returns the full body with all 29 patterns,
personality/soul, attribution, and Hermes tool refs (read_file, patch,
write_file) intact.
Plugins can now observe dangerous-command approval events in real time,
on both the CLI-interactive path and the async gateway path. This is the
missing hook surface external tools need to build approval notifiers
(macOS menu-bar allow/deny, Slack alerts, audit logs, etc.) without
forking Hermes or running a parallel gateway adapter.
Changes:
- hermes_cli/plugins.py: add two entries to VALID_HOOKS
- tools/approval.py: fire both hooks from check_all_command_guards --
around prompt_dangerous_approval (CLI surface) and around the
notify_cb + blocking event.wait loop (gateway surface)
- website/docs/user-guide/features/hooks.md: document both hooks with
a macOS-notification example
- tests/tools/test_approval_plugin_hooks.py: 5 tests covering CLI once,
CLI deny, plugin-crash resilience, gateway approve, gateway timeout
Hooks are observer-only: return values are ignored, so plugins cannot
veto or pre-answer an approval (use pre_tool_call for that). A crashing
plugin cannot break the approval flow -- invoke_hook swallows per-
callback errors, and the wrapper logs and swallows dispatch-layer
errors too.
Surface kwarg distinguishes "cli" from "gateway"; post hook reports
choice as one of once/session/always/deny/timeout.
A misconfigured auxiliary.compression.model is a user-fixable problem that silent recovery would hide. The previous retry-on-main logic transparently swallowed aux-model failures whenever the fallback succeeded, leaving the user's broken config in place and racking up future failures.
Track the aux-model failure on the compressor alongside the existing fallback-placeholder fields:
- _last_aux_model_failure_model: str | None
- _last_aux_model_failure_error: str | None
Both are set at the moment the aux model errors (captured before summary_model is cleared for retry), regardless of whether the retry succeeds. Cleared at compress() start and on on_session_reset() so a clean run doesn't leak stale warnings.
Surface at three places:
- gateway hygiene auto-compress: ℹ note to the platform adapter (thread_id preserved)
- gateway /compress command: ℹ line appended to the reply
- CLI via _emit_warning: deduped on (model, error) so repeat compactions don't spam
Distinct from the existing ⚠️ dropped-turns warning — different severity, different emoji, explicit 'context is intact' reassurance.
Adds four new reference docs covering common TD use cases not previously
documented in the skill:
- animation.md: LFOs, timers, keyframes, easing, time references
- midi-osc.md: MIDI controllers, OSC routing, TouchOSC, multi-machine sync
- particles.md: POPs and particleSOP — emission, forces, collisions, render
- projection-mapping.md: windowCOMP, corner pin, mesh warp, edge blending
Also clarifies the SKILL.md tool quick reference: adds td_screen_point_to_global
and notes that 4 admin/dev-mode tools (td_project_quit, td_test_session,
td_dev_log, td_clear_dev_log) live only in mcp-tools.md to keep the main
reference focused on creative workflows.
No SKILL.md workflow or critical-rules changes. References load on demand
so no token-budget impact at session start.
The existing retry-on-main path in _generate_summary only fires for errors that match the _is_model_not_found heuristic (404/503, 'model_not_found', 'does not exist', 'no available channel'). Other misconfiguration errors — 400s from aggregators, provider-specific 'no route' strings, opaque rejections — fall straight through to the transient-cooldown branch, which drops N turns of context and inserts a static placeholder.
Losing context is almost always worse than one extra summary attempt. Add a best-effort retry-on-main for the unknown-error branch, guarded by the same invariants as the existing fast-path retry: only when summary_model differs from main, and only once per compressor (_summary_model_fallen_back).
Tests cover: 404 fast-path fallback still works, unknown 400 now falls back, same-model aux skips retry (no infinite loop), and a double-failure (aux + main) stops at 2 calls.
PR #16333 added a warning to the manual /compress reply when the
auxiliary summariser fails and the static fallback placeholder is
used, but only the gateway-hygiene path had a test
(test_session_hygiene_warns_user_when_summary_generation_fails).
The /compress branch in _handle_compress_command was uncovered.
New test test_compress_command_appends_warning_when_summary_generation_fails
mocks the compressor's _last_summary_fallback_used /
_last_summary_dropped_count / _last_summary_error fields and
verifies the /compress reply contains the ⚠️ marker, the underlying
error string, the dropped message count, and the 'historical
message(s) were removed' wording — i.e. the same contract the
hygiene-path test enforces.
The per-call reset block at the top of compress() cleared
_last_summary_dropped_count and _last_summary_fallback_used but
not _last_summary_error. Functionally this didn't break the
gateway warning path (callers gate on _last_summary_fallback_used
first, and _last_summary_error is overwritten on the next failure),
but it left the three tracking fields inconsistent — anyone
reading _last_summary_error standalone after a successful compress
would see a stale value from a previous failed compress.
Reset all three together so the per-call contract is uniform.
The fallback placeholder said "N conversation turns were removed" while the
gateway warning said "N historical message(s) were removed". Use "messages"
in both so users don't wonder if the two counters refer to different things.
Address review feedback on PR #16333:
1. The hygiene-path warning send was missing metadata=_hyg_meta. On
Telegram topics / Slack threads / Discord threads the warning would
land in the main channel instead of the originating thread. Now
reuses the same _hyg_meta dict already computed for the hygiene
compaction itself.
2. New gateway-level test
test_session_hygiene_warns_user_when_summary_generation_fails
verifies end-to-end:
- When the compressor's _last_summary_fallback_used flag is True,
the gateway invokes adapter.send() exactly once.
- The warning message includes the dropped count and the underlying
error string.
- metadata={'thread_id': ...} is propagated so the warning lands
in the originating topic/thread.
Tests: 20 gateway hygiene + 54 context_compressor — all pass.
When auxiliary compression's summary LLM call fails (e.g. model 404,
auxiliary model misconfigured), the compressor still drops the selected
turns and inserts a static fallback placeholder — the dropped context
is unrecoverable.
Previously the only signal of this was a WARNING in agent.log. Gateway
users (Telegram/Discord/etc.) had no way to know context was lost
because the existing _emit_warning path requires a status_callback,
and the gateway hygiene path uses a temporary _hyg_agent with
quiet_mode=True and no callback wired up.
Changes:
- ContextCompressor: track _last_summary_fallback_used and
_last_summary_dropped_count on each compress() call. Cleared at the
start of compress() and on session reset.
- gateway/run.py hygiene: after auto-compress, inspect the temp
agent's compressor; if fallback was used, send a visible ⚠️ warning
to the user via the platform adapter (TG/Discord/etc.) including
dropped count and the underlying error.
- gateway/run.py /compress: append the same warning to the manual
compress reply so users running /compress see the failure too.
Acceptance:
- Summary success: no user-visible warning (unchanged).
- Summary failure on gateway hygiene: user receives a TG/Discord
message with dropped count + error + remediation hint.
- Summary failure on /compress: warning appended to the command reply.
- CLI status_callback / _emit_warning path is untouched.
- Test coverage: two new tests verify the tracking fields are set on
failure and cleared on subsequent success.
The typing-indicator refresh loop in BasePlatformAdapter._keep_typing
awaited each send_typing call unconditionally. Each call is an HTTP
round-trip to the platform API (Telegram/Discord), normally ~100ms. When
the same network instability that causes upstream provider timeouts
(e.g. Anthropic capacity blips slowing first-token latency past the
120s stream-read timeout) also slows the platform typing API to
multi-second response times, the refresh loop stalls inside the await.
Platform-side typing expires at ~5s, so the bubble dies and stays dead
until the stuck send_typing call returns — right when the user most
needs the 'still working' signal and instead sees a bot that looks
dead, then asks 'wtf are you doing' which itself interrupts the
eventually-recovering turn.
Bound each send_typing with asyncio.wait_for (1.5s cap, derived from
interval so it's always below the 2s cadence). Slow calls get abandoned
so the next scheduled tick fires a fresh send_typing on schedule. As
long as any one of them reaches the platform within its ~5s
typing-expiry window, the bubble stays visible across the stall.
Also catches non-timeout send_typing exceptions (transient HTTP errors)
so one bad tick doesn't terminate the whole loop.
Tests: 4 new in tests/gateway/test_keep_typing_timeout.py covering
slow-send non-blocking, fast-send still-awaited, exception resilience,
and paused-chat regression guard.
- moveCursor(extend=true) now collapses to the bare cursor when the
computed offset equals the existing anchor instead of leaving a
zero-length sel. Without this, Shift+Left at col 0 / Shift+Home at
start would silently hide the hardware cursor (selected truthy)
without rendering any highlight.
- _tui_need_npm_install also catches UnicodeDecodeError so a corrupted
/ non-UTF8 lockfile falls back to the mtime path the docstring
promises instead of crashing.
Made-with: Cursor
* feat(tui): auto copy-on-select for transcript text
Drag in the transcript already highlighted but you had to press Cmd+C to
land it on the clipboard, and the highlight cleared on copy — most users
never realised selection existed. Now drag-release fires copySelectionNoClear
so the text is on the clipboard immediately while the highlight stays put,
matching iTerm2's "Copy to pasteboard on selection" default. Esc clears.
Behaviour:
- Single click in the input still positions the cursor (TextInput onClick).
- Single click in the transcript still does nothing destructive.
- Double / triple click select word / line, then drag extends.
- /copyselect [on|off|toggle] (alias /cos) flips the setting at runtime,
HERMES_TUI_DISABLE_COPY_ON_SELECT=1 disables at startup, persists via
display.tui_copy_on_select in config.yaml.
Help overlay now lists drag-select, multi-click, and click-to-position
so the gestures are discoverable.
Made-with: Cursor
* fix(tui): support prompt text selection gestures
Add mouse drag selection and Shift+Arrow/Home/End extension inside the TUI composer so prompt text behaves like a normal editable field while keeping click-to-position and right-click paste intact.
Made-with: Cursor
* Revert "feat(tui): auto copy-on-select for transcript text"
This reverts commit 6701288fe0.
* fix(tui): allow composer selection from prompt whitespace
Give the composer a one-cell mouse capture pad before the editable text. The prompt glyph/gutter still does not become selectable, but dragging from the edge now anchors at input offset 0 so users do not need to hit the first character precisely.
Made-with: Cursor
* fix(tui): clear selections from blank composer space
Clicking blank space in the transcript or composer now clears active TUI/input selections like a normal text surface. TextInput clicks stop bubbling so cursor placement and selection gestures keep their local behavior.
Made-with: Cursor
* fix(tui): delegate prompt gutter drags to composer text
The prompt gutter is now an input gesture region, not selectable content. Dragging from the whitespace or prompt area anchors the composer selection at offset 0, while selection highlight/copy remains limited to actual input text.
Made-with: Cursor
* fix(tui): move composer cursor to end on selection clear
External clear actions now collapse the composer selection to the end of the input, matching normal text-field behavior after dismissing a selection.
Made-with: Cursor
* fix(tui): capture composer padding before prompt
Add an explicit mouse capture cell over the left padding before the prompt glyph. Drags starting there now delegate to the composer input at offset 0 instead of starting terminal-level selection over the prompt chrome.
Made-with: Cursor
* fix(tui): avoid npm install on lockfile mtime churn
Compare package-lock.json against npm's hidden node_modules lock by content instead of mtimes. Git checkouts and npm lock rewrites can make the root lockfile newer even when installed dependencies already match, causing hermes --tui to print Installing TUI dependencies on every launch.
Made-with: Cursor
* fix(tui): include prompt leading cell in gesture region
Use the prompt box's real layout region to cover the leading whitespace cell before the glyph. The cell now participates in mouse hit testing and delegates to composer selection instead of starting terminal-level selection.
Made-with: Cursor
* fix(tui): widen prompt-side gesture capture band
Capture a wider left-side band around the composer prompt row so drags starting in terminal gutter/padding cells are consumed and delegated to input selection, instead of triggering terminal-level selection chrome.
Made-with: Cursor
* fix(tui): make pre-prompt spacer non-selectable content
Replace the sticky-prompt fallback `Text(' ')` with an empty spacer box so the visual gap remains but no literal space character is rendered/copyable before the composer prompt.
Made-with: Cursor
* fix(tui): capture pre-prompt spacer without shifting prompt layout
Revert the widened negative-margin prompt capture band and instead capture drags on the dedicated spacer row above the prompt. This keeps prompt/text alignment stable while still delegating whitespace-start drags to composer selection.
Made-with: Cursor
* fix(tui): align prompt with status bar and capture full input row
Drop the leading prompt column from 3 to 2 so the input first character lines up with the status bar text. Wrap the prompt+input row in a single mouse-capture box and stop event propagation from TextInput's own handlers so any drag in that row delegates to composer selection without leaking to terminal-level selection.
Made-with: Cursor
* fix(tui): anchor hardware cursor during composer selection
When a composer selection covers a row exactly the column width, the rendered text fills the row and the terminal auto-wraps the hardware cursor to col 0 of the next row, leaving a ghost block beneath the prompt. Park the cursor at the start of the input box during selection so it can't escape the input region.
Made-with: Cursor
* fix(tui): hide hardware cursor during composer selection
Stop fighting auto-wrap by hiding the hardware cursor outright while the
composer has an active selection. This prevents both the ghost block under
the prompt (cursor wrapping past the last cell) and the parked-cursor block
on the first selected character. The cursor restores as soon as the
selection clears or focus changes.
Made-with: Cursor
* chore(tui): /clean — drop dead capture-pad path, dedupe gutter handlers
- TextInput: remove unused leftCaptureColumns prop and capture-pad math, drop
unused mouseApi.startAt, fold mouse offset into a single offsetAt helper,
share a MouseEventLite type across the four handlers.
- appLayout: hoist a GutterMouseEvent type and an endInputDrag callback so the
spacer/prompt/input rows share one shape.
- _tui_need_npm_install: lift the runtime-only key set to a module constant,
collapse nested isinstance checks, and document the mtime fallback.
Made-with: Cursor
* fix(tui): address copilot review on PR #16732
- Split InputSelection.clear() into clear() (cursor-preserving) and
collapseToEnd() (clear + jump to end). Cmd+C copy paths keep using
clear() so the cursor stays put; the blank-area click in useMainApp
switches to collapseToEnd() to match the requested UX.
- Spacer-row drags now force row=0 when forwarding into the input,
since the spacer's vertical origin doesn't align with the input box
and Ink mouse-capture keeps dispatching motion to the original
target. Prompt+input row drag keeps localRow because origins match.
Made-with: Cursor
* fix(tui): give TextInput Box an explicit width
After the /clean pass dropped the unused capture-pad math, the wrapping
Box also lost its explicit width and started sizing to its rendered
content. Clicks past the last character missed TextInput and fell
through to the parent prompt-row Box, which collapsed the cursor to
offset 0. Pin the Box back to `columns` so the input owns its full
column span regardless of value length.
Made-with: Cursor
* feat(tui): double-click select-all + hide cursor on terminal blur
- Track click time/offset in TextInput so a quick second click on the
same offset triggers select-all. Ink's screen-level multi-click is
bypassed once our onMouseDown captures, so the gesture has to be
detected locally.
- Extend the cursor-hide effect to also fire when the terminal loses
focus, so the hollow-rect ghost most terminals draw at the parked
cursor position disappears too.
Made-with: Cursor
* chore(tui): /clean — extract isMultiClickAt helper
Pull the click-recurrence math out of TextInput's onMouseDown into a
small isMultiClickAt(offset) helper so the handler reads as the gesture
list it actually is (multi-click → select-all, otherwise start).
Drop the redundant length>0 guard now that selectAll() already noops on
an empty value.
Made-with: Cursor
* docs(tui): explain _tui_need_npm_install content-vs-mtime comparison
Expand the docstring so future readers understand why we parse the
lockfiles instead of comparing mtimes, what the optional/peer skip
covers, how stale hidden-lock entries are handled, and when we fall
back to mtime.
The previous commit on this branch went through a layer that redacted
strings matching API-key patterns. Restore the original placeholder
values (sk-ant-..., ${ANTHROPIC_API_KEY}, etc.) that were already in
main so the diff is scoped strictly to the new Multi-profile support
section.
Clarifies that Hermes' built-in multi-profile feature is not recommended
when running under Docker. Recommends instead running one container per
profile, each bind-mounting its own host data directory as /opt/data.
Includes docker run examples, a rationale list (isolation, independent
lifecycle, port separation, concurrent-write safety), and a Compose
snippet showing two profile services side by side.
- Rename `removeAt` → `removeAtInPlace` and document the mutation
contract; the old name read like a non-mutating helper.
- Hotkey table + queue header: use `Ctrl+X` / `Esc` to match the
rest of the UI (was `⌃X` / `esc`).
- Render the queued header as a single template literal so JSX
text-node whitespace can't sneak into the rendered line.
- Make `Esc` while editing beat the `terminal.hasSelection` clear:
the header promises 'Esc cancel', so an active selection
shouldn't silently consume the keystroke.
The text input's ctrl-passthrough whitelist only listed Ctrl+C and
Ctrl+B. Ctrl+X fell through to the printable-char branch and got
inserted as 'x' alongside the queue-delete action firing in
useInputHandlers.
Add Ctrl+X to the same whitelist so it bypasses the readline-style
fallback and reaches the app-level handler unchanged. When not in
queue-edit mode it's a no-op, which is fine — typing 'x' on Ctrl+X
was the wrong default anyway.
Today there's no way to remove a queued message — ↑ loads it for edit,
ctrl-K dispatches the head, but a draft you no longer want stays put
forever. ctrl-C just clears the composer and exits edit mode without
touching the queue.
Two new bindings, both gated on queueEditIdx !== null so they're
inert when the user isn't pointing at a queue item:
- ctrl-X — delete the queue item being edited, clear composer, exit
edit mode. "cut" matches the mental model and doesn't collide with
any existing binding.
- esc — cancel the edit (composer clears, item stays in queue).
Mirrors ctrl-C's existing behavior so muscle memory has two paths.
Header line now reads `queued (3) · editing 2 · ⌃X delete · esc cancel`
when in edit mode, so the affordance is discoverable without /help.
The /help hotkey table also gets a Ctrl+X entry.
ctrl-C is intentionally unchanged: it should never destroy queued
content. Cancel is non-destructive (esc / ctrl-C); only ctrl-X
removes the item.
Same layering concern as the persisted-assistant scrub already removed:
_emit_interim_assistant_message and the final_response return path were
mutating model output broadly. Streaming scrubber covers real leaks
delta-by-delta; these post-stream scrubs were redundant.
Reviewer pushback on the original boundary-hardening commits — three
overreach points pulled plugin-specific policy into shared core paths:
1. gateway/run.py hardcoded a '## Honcho Context' literal split for
vision-LLM output. Plugin-format heading in framework code; could
truncate legitimate output naturally containing that header.
Drop the literal split; keep generic sanitize_context (the wrapper
strip is plugin-agnostic). Plugin-specific cleanup belongs at the
provider boundary, not the shared gateway path.
2. run_agent.run_conversation scrubbed user_message and
persist_user_message before the conversation loop. User text is
sacred — if a user types a literal <memory-context> tag we must
not silently delete it. The producer (build_memory_context_block)
is the only legitimate emitter; user input should never need the
reverse op.
3. _build_assistant_message scrubbed model output before persistence.
Same hazard: would silently mutate legitimate documentation/code
the model emits containing the literal markers. The streaming
scrubber catches real leaks delta-by-delta before content is
concatenated; persist-time scrub was redundant belt-and-suspenders.
4. _fire_stream_delta stripped leading newlines from every delta unless
a paragraph break flag was set. Mid-stream '\n' is legitimate
markdown — lists, code fences, paragraph breaks — and chunk
boundaries are arbitrary. Narrow lstrip to the very first delta
of the stream only (so stale provider preamble still gets cleaned
on turn start, but mid-stream formatting survives).
Plus: build_memory_context_block now logs a warning when its defensive
sanitize_context strips something — surfaces buggy providers returning
pre-wrapped text instead of silently double-fencing.
Net architectural change: scrub surface collapses from 8 sites to 3
(StreamingContextScrubber on output deltas, plugin→backend send,
build_memory_context_block input-validation). Plugin-specific strings
stay out of shared runtime paths. User input and persisted assistant
output are no longer mutated.
Tests: rescoped TestMemoryContextSanitization (helper-correctness only,
no source-inspection of removed call sites), updated vision tests to
drop '## Honcho Context' literal-split assertions, updated
_build_assistant_message persistence test to assert preservation.
Added: cross-turn scrubber reset, build_memory_context_block warn-on-
violation, mid-stream newline preservation (plain + code fence).
Closed PR #5137 addressed the retrieval path (peer cards via get_card()
instead of the session-scoped lookup that returned empty for per-session
messaging flows) — that architectural fix is already in main as
_fetch_peer_card / _fetch_peer_context.
What never got fixed is the user-visible side: honcho_profile returning
a flat 'No profile facts available yet.' leaves the model to guess at
why. The model then often surfaces it to the user as a cryptic error.
Adds a diagnostic hint next to the existing 'result' message, enumerating
the likely causes in rough order of frequency:
1. Observation disabled for this peer (user_observe_me/others off)
2. Peer card hasn't accumulated yet (fresh peer / dialectic cadence
hasn't fired enough turns — cards build over time)
3. Generic fallback: self-hosted Honcho < 3.x lacks peer cards
The hint also suggests alternative tools (honcho_reasoning / honcho_search)
so the model can route around the empty card rather than giving up.
Schema description updated so the model knows the hint field exists and
that an empty card is NOT an error state.
7 tests cover the hint paths: warmup, observation-disabled for user + ai,
generic fallback, populated card still returns plain result (no hint),
alternative-tool suggestion present.
The scheme-validation commit (e77a3f2c) was too strict: a user with
legacy ''baseUrl: localhost:8000'' (no ''http://'' prefix) in their
''~/.honcho/config.json'' would get ''No API key configured'' from the
CLI after that change, even though their setup worked before.
urlparse on a schemeless host:port treats the host segment as the
scheme and leaves netloc empty, so the http/https check rejected it.
Falls back to a lenient check for schemeless strings that look like
hosts: contain '.' or ':', aren't a boolean/null literal, aren't pure
digits. The SDK still rejects truly malformed URLs at connect time
with a clearer error than ours.
Three new tests: legacy schemeless hosts accepted; obvious garbage
literals (''true'', ''null'', ''12345'') still rejected. Reviewer
noted concern #1: schemeless regression for self-hosters with old
configs.
main's 6a957a74 added an optional 'metadata' kwarg to
MemoryProvider.on_memory_write so providers can distinguish tool-driven
memory writes from background-review writes. MemoryManager already
does a getfullargspec-based introspection, so the old 3-arg signature
didn't break at runtime — but it missed the origin hint entirely.
Updates HonchoMemoryProvider.on_memory_write to accept the kwarg. The
metadata isn't yet threaded into Honcho's create_conclusion payload —
that's worth its own PR once the consolidation lands and the new
metadata shape stabilises.
Two small follow-ups to the PR review:
- Hoist hashlib import from _enforce_session_id_limit() to module top.
stdlib imports are free after first cache, but keeping all imports at
module top matches the rest of the codebase.
- _resolve_api_key now URL-parses baseUrl and requires http/https +
non-empty netloc before returning the 'local' sentinel. A typo like
baseUrl: 'true' (or bare 'localhost') no longer silently passes the
credential guard; the CLI correctly reports 'not configured'.
Three new tests cover the new validation (garbage strings, non-http
schemes, valid https).
fixes#5719
The auxiliary vision LLM called by gateway._enrich_message_with_vision
can echo its injected Honcho system prompt back into the image
description. That description gets embedded verbatim into the enriched
user message, so recalled memory (personal facts, dialectic output)
surfaces into a user-visible bubble.
Strips both forms of leak before embedding:
- <memory-context>...</memory-context> fenced blocks (sanitize_context)
- trailing '## Honcho Context' sections (header + everything after)
Plus regression tests:
- tests/agent/test_streaming_context_scrubber.py — 13 tests on the
stateful scrubber (whole block, split tags, false-positive partial
tags, unterminated span, reset, case-insensitivity)
- tests/run_agent/test_run_agent_codex_responses.py — 2 new tests on
_fire_stream_delta covering the realistic 7-chunk leak scenario and
the cross-turn scrubber reset
- tests/gateway/test_vision_memory_leak.py — 4 tests covering the
vision auto-analysis boundary (clean pass-through, '## Honcho Context'
header, fenced block, both patterns together)
sanitize_context() uses a non-greedy block regex that needs both
<memory-context> open and close tags present in a single string. When a
provider streams the fenced memory block across multiple deltas (typical
for recalled-context leaks — the payload often arrives in 10+ 1-80 char
chunks), the per-delta sanitize stripped the lone open/close tags via
_FENCE_TAG_RE but let the payload in between flow straight to the UI.
Adds StreamingContextScrubber: a small stateful scrubber that tracks
open/close tag pairs across deltas, holds back partial-tag tails at
chunk boundaries, and discards span contents wholesale (including the
system-note line that fragments across deltas).
Wired into _fire_stream_delta; reset per user turn; benign trailing
partial-tag tails are flushed at the end of each model call. Mid-span
interruption (provider drops closing tag) drops the orphaned content
rather than leaking it — truncated answer > leaked memory.
Follow-up to #13672 (@dontcallmejames).
new_session() was popping the old cached session, releasing the lock,
calling get_or_create, then re-acquiring the lock to insert. A concurrent
caller could observe the empty-cache window and race-create its own
session, producing two divergent session objects for the same key.
_cache_lock is an RLock, so nested reacquisition inside get_or_create is
safe. Hold it across the whole pop/create/insert sequence.
Follow-up to #13510 (@hekaru-agent).
When no explicit timeout is configured (HonchoClientConfig.timeout,
honcho.timeout / requestTimeout, or HONCHO_TIMEOUT), get_honcho_client
previously constructed the SDK with no timeout kwarg, letting the
underlying httpx client hang indefinitely if the Honcho backend
became unreachable mid-request.
This is a silent-failure hazard on the post-response path of
run_conversation: the memory_manager.sync_all() / queue_prefetch_all()
calls fire after the agent has already generated its final reply, so
a stalled Honcho request blocks run_conversation from returning.
The gateway never logs "response ready" and never delivers the
response to the platform (Telegram, etc.), even though the text is
already saved to the session file.
Repro: unplug the network or block app.honcho.dev mid-turn after
the model has produced its final message. Without this change,
_run_agent never returns. With it, the call aborts after 30s,
run_conversation returns, and the gateway delivers the response
(Honcho sync failure is logged and swallowed as before).
The default applies only when nothing is configured, so any
deployment that has explicitly set timeout / HONCHO_TIMEOUT /
honcho.timeout / honcho.requestTimeout keeps its existing value.
Self-hosted deployments that genuinely need a longer ceiling can
still override via any of those knobs.
_resolve_api_key() only checks for apiKey / HONCHO_API_KEY, so all
CLI subcommands (identity --show, status, migrate, etc.) bail with
"No API key configured" on self-hosted instances that use baseUrl
without an API key.
Return "local" when baseUrl or HONCHO_BASE_URL is set, matching the
client.py behavior that already handles this case for the SDK.
Tested on: macOS, self-hosted Honcho (Docker, localhost:8000).
Wraps _session_cache mutations in threading.RLock. Without this, concurrent
gateway sessions (e.g., Telegram + Discord hitting Honcho at the same time)
can race on the cache and silently lose conclusions or memory writes.
Adopted from #13510 by @hekaru-agent; the off-topic cron/jobs.py cleanup
hunk from that PR is dropped here for scope isolation. Resolved a small
conflict with the pinPeerName guard (kept both).
Gateway session keys (Matrix "!room:server" + thread event IDs, Telegram
supergroup reply chains, Slack thread IDs with long workspace prefixes) can
exceed Honcho's 100-character session ID limit after sanitization. Every
Honcho API call for those sessions then 400s with "session_id too long".
Add a helper that enforces the 100-char limit after sanitization:
short keys (the common case) short-circuit unchanged; over-limit keys
keep a prefix and append a deterministic `-<8 hex>` SHA-256 suffix over
the original key so two long keys sharing a leading segment can't
collide onto the same truncated ID.
Adds 7 regression tests in tests/honcho_plugin/test_client.py covering
short / exact-limit / long / deterministic / collision-resistant /
allowlist-preserving / hash-suffix-present cases.
CI caught that ``test_session_manager_prefers_runtime_user_id_over_config_peer_name``
in ``tests/agent/test_memory_user_id.py`` failed after this branch: that
test passes a ``MagicMock`` for ``config``, where
``mock.pin_peer_name`` silently returns another ``MagicMock`` — truthy by
default. My ``getattr(..., "pin_peer_name", False)`` fallback was
supposed to guard against callers that haven't added the new attr, but
MagicMock *does* have the attr — it just returns a live mock for it.
Tightened the gate to ``getattr(..., False) is True``. Real configs
built via ``HonchoClientConfig.from_global_config`` always yield a
proper boolean, so strict equality matches the pinned case and rejects
both the unset-attr fallback and MagicMock stand-ins. Added a comment
explaining why ``is True`` is intentional, not paranoid.
Also tightened the ``peer_name`` existence check to
``getattr(..., None)`` so a MagicMock with ``peer_name`` left at its
default (also truthy) doesn't spuriously enable pinning either.
Verified against both the new ``test_pin_peer_name.py`` suite (13/13
pass) and the previously-failing
``TestHonchoUserIdScoping`` (3/3 pass). Zero behaviour change for real
``HonchoClientConfig`` values.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
When a gateway drives Hermes (Telegram, Discord, Slack, ...), it passes the
platform-native user ID as ``runtime_user_peer_name`` into the Honcho
session manager. That ID wins over ``peer_name`` in ``honcho.json``, so a
single user who connects over three platforms ends up as three separate
Honcho peers — one per platform — with fragmented memory and no cross-
platform context continuity.
For multi-user bots this is correct (and must not change): each user gets
their own peer scope. For the vast majority of personal Hermes deployments
the configured ``peer_name`` is an unambiguous identity, though, so the
reporter asked for an opt-in knob that pins the user peer to that value.
Fix: new ``pinPeerName`` boolean on the host config, default ``false``.
When ``true`` AND ``peerName`` is set, the configured peer_name beats the
gateway's runtime identity; every other resolution case is unchanged.
honcho.json:
{
"peerName": "Igor",
"hosts": {
"hermes": { "pinPeerName": true }
}
}
session.py (resolution order, pinned case):
runtime_user_peer_name → skipped (opt-in flag active)
config.peer_name → WINS "Igor"
session-key fallback → unreached
Parsing follows the same host-block-overrides-root pattern as every other
flag in HonchoClientConfig.from_global_config (``_resolve_bool`` helper).
Tests (tests/honcho_plugin/test_pin_peer_name.py — 13 cases, 5 groups):
- Config parsing: default, root true, host-block true, host overrides
root, explicit false.
- Peer resolution: runtime wins by default (regression guard for multi-
user bots), config wins when pinned, pin-without-peer_name is a no-op
(prevents silent peer-id collapse to session-key fallback), CLI path
where runtime is absent, deepest fallback intact, assistant peer
untouched by the flag.
- Cross-platform unification: Telegram UID + Discord snowflake collapse
to one peer when pinned; negative control confirms two distinct
runtime IDs still produce two peers when unpinned.
244 honcho_plugin tests pass, 3 pre-existing skips, zero regressions.
Defensive detail: session.py uses ``getattr(self._config, "pin_peer_name",
False)`` so callers building partial config objects (several test fixtures
across the codebase do this) don't break if they haven't updated yet.
Runtime cost: one attr lookup per new session.
Closes#14984
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* feat(nix): parameterize dependency-groups in python.nix
* refactor(nix): extract package to callPackage-able hermes-agent.nix
Makes the package overridable via .override{} and adds
extraPythonPackages parameter for PYTHONPATH injection.
Includes build-time collision check using PEP 503 name
canonicalization.
* feat(nix): add overlay for external NixOS consumption
External flakes can now add overlays = [ inputs.hermes-agent.overlays.default ]
to get pkgs.hermes-agent with full .override support.
* test(nix): add check for extraPythonPackages PYTHONPATH injection
Verifies wrapper has PYTHONPATH when extras provided, and
base package has no PYTHONPATH without extras.
* feat(nix): add extraPlugins option for directory-based plugins
Symlinks plugin packages into HERMES_HOME/plugins/ at activation time.
Validates plugin.yaml presence. Asserts unique plugin names at eval time.
Hermes discovers them automatically via its directory scan.
* feat(nix): add extraPythonPackages option for entry-point plugins
Overrides the hermes package with PYTHONPATH injection when
extraPythonPackages is non-empty. Plugin .dist-info directories
become visible to importlib.metadata for entry-point discovery.
Works in both native systemd and container modes.
* docs: add NixOS declarative plugin installation to nix-setup, plugins, and build-a-plugin guides
- nix-setup.md: new Plugins section with extraPlugins/extraPythonPackages
examples, overlay usage, collision checking note, options reference rows
- plugins.md: Nix row in discovery table, NixOS declarative plugins section
- build-a-hermes-plugin.md: Distribute for NixOS section after pip section
* fix: address review feedback — remove unrelated umask, fix fetchFromGitHub naming, simplify checks
- Remove accidentally introduced umask/migration changes (unrelated to plugins)
- Add pluginName helper, fix fetchFromGitHub producing name='source'
- Show name= in extraPlugins example docs
- Simplify checks.nix: use hermes-agent.override instead of re-callPackage
- Fix fragile grep shell logic in checks
* refactor: address simplify feedback — lib.getName, drop unused inputs', Python list for extras
- Use lib.getName instead of custom pluginName helper
- Drop unused inputs' from checks.nix perSystem args
- Pass extraPythonPackages as Python list literal instead of colon-split string
* fix: walk propagatedBuildInputs for plugin PYTHONPATH and collision check
Uses python312.pkgs.requiredPythonModules to resolve the full transitive
closure of extraPythonPackages. Without this, a plugin with third-party
deps (e.g. requests) would fail at runtime if those deps weren't already
in the sealed uv2nix venv. The collision check now also scans the full
closure, catching transitive conflicts.
* cleanup: fold plugins into subdir loop, use find for symlink cleanup, inline lib.getName
- Add 'plugins' to the existing cron/sessions/logs/memories subdir loop
instead of a separate mkdir/chown/chmod block
- Replace fragile for-glob with find -delete for stale symlink cleanup
- Inline lib.getName at both call sites, remove pluginName wrapper
* fix: bypass FTS5 for CJK queries in session_search
FTS5 default tokenizer splits CJK characters into individual tokens,
so multi-character queries like "大别山项目" become AND of single chars.
This produces few/no results compared to LIKE substring search.
For CJK queries, skip FTS5 entirely and use LIKE for accurate
phrase matching.
Fixes NousResearch/hermes-agent#15500
* fix: cache _contains_cjk, escape LIKE wildcards, add regression tests
On top of the CJK FTS5 bypass from #15509:
- Cache _contains_cjk() result in a local var to avoid redundant O(n)
scans on every CJK query
- Escape %, _ in LIKE queries so literal wildcards in user input are
not treated as SQL wildcards (consistent with other LIKE queries in
hermes_state.py that use ESCAPE '\')
- Fix misleading comment ('or CJK fallback' → accurate description)
- Add 3 regression tests:
- test_cjk_partial_fts5_results_supplemented_by_like (#15500 / #14829)
- test_cjk_like_dedup_no_duplicates
- test_cjk_like_escapes_wildcards (new wildcard escaping)
* feat: trigram FTS5 index for CJK search, replace LIKE fallback
Replace the LIKE '%query%' full-table-scan fallback for CJK queries with
a proper trigram FTS5 index (messages_fts_trigram). The trigram tokenizer
creates overlapping 3-byte sequences so substring matching works natively
for any script — CJK, Thai, etc.
For queries with 3+ CJK characters: uses the trigram FTS5 table with
proper ranking, snippets, and indexed lookups. For shorter queries
(1-2 CJK chars): falls back to LIKE since the trigram tokenizer needs
≥9 UTF-8 bytes (3 CJK chars) minimum.
Schema v10 migration creates the trigram table and backfills existing
messages. Triggers keep the index in sync on INSERT/UPDATE/DELETE.
Builds on top of #16276 (bypass FTS5 for CJK, escape LIKE wildcards).
---------
Co-authored-by: vominh1919 <vominh1919@gmail.com>
Keep the parity test backed by the real Python command registry while avoiding hard failures in Node-only Vitest environments that cannot import hermes_cli.commands.
- config.py: remove dead ENV_VARS_BY_VERSION[17] entry (current _config_version
is 22, so all users are past version 17 and would never be prompted for
GMI_API_KEY on upgrade — consistent with how arcee was added)
- auxiliary_client.py: use google/gemini-3.1-flash-lite-preview as GMI aux
model instead of anthropic/claude-opus-4.6 (matches cheap fast-model pattern
used by all other providers: zai→glm-4.5-flash, kimi→kimi-k2-turbo-preview,
stepfun→step-3.5-flash, kilocode→google/gemini-3-flash-preview)
- test_gmi_provider.py: fix malformed write_text() call in doctor test
(was: write_text("GMI_API_KEY=*** encoding="utf-8") → missing closing quote,
wrote literal string 'GMI_API_KEY=*** encoding=' to .env file)
- test_gmi_provider.py + test_auxiliary_client.py: update aux model assertions
to match new cheaper default
- docs/integrations/providers.md: add 'gmi' to inline 'Supported providers'
fallback list (was only in the table, not the inline list at line ~1181)
- docs/reference/cli-commands.md: add 'gmi' to --provider choices list
Distinguish missing model from unsupported model before enabling fast mode and cover both cases so config and live agent state remain untouched on invalid fast toggles.
Match classic CLI parity by refusing to enable fast mode when the active model cannot produce fast request overrides, avoiding a misleading fast status with no runtime effect.
Make `config.set fast status` read-only and keep live agent request overrides in sync with fast-mode toggles so runtime API kwargs match the selected mode.
Expose a small forceRedraw API from @hermes/ink and use it for Ctrl/Cmd+L so the hotkey performs a real terminal clear + full repaint instead of a no-op state patch.
Use explicit repaint patch semantics for Ctrl/Cmd+L and narrow the hotkey assertion to the actual +L entry so unrelated descriptions do not cause false failures.
Harden busy mode config reads against invalid display config shapes and align /fast help+usage text with accepted aliases, with regression coverage for non-dict display values.
Make Ctrl+L non-destructive by redrawing the current screen state instead of starting a new session, and stop auto-appending --global for typed /model commands so session scope remains the default unless explicitly requested.
Route /browser, /reload-mcp, /rollback, /stop, /fast, and /busy through direct TUI RPC handlers so state changes hit the live gateway session instead of slash-worker fallback. Add TUI session finalize/reset parity hooks (memory commit + plugin boundaries) and parity matrix tests to keep mutating commands off fallback.
Handle queued-title ValueError cleanup during session init, harden Discord message source building for test stubs, and fix the Dockerfile contract test syntax error. Also refresh the TUI lockfile and Nix build flags so nix ubuntu-latest no longer fails on npm lock/peer resolution drift.
Retry queued pending titles even when the DB already has a non-empty title so explicit user title intents are not silently lost (for example after auto-title). Includes regression coverage.
Tighten pending-title flush during session init and treat row lookup failures during title-set no-op detection as RPC errors instead of silently queueing.
Handle session.title read failures without crashing, distinguish no-op title writes from missing session rows, and use a distinct empty-title error code with regression coverage.
- create HERMES_TUI_ACTIVE_SESSION_FILE with mkstemp instead of a predictable tmp path and always cleanup in finally
- add assertions that launch wiring uses a randomized session file path and removes it on exit
- use a grouped last_active join in search_sessions to avoid per-row correlated max lookups
- always close SessionDB in _resolve_last_session via finally and add regression coverage for search failure cleanup
- order session listing by computed last_active in SessionDB so callers get MRU rows directly
- keep _resolve_last_session as a single-row lookup and add regression coverage for >20 session sampling
Route TUI /title through session.title RPC and queue titles when the session DB row is still initializing, so renamed sessions reliably appear in /resume and browse flows.
The auto-lowered-threshold warning only named the compression model,
making it confusing when the main and aux models are configured with
the same slug but end up with different resolved context lengths (e.g.
OpenRouter's stepfun/step-3.5-flash catalog value vs. a main-model
context_length override). Users couldn't tell whether the warning
reflected two different models or a context-resolution mismatch.
Now includes both 'model (provider)' labels. The aux provider falls
back to the client's base_url hostname when the configured provider
is 'auto', so users see where compression is actually being called.
Thread a vision-request flag through auxiliary provider resolution so Copilot clients can include Copilot-Vision-Request only for vision tasks. This preserves normal text requests while ensuring Copilot vision payloads reach the vision-capable route.
Add regression coverage for Copilot vision routing and keep cached text and vision clients separate so a text client without the header is not reused for vision.
Co-authored-by: dhabibi <9087935+dhabibi@users.noreply.github.com>
* fix: clean gateway auxiliary client caches on teardown
* fix(gateway): recover from stale pid files and close cron agents
Two issues were keeping the gateway from surviving long runs:
1. `_cleanup_invalid_pid_path` delegated to `remove_pid_file`, which
refuses to unlink when the file's pid differs from our own. That
safety check exists for the --replace atexit handoff, but it also
applied to stale-record cleanup, so after a crashy exit the pid
file was orphaned: `write_pid_file()`'s O_EXCL create then failed
with `FileExistsError`, and systemd looped on "PID file race lost
to another gateway instance". Unlink unconditionally from this
helper since the caller has already verified the record is dead.
2. The cron scheduler never closed the ephemeral `AIAgent` it creates
per tick, and never swept the process-global auxiliary-client
cache. Over days of 10-minute ticks this leaked subprocesses and
async httpx transports until the gateway hit EMFILE. Release the
agent and call `cleanup_stale_async_clients()` in `run_job`'s
outer `finally`, matching the gateway's own per-turn cleanup.
* chore(release): map bloodcarter@gmail.com -> bloodcarter
---------
Co-authored-by: bloodcarter <bloodcarter@gmail.com>
When a paste takes longer than 500ms to process on the prompt_toolkit
event-loop thread, emit a logger.warning with elapsed time, byte size,
line count, and sys.platform. Gives us concrete repro data for the
recurring 'CLI freezes after paste on macOS' class of reports (issue
#16263, plus sibling reports across Claude Code / Cursor / Lightroom
against macOS Tahoe 26).
Pure diagnostic — no behavior change. Two time.perf_counter() calls
and one conditional per paste event. Log line only fires when the
handler is actually slow, so normal pastes add no log noise.
The backup takes a consistent snapshot of each .db via sqlite3.backup(),
so shipping the live .db-wal / .db-shm / .db-journal alongside pairs the
fresh snapshot with stale sidecar state and produces a torn restore on
first open. Sidecars are transient and SQLite regenerates them on next
connection anyway.
This also trims multi-MB of junk from every zip — state.db-wal alone was
~9 MB here, doubled by the fact the WAL is the live write-ahead log, not
data.
PR #13734 fixed the concurrent-tool-executor vector (ThreadPoolExecutor
workers didn't inherit the CLI's TLS approval callback). Two vectors
remained that could still land in the deadlocking input() fallback:
1. _spawn_background_review spawns a raw threading.Thread with no
approval callback installed, so any dangerous-command guard the
review agent trips falls back to input() -> deadlock against the
parent's prompt_toolkit TUI (same class as delegate_task subagents,
fixed in 023b1bff1 / #15491). Install a _bg_review_auto_deny
callback at thread start, clear on finally.
2. prompt_dangerous_approval's fallback unconditionally spawned a
daemon thread calling input() when approval_callback was None.
That fallback can never succeed under prompt_toolkit because the
user's Enter goes to pt's raw-mode stdin capture. Detect an active
pt Application via get_app_or_none() and fail closed (deny + log)
instead, so future threads that forget to install a callback
degrade gracefully instead of hanging 60s invisibly.
Regression guards:
- tests/run_agent/test_background_review.py verifies the review
worker thread sees a callable auto-deny callback mid-run and that
the slot is cleared in the finally block.
- tests/tools/test_approval.py TestFailClosedUnderPromptToolkit
verifies prompt_dangerous_approval returns 'deny' fast under a
mocked pt Application, and that a real callback still wins over
the guard.
When tools execute concurrently via ThreadPoolExecutor, worker threads
could not see the thread-local approval/sudo callbacks registered by
the CLI. This caused dangerous-command prompts to fall back to plain
input(), which deadlocks against prompt_toolkit's raw terminal mode.
Capture parent-thread callbacks before launching workers, register
them locally in each _run_tool thread, and clear them on exit.
Mirrors the existing fix pattern from cli.py run_agent() for the
main agent worker thread (GHSA-qg5c-hvr5-hjgr / #13617).
The background skill/memory review agent was created without toolset
restrictions, inheriting the full default tool set. This allowed it to
use terminal, send_message, delegate_task, and other tools outside its
intended scope, potentially performing unrelated side effects after
skill creation.
Restrict the review agent to only memory and skills toolsets by passing
enabled_toolsets=['memory', 'skills'] during AIAgent construction.
Fixes#15204
The gateway fix in the previous commit forwards _session_messages on
gateway session teardown. The CLI exit cleanup path had the same bug:
it read getattr(agent, 'conversation_history', None) or [] — but AIAgent
has no conversation_history attribute, so providers always received [].
Switch to _session_messages (same attribute the gateway now uses),
guarded by isinstance(..., list) to preserve the no-arg fallback for
MagicMock-based CLI test stubs.
Adds tests/cli/test_cli_shutdown_memory_messages.py (4 cases mirroring
the gateway suite).
``_cleanup_agent_resources`` previously invoked
``agent.shutdown_memory_provider()`` with no arguments, so every memory
provider's ``on_session_end`` hook received an empty list. Providers
with an early-return guard on empty input (Holographic, Hindsight) never
extracted facts from the conversation, and users hit
"抱歉，找不到相關的對話記錄" on the first turn after any gateway
restart, session reset, or idle expiry.
Forward ``agent._session_messages`` — the transcript the agent itself
maintains and refreshes every turn via ``_persist_session`` — so
providers see the actual conversation. Falls back to the legacy no-arg
call whenever the attribute is absent or not a list (test stubs built
via ``object.__new__`` or ``MagicMock``) to preserve backward
compatibility with existing suites. ``AIAgent.shutdown_memory_provider``
already accepts ``messages: list = None`` (run_agent.py:4126), so this
is a pure caller-side fix.
Paths that use ``skip_memory=True`` temporary agents (memory flush,
hygiene auto-compress, ``/compress``) are no-ops inside
``shutdown_memory_provider`` because ``self._memory_manager`` is None —
no behaviour change for them.
Covers Part A of the bug report. Part B (adding ``on_session_end`` to
the Hindsight plugin) is a separate concern that would benefit from
this fix landing first.
Regression test added at
``tests/gateway/test_shutdown_memory_provider_messages.py`` covering:
populated messages forwarded, empty list still forwarded, attribute
missing falls back, non-list (MagicMock) falls back, provider
exceptions don't block ``close()``, None agent no-op, and agent
without ``shutdown_memory_provider`` tolerated.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Session-local trajectory cache — keyed by session hash, regenerated
per-session, won't port to another machine anyway. On a large install
this was multiple GB of pure noise in every zip.
Also adds a regression test for the pre-existing backups/ exclusion
so the two machine-local dirs share coverage.
The zip backup could add minutes to every 'hermes update' on large
HERMES_HOME directories. Flip the default to off and add a --backup
flag for one-off opt-in runs.
- updates.pre_update_backup default: True -> False
- hermes update: new --backup flag (opposite of existing --no-backup)
- Silent no-op when disabled (no message spam on every update)
- Existing --no-backup still works and wins over --backup
- Users who explicitly set pre_update_backup: true keep the old behavior
- Tests updated to cover default-off, --backup opt-in, and config-enabled paths
* feat(image-input): native multimodal routing based on model vision capability
Attach user-sent images as OpenAI-style content parts on the user turn when
the active model supports native vision, so vision-capable models see real
pixels instead of a lossy text description from vision_analyze.
Routing decision (agent/image_routing.py::decide_image_input_mode):
agent.image_input_mode = auto | native | text (default: auto)
In auto mode:
- If auxiliary.vision.provider/model is explicitly configured, keep the
text pipeline (user paid for a dedicated vision backend).
- Else if models.dev reports supports_vision=True for the active
provider/model, attach natively.
- Else fall back to text (current behaviour).
Call sites updated: gateway/run.py (all messaging platforms), tui_gateway
(dashboard/Ink), cli.py (interactive /attach + drag-drop).
run_agent.py changes:
- _prepare_anthropic_messages_for_api now passes image parts through
unchanged when the model supports vision — the Anthropic adapter
translates them to native image blocks. Previous behaviour
(vision_analyze → text) only runs for non-vision Anthropic models.
- New _prepare_messages_for_non_vision_model mirrors the same contract
for chat.completions and codex_responses paths, so non-vision models
on any provider get text-fallback instead of failing at the provider.
- New _model_supports_vision() helper reads models.dev caps.
vision_analyze description rewritten: positions it as a tool for images
NOT already visible in the conversation (URLs, tool output, deeper
inspection). Prevents the model from redundantly calling it on images
already attached natively.
Config default: agent.image_input_mode = auto.
Tests: 35 new (test_image_routing.py + test_vision_aware_preprocessing.py),
all existing tests that reference _prepare_anthropic_messages_for_api
still pass (198 targeted + new tests green).
* feat(image-input): size-cap + resize oversized images, charge image tokens in compressor
Two follow-ups that make the native image routing safer for long / heavy
sessions:
1) Oversize handling in build_native_content_parts:
- 20 MB ceiling per image (matches vision_tools._MAX_BASE64_BYTES,
the most restrictive provider — Gemini inline data).
- Delegates to vision_tools._resize_image_for_vision (Pillow-based,
already battle-tested) to downscale to 5 MB first-try.
- If Pillow is missing or resize still overshoots, the image is
dropped and reported back in skipped[]; caller falls back to text
enrichment for that image.
2) Image-token accounting in context_compressor:
- New _IMAGE_TOKEN_ESTIMATE = 1600 (matches Claude Code's constant;
within the realistic range for Anthropic/GPT-4o/Gemini billing).
- _content_length_for_budget() helper: sums text-part lengths and
charges _IMAGE_CHAR_EQUIVALENT (1600 * 4 chars) per image/image_url/
input_image part. Base64 payload inside image_url is NOT counted
as chars — dimensions don't matter, only image-presence.
- Both tail-cut sites (_prune_old_tool_results L527 and
_find_tail_cut_by_tokens L1126) now call the helper so multi-image
conversations don't slip past compression budget.
Tests: 9 new in test_image_routing.py (oversize triggers resize,
resize-fails-returns-None, oversize-skipped-reported), 11 new in
test_compressor_image_tokens.py (flat charge per image, multiple images,
Responses-API / Anthropic-native / OpenAI-chat shapes, no-inflation on
raw base64, bounds-check on the constant, integration test that an
image-heavy tail actually gets trimmed).
* fix(image-input): replace blanket 20MB ceiling with empirically-verified per-provider limits
The previous commit imposed a hardcoded 20 MB base64 ceiling on all
providers, triggering auto-resize on anything larger. This was wrong in
both directions:
* Too loose for Anthropic — actual limit is 5 MB (returns HTTP 400
'image exceeds 5 MB maximum' above that).
* Too strict for OpenAI / Codex / OpenRouter — accept 49 MB+ without
complaint (empirically verified April 2026 with progressive PNG
sizes).
New behaviour:
* _PROVIDER_BASE64_CEILING table: only anthropic and bedrock have a
ceiling (5 MB, since bedrock-on-Claude shares Anthropic's decoder).
* Providers NOT in the table get no ceiling — images attach at native
size and we trust the provider to return its own error if it
disagrees. A provider-specific 400 message is clearer than us
guessing wrong and silently degrading image quality.
* build_native_content_parts() gains a keyword-only provider arg;
gateway/CLI/TUI pass the active provider so Anthropic users get
auto-resize protection while OpenAI users don't pay it.
* Resize target dropped from 5 MB to 4 MB to slide safely under
Anthropic's boundary with header overhead.
Empirical measurements (direct API, no Hermes in the loop):
image b64 anthropic openrouter/gpt5.5 codex-oauth/gpt5.5
0.19 MB ✓ ✓ ✓
12.37 MB ✗ 400 5MB ✓ ✓
23.85 MB ✗ 400 5MB ✓ ✓
49.46 MB ✗ 413 ✓ ✓
Tests: rewrote TestOversizeHandling (5 tests): no-ceiling pass-through,
Anthropic resize fires, Anthropic skip on resize-fail, build_native_parts
routes ceiling by provider, unknown provider gets no ceiling. All 52
targeted tests pass.
* refactor(image-input): attempt native, shrink-and-retry on provider reject
Replace proactive per-provider size ceilings with a reactive shrink path
on the provider's actual rejection. All providers now attempt native
full-size attachment first; if the provider returns an image-too-large
error, the agent silently shrinks and retries once.
Why the previous design was wrong: hardcoding provider ceilings
(anthropic=5MB, others=unlimited) meant OpenAI users on a 10MB image
paid no tax, but Anthropic users lost quality on anything >5MB even
though the empirical behaviour at provider-reject time is the same
(shrink + retry). Baking the table into the routing layer also
requires updating Hermes every time a provider's limit changes.
Reactive design:
- image_routing.py: _file_to_data_url encodes native size, no ceiling.
build_native_content_parts drops its provider kwarg.
- error_classifier.py: new FailoverReason.image_too_large + pattern
match ("image exceeds", "image too large", etc.) checked BEFORE
context_overflow so Anthropic's 5MB rejection lands in the right
bucket.
- run_agent.py: new _try_shrink_image_parts_in_messages walks api
messages in-place, re-encodes oversized data: URL image parts
through vision_tools._resize_image_for_vision to fit under 4MB,
handles both chat.completions (dict image_url) and Responses
(string image_url) shapes, ignores http URLs (provider-fetched).
New image_shrink_retry_attempted flag in the retry loop fires the
shrink exactly once per turn after credential-pool recovery but
before auth retries.
E2E verified live against Anthropic claude-sonnet-4-6:
- 17.9MB PNG (23.9MB b64) attached at native size
- Anthropic returns 400 "image exceeds 5 MB maximum"
- Agent logs '📐 Image(s) exceeded provider size limit — shrank and
retrying...'
- Retry succeeds, correct response delivered in 6.8s total.
Tests: 12 new (8 shrink-helper shapes + 4 classifier signals),
replaces 5 proactive-ceiling tests with 3 simpler 'native attach works'
tests. 181 targeted tests pass. test_enum_members_exist in
test_error_classifier.py updated for the new enum value.
* feat(plugins): google_meet — bundled plugin for join+transcribe Meet calls
v1 shipping transcribe-only. Spawns headless Chromium via Playwright,
joins an explicit https://meet.google.com/ URL, enables live captions,
and scrapes them into a transcript file the agent can read across turns.
The agent then has the meeting content in context and can do followup
work (send recap, file issues, schedule followups) with its regular tools.
Surface:
- Tools: meet_join, meet_status, meet_transcript, meet_leave, meet_say
(meet_say is a v1 stub — returns not-implemented; v2 will wire
realtime duplex audio via OpenAI Realtime / Gemini Live +
BlackHole / PulseAudio null-sink.)
- CLI: hermes meet setup | auth | join | status | transcript | stop
- Lifecycle: on_session_end auto-leaves any still-running bot.
Safety:
- URL regex rejects anything that isn't https://meet.google.com/...
- No calendar scanning, no auto-dial, no auto-consent announcement.
- Single active meeting per install; a second meet_join leaves the first.
- Platform-gated to Linux + macOS (Windows audio routing for v2 untested).
- Opt-in: standalone plugin, user must add 'google_meet' to
plugins.enabled in config.yaml.
Zero core changes. Plugin uses existing register_tool /
register_cli_command / register_hook surfaces. 21 new unit tests cover the
URL safety gate, transcript dedup + status round-trip, process-manager
refusals/start/stop paths, tool-handler JSON shape under each branch,
session-end cleanup, and platform-gated register().
* feat(plugins/google_meet): v2 realtime audio + v3 remote node host
v2 \u2014 agent speaks in-meeting
audio_bridge.py: PulseAudio null-sink (Linux) + BlackHole probe (macOS).
On Linux we load pactl module-null-sink + module-virtual-source, track
module ids for teardown; Chrome gets PULSE_SOURCE=<virt src> env so its
fake mic reads what we write to the sink. macOS just probes BlackHole
2ch and returns its device name \u2014 the plugin refuses to switch the
user's default audio input (that would surprise them).
realtime/openai_client.py: sync WebSocket client for the OpenAI Realtime
API. RealtimeSession.speak(text) sends conversation.item.create +
response.create, accumulates response.audio.delta PCM bytes, appends
them to a file. RealtimeSpeaker runs a JSONL-queue loop consuming
meet_say calls. 'websockets' is an optional dep imported lazily.
meet_bot.py: when HERMES_MEET_MODE=realtime, provisions AudioBridge,
starts RealtimeSession + speaker thread, spawns paplay to pump PCM
into the null-sink, then cleans everything up on SIGTERM. If any
realtime setup step fails, falls back cleanly to transcribe mode
with an error flagged in status.json.
process_manager.enqueue_say(): writes a JSONL line to say_queue.jsonl;
refuses when no active meeting or active meeting is transcribe-only.
tools.meet_say: real implementation; requires active mode='realtime'.
meet_join: adds mode='transcribe'|'realtime' param.
v3 \u2014 remote node host
node/protocol.py: JSON envelope (type, id, token, payload) + validate.
node/registry.py: $HERMES_HOME/workspace/meetings/nodes.json, with
resolve() auto-selecting the sole registered node when name is None.
node/server.py: NodeServer \u2014 websockets.serve, bearer-token auth,
dispatches start_bot/stop/status/transcript/say/ping onto the local
process_manager. Token auto-generated + persisted on first run.
node/client.py: NodeClient \u2014 short-lived sync WS per RPC, raises
RuntimeError on error envelopes, clean API matching the server.
node/cli.py: 'hermes meet node {run,list,approve,remove,status,ping}'
subtree; wired into the main meet CLI by cli.py so 'hermes meet node'
Just Works.
tools.py: every meet_* tool accepts node='<name>'|'auto'; when set,
routes through NodeClient to the remote bot instead of running
locally. Unknown node \u2192 clear 'no registered meet node matches ...'
error.
cli.py: 'hermes meet join --node my-mac --mode realtime' and
'hermes meet say "..." --node my-mac' route to the node; 'hermes
meet node approve <name> <url> <token>' registers one.
Tests
21 v1 tests updated (meet_say is no longer a stub; active-record now
carries mode).
20 new audio_bridge + realtime tests.
42 new node tests (protocol/registry/server/client/cli).
17 new v1/v2/v3 integration tests at the plugin level covering
enqueue_say edge cases, env var passthrough, mode validation, node
routing (known/unknown/auto/ambiguous), and argparse wiring for
`hermes meet say` + `hermes meet node` + --mode/--node flags.
Total: 100 plugin tests + 58 plugin-system tests = 158 passing.
E2E verified on Linux with fresh HERMES_HOME: plugin loads, 5 tools
register, on_session_end hook wires, 'hermes meet' CLI tree wires
including the node subtree, NodeRegistry round-trips, meet_join routes
correctly to NodeClient under node='my-mac' with mode='realtime',
enqueue_say accepts realtime/rejects transcribe, argparse parses every
new flag cleanly.
Zero changes to core. All new code lives under plugins/google_meet/.
* feat(plugins/google_meet): auto-install, admission detect, mac PCM pump, barge-in, richer status
Ready-for-live-test follow-up on PR #16364. Five additions that matter for
the first live run on a real Meet, in priority order:
1. hermes meet install [--realtime] [--yes]
pip install playwright websockets + python -m playwright install chromium
--realtime: installs platform audio deps (pulseaudio-utils on Linux via
sudo apt, blackhole-2ch + ffmpeg on macOS via brew). Prompts before
sudo/brew unless --yes. Refuses on Windows. Refuses to auto-flip the
macOS default input — user still selects BlackHole in System Settings
(deliberate; surprise audio rerouting is worse than a manual step).
2. Admission detection
_detect_admission(page): Leave-button visible OR caption region
attached OR participants list present → we're in-call.
_detect_denied(page): 'You can\'t join this video call' / 'You were
removed' / 'No one responded to your request' → bail out.
HERMES_MEET_LOBBY_TIMEOUT (default 300s) caps how long we sit in
the lobby before giving up. in_call stays False until admitted.
Status surfaces leaveReason: duration_expired | lobby_timeout |
denied | page_closed.
3. macOS PCM pump
ffmpeg reads speaker.pcm (24kHz s16le mono) and writes to the
BlackHole AVFoundation output via -f audiotoolbox
-audio_device_index <N>. _mac_audio_device_index() probes
ffmpeg -f avfoundation -list_devices true to resolve 'BlackHole 2ch'
→ numeric index. Falls back to index 0 on probe failure. Linux
paplay pump unchanged.
4. Richer status dict
_BotState now tracks realtime, realtimeReady, realtimeDevice,
audioBytesOut, lastAudioOutAt, lastBargeInAt, joinAttemptedAt,
leaveReason. RealtimeSession.audio_bytes_out / last_audio_out_at
counters fold into the status file once a second so meet_status()
can show the agent's voice activity in near-real-time.
5. Barge-in
RealtimeSession.cancel_response() sends type='response.cancel' over
the same WS (lock-guarded so it's safe to call from the caption
thread while speak() is reading frames). Handles response.cancelled
as a terminal frame type. _looks_like_human_speaker() gates triggers
so the bot's own name, 'You', 'Unknown', and blanks don't self-cancel.
Called from the caption drain loop: when a new caption arrives
attributed to a real participant while rt.session exists, we fire
cancel_response() and stamp lastBargeInAt.
Tests: 20 new unit tests across _BotState telemetry, barge-in gating,
admission/denied probe error handling, cancel_response with and without
a connected WS, and `hermes meet install` CLI wiring (flag parsing +
end-to-end subprocess.run verification + Linux-already-installed fast
path). Total 171 passing across all google_meet test files + the
plugin-system regression suite.
E2E verified on Linux: plugin loads, all 5 tools register,
`hermes meet install --realtime --yes` parses, fresh-bot status.json
has every new telemetry key, cancel_response on a disconnected session
returns False without raising, barge-in helper gates the bot's own
name correctly.
Still out of scope (for a future PR, not blocking live test):
mic → Realtime duplex (the agent listening to meeting audio via
WebRTC), node-host TLS/pairing UX, Windows audio, Meet create+Twilio.
Docs updated: SKILL.md now lists the installer subcommand, lobby
timeout, barge-in caveat, and the full status-dict reference table.
README.md quick-start uses hermes meet install.
Every 'hermes update' now runs a full backup of ~/.hermes/ first, so
users can always roll back to the exact state they had before the
update if anything goes wrong (corrupted sessions.db, broken skills,
config migrations that don't round-trip, etc.).
Changes:
- hermes_cli/backup.py: new create_pre_update_backup() helper. Writes
to <HERMES_HOME>/backups/pre-update-<stamp>.zip using the same
exclusion rules and SQLite safe-copy as 'hermes backup'. Auto-rotates
(keep last N, pre-update-*.zip only — hand-dropped zips in backups/
are untouched). Adds 'backups' to _EXCLUDED_DIRS so subsequent backups
don't nest prior ones.
- hermes_cli/main.py: _run_pre_update_backup() wired into
_cmd_update_impl before any git operation. Prints save path, restore
command, and how to disable. Swallows failures so a broken backup
never blocks the update itself. New --no-backup flag on 'hermes
update' for one-off override.
- hermes_cli/config.py: new 'updates' section in DEFAULT_CONFIG with
pre_update_backup (default true) and backup_keep (default 5).
Auto-surfaces in the dashboard config UI.
- tests/hermes_cli/test_backup.py: +11 tests covering backup location,
content parity with 'hermes backup', no-recursion, rotation, manual
file preservation, config gate, --no-backup flag, flag-wins-over-config.
Adds a short always-on pointer to the system prompt: when the user asks
about configuring, setting up, troubleshooting, or using Hermes Agent
itself, load the hermes-agent skill via skill_view(name='hermes-agent')
and fall back to https://hermes-agent.nousresearch.com/docs via
web_extract. Keeps sessions without skill_view loaded useful too — the
docs URL + web_extract is enough to answer most questions.
The guidance is appended right after DEFAULT_AGENT_IDENTITY (or SOUL.md)
so it ships regardless of which toolset profile is active. Footprint is
~560 chars, behind the existing prompt cache.
The CLI renders through prompt_toolkit in non-full-screen mode, so every
repaint uses the renderer's tracked _cursor_pos.y to cursor_up() + erase
before drawing the new frame. Any time that tracked position drifts from
terminal reality, redraws stack on top of stale content instead of
overwriting it. Four user-visible bugs share this root cause.
Fixes:
- #5474 (SIGWINCH ghosts): the resize wrapper previously only handled
column-shrink reflow. Generalize it to force a full screen-clear
(erase_screen + cursor_goto(0,0)) and renderer.reset() on every resize
— covers widen, row-shrink, and multiplexer SIGWINCH-less redraws.
- #8688 (cmux/tmux tab switch): no SIGWINCH fires on focus regain, so
prompt_toolkit has no signal to recover. Add a _force_full_redraw()
helper, bound to Ctrl+L (standard bash/zsh/vim convention) and exposed
as /redraw. Users can manually clear drift without restarting Hermes.
- #14692 (DSR response leaks — ^[[53;1R): resize storms make
prompt_toolkit's CSI 6n queries race past the input parser; the
terminal's reply ends up as literal input text. Add a sibling of the
bracketed-paste sanitizer that strips \x1b[<row>;<col>R and the
caret-escape visible form from paste text, buffer text-filter, and
the input-processing loop.
The idle-redraw removal (#12641) is in the preceding commit from
@foxion37 — keeping them as separate commits preserves attribution.
On provider switches mid-session (e.g. MiniMax -> DeepSeek), the source
assistant turn carries a 'reasoning' field written by the prior provider
but no 'reasoning_content' key. _copy_reasoning_content_for_api would
promote that foreign 'reasoning' to 'reasoning_content' on the outbound
DeepSeek request, leaking a cross-provider chain of thought and in
practice causing HTTP 400.
DeepSeek's own _build_assistant_message always pins reasoning_content=''
at creation time for tool-call turns, so the shape (reasoning set,
reasoning_content absent, tool_calls present) is unreachable from
same-provider DeepSeek history — it can only come from a prior provider.
Pad with '' in that case instead of promoting.
Healthy same-provider 'reasoning' promotion (no tool_calls, or on
providers that do not require the empty-string pin) is unchanged.
Defensive: when the generator encounters a fenced code block containing
Unicode box-drawing characters, wrap it in `<!-- ascii-guard-ignore -->`
markers so the docs-site-checks lint (which scans inside code fences)
can't reject the page for a skill's own diagram.
Plain bash/python code blocks stay uncluttered — only blocks with box
chars get wrapped. Skill authors no longer have to remember to add the
ignore markers in every SKILL.md with ASCII art.
Fixes#15305.
Previously 'hermes debug share' uploads only got DELETEd when the user
ran 'hermes debug share' again — opportunistic-sweep-on-invoke was the
only cleanup path. A user who uploaded once and never ran debug again
left pastes up until paste.rs's retention kicked in (which, empirically,
never actually expires them).
Hook _sweep_expired_pastes into the gateway cron ticker at the same
hourly cadence as the image/document cache cleanups. The opportunistic
sweep in 'hermes debug share' stays as a fallback for CLI-only users
who never start the gateway.
On macOS (bash 3.2 and some Homebrew bash builds) `source`ing a file that
contains `declare -x` statements prints each declaration to stdout. The
persistent-shell wrapper in tools/environments/base.py was only redirecting
stderr when sourcing the session snapshot, so ~60 lines of env vars leaked
into every terminal tool response — blowing out context and triggering
HTTP 400s on context-limited providers.
Fix: redirect both stdout and stderr when sourcing the snapshot. Linux
bash is silent here, so the redirect is harmless there; macOS no longer
leaks.
Closes#15459
Co-authored-by: Sanjays2402 <51058514+Sanjays2402@users.noreply.github.com>
Quick state snapshot now includes pairing JSONs (generic + legacy +
Feishu comment pairing), and `hermes update` takes a pre-update
snapshot labeled `pre-update` before pulling.
Pairing data lives outside state.db in platform-specific JSONs under
~/.hermes/pairing/, ~/.hermes/platforms/pairing/, and
~/.hermes/feishu_comment_pairing.json. The update command already
couldn't touch $HERMES_HOME, but #15733 reports lost pairing after
an update — this gives users something to restore from via
`/snapshot list` / `/snapshot restore <id>` if anything clobbers
the approved-user lists.
- Extend _QUICK_STATE_FILES with pairing paths (files + dirs)
- Snapshot walks directories recursively and records each file in the
manifest individually so restore logic is unchanged
- _cmd_update_impl calls create_quick_snapshot(label='pre-update')
after 'Found N new commits' and before 'Pulling updates'
- Snapshot failures are logged at debug and never block the update
Refs #15733.
read_file's dedup path returned a lightweight stub on re-reads of an
unchanged file, then returned early — so the consecutive-read loop
guard (hard block at count>=4) at the bottom of read_file_tool never
ran for stub-looped calls. Weaker tool-following models (local Qwen3.6
variants in the reported case) ignore the passive 'refer to earlier
result' hint and hammer the same read_file call until iteration budget
runs out.
Track per-key stub returns in task_data['dedup_hits'] and, on the
second stub for the same (path, offset, limit), return a hard BLOCKED
error mirroring the wording the real-read path already uses. A real
read, an intervening non-read tool call (notify_other_tool_call), or
reset_file_dedup (on context compression) all clear the counter so
the guard never stays engaged longer than the actual loop.
Closes#15759
Telegram groups emit a single bot_command entity covering the whole
/cmd@botname span with no accompanying mention entity, so the existing
mention gate in _message_mentions_bot dropped slash commands sent via
the bot-menu autocomplete whenever require_mention is enabled.
Recognise bot_command entities whose @botname suffix matches the bot
username (case-insensitive) as a direct mention, and keep rejecting
commands addressed at other bots. Fixes#15415.
When 'hermes model' runs against a providers: (keyed-schema) entry that
relies only on key_env, the picker resolves the env var for the live
/models request and then wrote a synthesized 'api_key: ${KEY_ENV}' back
to the providers.<key> entry. That's redundant — the runtime already
resolves from key_env directly — and it clutters configs that
intentionally keep credentials out of config.yaml.
Only persist provider_entry['api_key'] when the user originally had an
inline value (literal secret or ${VAR} template). Entries that declared
only key_env stay clean on save.
Fixes#15803.
For 14 of 74 compressed skills, the original description contained
trigger keywords, technique counts, attribution, or use-case phrases
not covered by the existing body content. Prepends a 'When to use' /
'What's inside' block near the top so the agent still has the full
context when the skill is loaded.
Skills salvaged:
- codex, ascii-video, creative-ideation, excalidraw, manim-video, p5js
- gif-search, heartmula, youtube-content
- lm-evaluation-harness, obliteratus, vllm, axolotl
- powerpoint
Remaining 60 skills were verified to already cover the dropped content
in their existing body sections (When to Use, overview, intro prose)
or had short descriptions fully captured by the new compressed form.
Target: every skill's description fits in a one-line gateway menu and
leads with trigger keywords an agent would match on. Drops filler like
'Use this skill to', 'A skill for', 'This skill provides'.
Before: max description length was 791 chars (architecture-diagram),
74 of 81 built-in skills were >60 chars.
After: max 60, mean 54, all 81 built-in skills <=60.
Rewritten with double-quoted YAML scalars to preserve Chinese/arrow
glyphs (baoyu-comic, yuanbao, youtube-content).
- claude-design: 'Design one-off HTML artifacts (landing, deck, prototype).' (57)
- popular-web-designs: '54 real design systems (Stripe, Linear, Vercel) as HTML/CSS.' (60)
- design-md: "Author/validate/export Google's DESIGN.md token spec files." (59)
Also adds an inline callout near the top of claude-design pointing to
popular-web-designs and design-md so the cross-reference lands even
without reading the full decision table.
- claude-design: design process + taste for one-off HTML artifacts
- popular-web-designs: 54 ready-to-paste design systems (Stripe/Linear/etc.)
- design-md: formal DESIGN.md token spec file authoring
Adds a comparison table to claude-design's 'When To Use' section and
reciprocal pointers in design-md and popular-web-designs. Also corrects
claude-design author attribution to BadTechBandit.
Harden the Matrix adapter's sender-drop guards so bot-self events and
appservice/bridge identities never reach the gateway's pairing flow or
the agent loop.
Two filters, applied as early as possible in _on_room_message (and
_on_reaction for the self-filter):
1. _is_self_sender(sender) — case-insensitive + whitespace-trimmed
equality with self._user_id. When self._user_id is still empty
(whoami has not resolved, or login failed), returns True
defensively: an unidentified bot dropping its own events is always
preferable to falling into an echo loop. The previous byte-for-byte
equality check let differently-cased copies of the bot's MXID slip
through, and an unresolved self-ID silently disabled the guard.
2. _is_system_or_bridge_sender(sender) — drops appservice namespace
puppets (conventional @_bridge_...:server form) and malformed
senders with an empty localpart. These identities used to fall
through to the gateway's unauthorized-user path, trigger a pairing
code, and — once an operator approved the bridge — every outbound
message the bridge relayed would loop back as an authorized user
message. This was the root of the 'hall of mirrors' symptom.
Fixes#15763
Test plan
---------
scripts/run_tests.sh tests/gateway/test_matrix.py
scripts/run_tests.sh tests/gateway/test_matrix_mention.py tests/gateway/test_matrix_voice.py
All 182 tests pass. 14 new regression tests cover exact / case-insensitive
/ whitespace / unresolved-self-id matches, bridge prefix detection, empty
sender, and the full _on_room_message drop path.
Closes#15775.
Title generation swallowed exceptions at debug level and returned None,
so a depleted auxiliary provider (e.g. OpenRouter 402) silently left
sessions with NULL titles. Reporter observed 45 untitled sessions
accumulated over 19 days with no user-visible indication.
- agent/title_generator.py: accept optional failure_callback, bump log
to WARNING, invoke callback on call_llm exception (swallowing callback
errors so nothing can crash the fire-and-forget worker thread).
- cli.py, gateway/run.py: pass agent._emit_auxiliary_failure as the
callback so failures route through the existing user-visible warning
channel.
- tests: cover callback fires / errors are swallowed / no-callback
legacy behavior / maybe_auto_title forwards kwarg to worker.
The bare-string isinstance guard added in 80ae2621 covered _find_tail_cut_by_tokens
(line 1084) but missed the identical pattern in _calculate_protect_tail_boundary
(line 487, the protect-tail scan loop). Both loops call .get("text", "") on every
list item in message["content"]; both crash with AttributeError when that list
contains a bare string.
Apply the same dict/str/fallback isinstance guard to the protect-tail path.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
raw_content from message["content"] can be a list that contains bare
strings, not only dicts. The previous `p.get("text", "")` call raised
AttributeError on string items, crashing context compression for any
session that had a message with mixed content.
Guard with isinstance checks: dict → .get("text"), str → len(p),
fallback → len(str(p)). Adds a regression test covering the bare-string
case that would have AttributeError'd on the pre-fix code.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
_find_tail_cut_by_tokens called len(content) to estimate message tokens.
When content is a list of blocks (multimodal: text + image_url), len()
returns block count (e.g. 2) rather than character count, so a message
with 500 chars of text was counted as ~10 tokens instead of ~135.
This caused the backward walk to exhaust all messages before hitting the
budget ceiling; the head_end safeguard then forced cut = n - min_tail,
shrinking the protected tail to the bare minimum and preventing effective
compression of long multimodal conversations.
Fix mirrors the existing pattern in _prune_old_tool_results (line 487):
sum(len(p.get("text", "")) for p in raw_content)
if isinstance(raw_content, list) else len(raw_content)
Tests: 3 new cases in TestTokenBudgetTailProtection — regression guard
(confirms the test fails with the bug), plain-string regression guard,
and image-only block edge case.
Fixes#16087.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
If the gateway's Python env loses access to 'croniter' between when a
cron job was created and when mark_job_run() fires, compute_next_run()
returns None for cron schedules. mark_job_run() treated that as terminal
completion and wrote enabled=false, state=completed — turning a missing
runtime dep into a silent, permanent job-off.
That behaviour is safe for one-shot jobs but wrong for recurring ones. A
missing dep should surface as an error the user can see, not as successful
completion of a job that is about to stop firing.
mark_job_run() now only disables the job on next_run_at=None when the
schedule is one-shot. For recurring (cron/interval) schedules it keeps
enabled=true, sets state=error, and records last_error so the user can
see why the job isn't advancing. compute_next_run() also logs a warning
the first time cron+no-croniter hits, so the underlying cause is visible
in the gateway log.
Tests cover:
- recurring cron job stays enabled with state=error when HAS_CRONITER=False
- recurring interval stays enabled when compute_next_run returns None
- one-shot jobs still flip to enabled=false, state=completed (no regression)
Fixes#16265
Azure Foundry deploys GPT-5.x, codex-*, and o1/o3/o4 reasoning models as
Responses-API-only. Calling /chat/completions against these deployments
returns 400 'The requested operation is unsupported.', which broke any
user who ran 'hermes model' on Azure, picked a gpt-5/codex deployment,
and kept the default api_mode: chat_completions. Verified in a user
debug bundle on 2026-04-26: gpt-5.3-codex failed on synopsisse.openai.azure.com
with that exact payload while gpt-4o-pure on the same endpoint worked.
Adds azure_foundry_model_api_mode(model_name) that returns
codex_responses when the model name starts with gpt-5, codex, o1, o3,
or o4 — otherwise None so chat_completions / anthropic_messages stay
untouched for gpt-4o, Llama, Claude-via-Anthropic, etc.
Resolver (both the direct Azure Foundry path and the pool-entry path)
consults it and upgrades api_mode unless the user explicitly picked
anthropic_messages. target_model (from /model mid-session switch)
takes precedence over the persisted default so switching from gpt-4o
to gpt-5.3-codex routes correctly before the next request.
Docs: correct the azure-foundry guide which previously claimed Azure
keeps gpt-5.x on chat completions — that was only true for early Azure
OpenAI, not Azure Foundry codex/o-series deployments.
Tests: 14 unit tests for azure_foundry_model_api_mode + 6 integration
tests in TestAzureFoundryResolution covering Bob's exact scenario,
target_model override, anthropic_messages guard, and o3-mini.
Follow-up to #16323 — the UrlSource adapter is shipped but four
user-facing docs surfaces still only listed the hub-identifier forms.
- user-guide/features/skills.md: add ``url`` to the Supported-hub-sources
table; add a new "#### 8. Direct URL (`url`)" section explaining scope
(single-file SKILL.md only), name-resolution order (frontmatter → URL
slug → interactive prompt → --name flag), and both TTY and
non-interactive usage. Add two URL examples to the install-examples
block near the top of the page.
- reference/cli-commands.md: two URL install examples + one note
explaining the name-resolution fallback chain.
- guides/work-with-skills.md: one URL-install example alongside the
existing hub-identifier examples.
- skills/autonomous-ai-agents/hermes-agent/SKILL.md: Quick Reference
block's ``hermes skills install`` line now spells out that ID can be
a hub identifier OR a direct SKILL.md URL, and mentions --name for
frontmatter-less skills.
No code changes. No new dependencies. Website builds via the usual
Docusaurus pipeline.
Co-authored-by: teknium1 <teknium@noreply.github.com>
Parse scope from the raw callback URL before stripping the auth code so Flow.fetch_token matches user-granted scopes. Add regression test for dual-scope callbacks.
Made-with: Cursor
Two related fixes for OpenClaw-residue problems after an OpenClaw→Hermes
migration (especially migrations done via OpenClaw's own tool, which
doesn't archive the source directory).
1. optional-skills/migration/openclaw-migration/scripts/openclaw_to_hermes.py:
rebrand_text() was rewriting ~/.openclaw/config.yaml → ~/.Hermes/config.yaml
(capital H — a directory that doesn't exist). Now case-preserving:
"OpenClaw" → "Hermes" (prose), but "openclaw" → "hermes" (so filesystem
paths land on the real Hermes home). Regex logic unchanged — replacement
function now checks if the matched text was all-lowercase and emits the
replacement in the matching case.
2. agent/onboarding.py + cli.py: one-time startup banner the first time
Hermes launches and finds ~/.openclaw/. Tells the user to run
`hermes claw cleanup` to archive it, gated on the existing onboarding
seen-flag framework (onboarding.seen.openclaw_residue_cleanup in
config.yaml). Fires once per install; re-running requires wiping that
flag or running cleanup directly.
Tests:
- 4 new TestDetectOpenclawResidue tests (present / absent / file-instead-
of-dir / default-home smoke)
- 2 TestOpenclawResidueHint tests (content check)
- 2 TestOpenclawResidueSeenFlag tests (flag isolation + round-trip)
- test_rebrand_text_preserves_filesystem_path_casing regression test
with 4 scenarios including the exact ~/.openclaw/config.yaml case
- Existing test_rebrand_text_* tests updated to the new case-preserving
contract (lowercase input → lowercase output)
Co-authored-by: teknium1 <teknium@noreply.github.com>
Four small tool-description / skill-content tweaks addressing recurring
model mistakes seen in @versun's docx feedback (Kimi 2.6, but the patterns
apply to every model):
1. browser_navigate description: call out .md/.txt/.json/.yaml/.csv/.xml,
raw.githubusercontent.com, and API endpoints as specifically preferring
curl or web_extract. The generic "prefer web_search or web_extract" was
too weak; models kept firing up the browser for plain-text URLs.
2. delegate_task description: two additions.
(a) Pass user language / output-style preferences in 'context' when they
differ from English — otherwise subagents default to English and their
summaries contaminate the final reply (caused the bilingual digest bug).
(b) Subagent summaries are self-reports, not verified facts. For
operations with external side-effects (HTTP uploads, remote writes,
file creation at shared paths), require a verifiable handle (URL, ID,
path) and verify it yourself before claiming success.
3. agent/prompt_builder.py Skills-mandatory block: new explicit line
"Whenever the user asks to configure / set up / modify / install /
enable / disable / troubleshoot Hermes Agent itself, load the
`hermes-agent` skill first." The generic "load what's relevant" didn't
route Hermes-meta questions (like "how do I turn off redaction?") to
the one skill that has the answer.
4. skills/autonomous-ai-agents/hermes-agent/SKILL.md: new "Security &
Privacy Toggles" section covering security.redact_secrets (with the
import-time-snapshot restart-required caveat), privacy.redact_pii,
approvals.mode (manual/smart/off) + --yolo + HERMES_YOLO_MODE, shell
hooks allowlist, and how to disable network/media tools entirely.
Every command verified against the actual config keys — no invented
knobs.
Co-authored-by: teknium1 <teknium@noreply.github.com>
* feat(skills): install skills from a direct HTTP(S) URL
Adds UrlSource adapter so `hermes skills install <url-to-SKILL.md>` and
`/skills install <url>` work as first-class operations — no more
improvising with curl + patch + cp.
- Claims identifiers that start with http(s):// and end in .md
- Skips /.well-known/skills/ URLs (WellKnownSkillSource handles those)
- Skill name from YAML frontmatter, URL-slug fallback
- Single-file SKILL.md only (v1 scope — multi-file skills need a manifest)
- Trust level 'community'; full security scan still runs
- Lock file stores the URL as identifier so `hermes skills update`
re-fetches from the same URL cleanly
Scope matches real user need from @versun's docx feedback where
`https://sharethis.chat/SKILL.md` had no first-class install path.
* feat(skills): interactive name/category for URL installs + --name override
Follow-up to the UrlSource adapter. The previous commit fell back to weak
heuristics when frontmatter had no ``name:`` and could produce garbage names
like ``SKILL`` or ``unnamed-skill``. Now:
tools/skills_hub.py
- ``UrlSource._is_valid_skill_name()`` — strict identifier check
(``^[a-z][a-z0-9_-]*$``), rejects sentinel values (``SKILL``, ``README``,
``INDEX``, ``unnamed-skill``, empty, non-strings).
- ``_resolve_skill_name()`` returns ``Optional[str]`` — ``None`` when
nothing valid is resolvable. Also ignores unsafe frontmatter names
(``../evil``) and falls through to URL slug instead of returning None
immediately, so a URL with a bad frontmatter but a good path still
works.
- ``fetch()``/``inspect()`` carry an ``awaiting_name=True`` marker in
metadata/extra when resolution fails, letting ``do_install`` decide
whether to prompt, apply an override, or error out.
hermes_cli/skills_hub.py
- ``do_install`` gains a ``name_override`` parameter.
- On URL-sourced bundles with ``awaiting_name=True``:
1. If ``name_override`` is valid → use it.
2. If ``name_override`` is invalid → refuse with a clear error.
3. Else if ``skip_confirm=True`` (non-interactive: slash / TUI /
gateway / scripts) → refuse with an actionable retry hint pointing
at ``--name <your-name>`` on both CLI and slash forms.
4. Else (interactive TTY) → prompt for the name.
- Interactive TTY also prompts for a category when none is given for a
URL-sourced install, hinting existing category buckets so users can
reuse ``productivity``, ``devops``, etc. Empty input → flat install.
- ``_existing_categories()`` scans ``~/.hermes/skills/`` for subdirs that
look like category buckets (contain nested SKILL.md files); skips
top-level skills and hidden dirs.
- ``_prompt_for_skill_name()`` / ``_prompt_for_category()`` helpers
(EOF/Ctrl-C-safe, match the existing ``Confirm [y/N]`` prompt style).
hermes_cli/main.py
- ``hermes skills install`` argparse gains ``--name <name>``.
hermes_cli/skills_hub.py (slash)
- ``/skills install <url> --name <x>`` parsing added.
Tests
- tests/tools/test_skills_hub.py: updated ``UrlSource`` tests to assert
the new ``awaiting_name`` metadata; added 4 new tests for
``_is_valid_skill_name`` rejection sets and the awaiting-name marker.
- tests/hermes_cli/test_skills_hub.py: 8 new tests covering --name
override accept/reject, non-interactive error, interactive name prompt,
interactive category prompt, cancel-aborts-install, and
``_existing_categories`` scan behavior (buckets vs flat skills).
- E2E verified all four paths (no-name/no-override → error;
--name override → install; frontmatter name → install;
invalid --name → rejection).
---------
Co-authored-by: teknium1 <teknium@noreply.github.com>
_search_members() and _fetch_messages() call min(limit, 100) assuming
limit is int. Models can pass limit as a string (e.g. "10"), causing
TypeError: '<' not supported between instances of 'str' and 'int'.
Add try/except int() coercion with safe defaults at the top of both
functions, matching the pattern used in session_search fix (#10522).
`_resolve_effective_accept()` used `return bool(cfg_val)` for the
`hooks_auto_accept` config key. In Python, `bool("false")` is `True`,
so a user setting `hooks_auto_accept: "false"` (quoted YAML string)
in `config.yaml` would silently enable auto-approval of every shell
hook, bypassing the consent prompt entirely.
Replace the coercion with the same type-aware parsing already used for
the HERMES_ACCEPT_HOOKS env var three lines above: bool passthrough,
strings checked against {1,true,yes,on} case-insensitively, everything
else (including "false", None, 0, ints) rejected.
Add TestHooksAutoAcceptParsing guarding the regression across all four
value shapes (bool, string-truthy, string-falsy, missing/None).
Reported by @sprmn24 in #16244.
Follow-up on top of #16243. Two small tweaks:
- Compile the regex once as `_SAFE_IDENTIFIER_RE` and pin it to
`[A-Za-z0-9@.+\-]`. The previous `\w` accepts Unicode word chars
(full-width digits, accented letters) which aren't valid WhatsApp
identifiers and shouldn't reach the mapping-file lookup.
- Add a comment clarifying this is defense-in-depth, not a live
traversal. The hardcoded `lid-mapping-{current}{suffix}.json`
prefix already prevents escape via pathlib's component split —
with `current='../secrets'`, the first path component under
`session/` is the literal directory name `lid-mapping-..`,
which the attacker cannot create.
E2E verified: legit mapping chains still resolve, all probed attack
shapes (`../`, absolute paths, shell metacharacters, Unicode digit
tricks) are rejected before any file access.
expand_whatsapp_aliases() interpolated untrusted identifiers directly
into filenames (lid-mapping-{current}.json) without validation.
An identifier containing ../ or / could escape the session directory.
Also replaced bare except Exception: continue with targeted
(OSError, json.JSONDecodeError) and a debug log so mapping
corruption is diagnosable instead of silently skipped.
Fixes:
- Reject identifiers with unsafe characters via re.match guard
- Replace broad exception swallow with specific catch + debug log
Both get_provider_request_timeout() and get_provider_stale_timeout()
wrapped the load_config import in try/except ImportError but left the
actual load_config() call unprotected. A corrupt config file, YAML
parse error, or permission failure would raise instead of returning
None safely.
Move load_config() inside the try block so any exception returns None.
- remove the temporary -c MRU logic and companion test from this branch so PR #15926 stays focused on TUI perf work
- keep the resume-ordering change isolated in the dedicated follow-up PR
- drop unused TUI helpers, test-only layout scaffolding, and stale public debug exports
- remove an unused profiler import and trim test-only coverage for deleted helpers
- gateway handler: turnController always archives in recordMessageComplete,
so the post-complete archiveTodosAtTurnEnd().forEach is dead code. Drop
it and the now-unused import.
- turnController: collapse archive prepend into a single spread expression.
- gateway server: one-line comment for the tool.start todo skip.
Two bugs surfaced together while the model fired the todo tool:
1. Count flickered (e.g. 3 → 1 → 3) because tool.start echoed
args.todos as the live state. With merge=true (or any partial
replacement) args.todos is just the items being updated, not the
full list. Drop the early echo — tool.complete already carries the
canonical full list from the tool result.
2. After turn end the panel jumped from under the user prompt to below
thinking/tools because archiveDoneTodos() was pushed AFTER segments
in finalMessages. Prepend the archive trail msg so it sits right
after the user prompt — same visual slot the live panel occupied
during streaming.
CPU profiling showed the built TUI loading React development modules unless NODE_ENV was set. Default CLI and dashboard TUI children to production while preserving explicit user overrides.
Keep history metadata consistent with lineage replay, globally order replayed lineage messages, and make Ink cache eviction report post-eviction sizes. Also keys TUI config cache by path to avoid cross-home test leakage.
When _compress_context rotates session_id (compression split), fire
on_session_start(new_sid, boundary_reason="compression",
old_session_id=<old>) on the active context engine. Plugin engines
(e.g. hermes-lcm) use this to preserve DAG lineage across the rollover
instead of re-initializing fresh per-session state.
Built-in ContextCompressor.on_session_start accepts **kwargs and ignores
them — no behavior change for default users.
Closes hermes-lcm#68 symptom: after Hermes compressed and minted a new
physical session, LCM was treating the split as a fresh /new and losing
continuity (compression_count: 1, store_messages: 0, dag_nodes: 0).
Credit: @Tosko4 (PR #13370) — minimized scope to the boundary_reason
signal only; the broader session-lifecycle refactor will be taken in
separate PRs if justified by concrete plugin need.
Every working dir hermes ever touches gets its own shadow git repo under
~/.hermes/checkpoints/{sha256(abs_dir)[:16]}/. The per-repo _prune is a
no-op (comment in CheckpointManager._prune says so), so abandoned repos
from deleted/moved projects or one-off tmp dirs pile up forever. Field
reports put the typical offender at 1000+ repos / ~12 GB on active
contributor machines.
Adds an opt-in startup sweep that mirrors the sessions.auto_prune
pattern from #13861 / #16286:
- tools/checkpoint_manager.py: new prune_checkpoints() and
maybe_auto_prune_checkpoints() helpers. Deletes shadow repos that
are orphan (HERMES_WORKDIR marker points to a path that no longer
exists) or stale (newest in-repo mtime older than retention_days).
Idempotent via a CHECKPOINT_BASE/.last_prune marker file so it only
runs once per min_interval_hours regardless of how many hermes
processes start up.
- hermes_cli/config.py: new checkpoints.auto_prune /
retention_days / delete_orphans / min_interval_hours knobs.
Default auto_prune: false so users who rely on /rollback against
long-ago sessions never lose data silently.
- cli.py / gateway/run.py: startup hooks gated on checkpoints.auto_prune,
called right next to the existing state.db maintenance block.
- Docs updated with the new config knobs.
- 11 regression tests: orphan/stale deletion, precedence, byte-freed
tracking, non-shadow dir skip, interval gating, corrupt marker
recovery.
Refs #3015 (session-file disk growth was fixed in #16286; this covers
the checkpoint side noted out-of-scope there).
The write_file guard added in #16223 used strict equality against the
internal dedup status message. In practice, the model sometimes
prepends a short note or appends a trailing comment before calling
write_file, which slipped past the strict check.
Broaden the heuristic: reject writes whose stripped content equals
the status message OR contains it and is <=2x its length. Short,
status-dominated writes are always corruption; legitimate docs that
quote the message verbatim are always much longer.
Adds two tests: one for the small-wrapper corruption shape, one
confirming large legitimate files that quote the status still write.
write_file_tool and patch_tool both call _update_read_timestamp to
refresh the staleness tracker after writing, but they never invalidate
the dedup cache entries for the written path. The dedup cache keys are
(resolved_path, offset, limit) → mtime tuples populated by read_file_tool.
On filesystems where a read and write land in the same mtime second (or
when mtime granularity is 1s), the cached and current mtime are equal,
so the dedup check incorrectly returns a 'File unchanged since last
read' stub — even though the file was just overwritten.
The agent then sees stale content (or a stale 'File not found' error)
and enters expensive error-recovery loops, burning API calls.
Fix: add _invalidate_dedup_for_path(filepath, task_id) that removes all
dedup entries whose resolved path matches the written file. Called from
_update_read_timestamp so both write_file_tool and patch_tool benefit
automatically. Scoped to the writing task_id — other tasks' caches are
not affected.
6 regression tests added covering:
- read→write→read within same mtime second (core #13144 scenario)
- invalidation across all offset/limit combinations
- isolation: writing file A does not invalidate file B's cache
- isolation: writing in task A does not invalidate task B's cache
- _invalidate_dedup_for_path safety on missing task / empty dedup
All 25 tests pass (19 existing + 6 new).
Fixes#13144
Follow-up to #15960 — the provider-active detection in tools_config.py
also read use_gateway with raw truthiness (is False, not dict.get), so
quoted 'false' caused the FAL-direct row to show wrong active status in
the hermes tools picker. Route both sites through is_truthy_value().
PR #16013 plugged the leak in `/new`, but two sibling session-boundary
resets had the same bug:
1. Inactivity / suspended-session auto-reset (top of `_handle_message`)
previously cleared only reasoning. Now drops model override and the
queued "/model switched" note as well.
2. Compression-exhaustion auto-reset now also drops the pending note
alongside the existing model/reasoning cleanup.
All three session-boundary sites now use the identical cleanup idiom.
`npm install --silent` (used by `_build_web_ui` and `_update_node_dependencies`)
silently rewrites package-lock.json on npm ≥ 10 (strips "peer": true etc.),
leaving the working tree dirty after every `hermes update`. The next update
then detects the dirty lockfile and stashes it — producing a trail of
hermes-update-autostash entries for web/package-lock.json, ui-tui/package-lock.json,
and root package-lock.json.
Switch to `npm ci` (strict, lockfile-preserving) via a new
`_run_npm_install_deterministic` helper that falls back to `npm install`
when the lockfile is missing or out of sync (WIP forks).
Verified locally: all three lockfiles stay byte-identical after the real
_build_web_ui / _update_node_dependencies run twice back-to-back. Fallback
path tested with a deliberately out-of-sync lockfile and a no-lockfile case.
Four independent session-UX bugs reported by an external user (#16294).
/save wrote hermes_conversation_<ts>.json to CWD — invisible to
'hermes sessions browse' and easy to lose. Snapshots now write under
~/.hermes/sessions/saved/ and the command prints the absolute path plus
a 'hermes --resume <id>' hint for the live DB-indexed session.
'hermes sessions browse' default --limit raised from 50 to 500. With the
old ceiling, users with moderately long histories saw only the most
recent 50 rows and assumed older sessions had been lost.
TUI session.list (`/resume` picker) switched from a hardcoded allow-list
of 13 gateway source names to a deny-list of just { 'tool' }. Sessions
tagged acp / webhook / user-defined HERMES_SESSION_SOURCE values and
any newly-added platform now surface. Default limit 20 → 200.
ollama-cloud provider setup passes force_refresh=True to
fetch_ollama_cloud_models() so a user entering their API key sees the
fresh catalog (e.g. deepseek v4 flash, kimi k2.6) immediately instead
of waiting up to an hour for the disk cache TTL to expire.
Closes#16294.
Expand the airtable skill from bare CRUD to a full Hermes-shaped
cookbook matching the linear/notion neighbors, and trim the
description to fit the 60-char system-prompt cutoff.
Hermes-specific additions:
- Explicit 'use the terminal tool with curl — not web_extract or
browser_navigate' guidance, matching the same note in linear.
- Note that AIRTABLE_API_KEY flows from ~/.hermes/.env into the
subprocess automatically via env_passthrough, so curl calls don't
need to re-export it.
- Prefer 'python3 -m json.tool' (always present) over jq (optional)
for pretty-printing, with -s on every curl to keep output clean.
- Read-before-write workflow that resolves record IDs via
filterByFormula instead of guessing.
Cookbook expansion (new vs original):
- Field-type reference table (text, select, multi-select, attachment,
linked record, user) with the exact write-shape Airtable expects.
- typecast flag for auto-coercing values / auto-creating select options.
- performUpsert PATCH for idempotent sync by merge field.
- Batch create/delete endpoints (10-record cap per call).
- Sort + fields query params with URL-encoding (%5B / %5D).
- Named-view query that applies saved filter/sort server-side.
- Full pagination loop template (while loop with offset).
- Common filterByFormula patterns (exact match, contains, AND/OR,
date comparison, NOT empty).
- Rate-limit backoff guidance (Retry-After header, per-base budget).
- Airtable error-code reference (AUTHENTICATION_REQUIRED,
INVALID_PERMISSIONS, MODEL_ID_NOT_FOUND,
INVALID_MULTIPLE_CHOICE_OPTIONS) so the agent can map failures to
user-actionable fixes instead of just retrying.
Also: description trimmed from 183 chars (truncated to 60 in system
prompt, losing 'filter/upsert/delete' trigger terms) down to 59 chars
that render whole: 'Airtable REST API via curl. Records CRUD, filters,
upserts.' Catalog row updated to match.
SKILL.md grew from 115 to 228 lines — still under the 500-line soft
cap and below the linear skill (297 lines) which serves the same
role for GraphQL.
- scripts/release.py: map sonoyuncudmr@gmail.com -> Sonoyunchu so the
check-attribution CI job and release notes credit Soynchu correctly.
- website/docs/reference/skills-catalog.md: add the airtable row to
the productivity bundled-skills table.
Adds NOTION_API_KEY, LINEAR_API_KEY, TENOR_API_KEY, and AIRTABLE_API_KEY
to OPTIONAL_ENV_VARS so:
- They persist to ~/.hermes/.env via save_env_value like every other
key Hermes knows about, instead of being ad-hoc variables the user
has to hand-edit the dotfile for.
- load_env() / reload_env() populate os.environ from .env on every
startup — the user sets the key once, skills keep working across
restarts without losing access.
- hermes setup / hermes config show surface them as known optional
vars with the correct signup URL (linear.app/settings/api,
airtable.com/create/tokens, etc.).
These four entries use category="skill" (new) rather than "tool".
tools/environments/local.py auto-adds every category=tool/messaging
entry to _HERMES_PROVIDER_ENV_BLOCKLIST, which stops env passthrough
from leaking provider credentials into the execute_code sandbox
(GHSA-rhgp-j443-p4rf). Skill API keys are the opposite case — the
point is for the agent's subprocess to see them so curl can read
Authorization headers — so they must be outside the blocklist. The
new category is inert for that check.
All four entries are advanced=True: they show up in 'hermes config'
and 'hermes status' displays, but do not nag users who have never
touched those skills during setup checklists.
E2E verified: save_env_value → reload_env → os.environ populated →
skill_view reports setup_needed=False → env_passthrough registers
the key for subprocess inheritance.
Convert the airtable skill from 'skills.config.airtable.api_key'
(config.yaml, wrong bucket for a secret) to 'prerequisites.env_vars:
[AIRTABLE_API_KEY]' (~/.hermes/.env), matching every other bundled
skill that authenticates with an API token.
Why the original shape was wrong:
- metadata.hermes.config is for non-secret skill settings (paths,
preferences) per references/skill-config-interface.md. Storing a
bearer token under skills.config.* also triggered the documented
'hermes config migrate' nag-on-every-run problem.
- The Quick Reference's 'AIRTABLE_API_KEY=...' bash line couldn't
read skills.config.airtable.api_key anyway — it's a yaml path, not
an env var.
Follow-up polish on the same pass:
- Added version/author/license frontmatter to match notion/linear.
- Added prerequisites.commands: [curl].
- Setup section now specifies the PAT format (pat...) that replaced
legacy 'key...' API keys in Feb 2024, plus the three required scopes
(data.records:read/write, schema.bases:read) and the per-base Access
list requirement.
- Clarified PATCH vs PUT and pagination (100 records/page cap).
- Swapped verification from 'hermes -q ...' (non-deterministic) to a
curl /v0/meta/bases call that returns a verifiable HTTP status code.
_web_ui_build_needed() in PR #14914 checked web_dir/"dist" as the
sentinel, but vite.config.ts sets outDir: "../hermes_cli/web_dist" so
the build output lands in hermes_cli/web_dist/, never in web/dist/.
The sentinel was therefore always missing → _web_ui_build_needed always
returned True → npm install + Vite build ran on every startup → OOM on
low-memory VPS persisted unchanged.
Fix: derive dist_dir as web_dir.parent / "hermes_cli" / "web_dist" so
the sentinel points to the actual build output directory.
Fixes#14898
When the gateway intercepts a pending /update prompt and the user sends
a recognized slash command (/new, /help, ...), the command now dispatches
normally AND the detached update subprocess is unblocked by writing a
blank .update_response. _gateway_prompt reads '' → strips → returns the
prompt's default (typically a safe 'n' / skip), so the update process
exits cleanly instead of blocking on stdin until the 30-minute watcher
timeout.
Also clears _update_prompt_pending[session_key] on this path so stray
future input for the same session isn't re-intercepted.
Extends PR #15849 with tests for the new cancel-write + a regression
test pinning the legacy behavior of unrecognized /foo slash commands
still being consumed as the response.
Slack Bolt posts are not editable like CLI spinners; medium-tier new still emitted a permanent line per tool start (issue #14663).
- Built-in slack default: off; other tier-2 platforms unchanged.
- Adjust /verbose isolation test for off to new cycle.
- Migration tests: read/write config.yaml as UTF-8 (Windows locale).
Previously, setting SLACK_BOT_TOKEN in .env would unconditionally enable
the Slack gateway adapter regardless of `slack.enabled: false` in config.yaml.
This caused spurious "SLACK_APP_TOKEN not set" errors when the token was
used only by skills (e.g. cron jobs that send Slack messages) rather than
for the Hermes messaging gateway.
Now, enabled: false in config.yaml is respected — the token is stored so
skills can still use it, but the gateway adapter is not activated.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- TestAutoMaintenance gains 3 tests: auto-prune deletes transcript files
when sessions_dir is passed, preserves them when it isn't (backward-
compat), and never touches active-session files during prune.
- FakeDB helpers in test_sessions_delete.py accept **kwargs so they
don't break when delete_session signature gains sessions_dir.
`delete_session()` and `prune_sessions()` only removed SQLite records,
leaving .json/.jsonl transcript files on disk forever. Over time this
causes unbounded disk growth (~27MB/day observed).
Changes:
- Add `_remove_session_files()` static helper that cleans up
`{session_id}.json`, `.jsonl`, and `request_dump_{session_id}_*.json`
- `delete_session()` accepts optional `sessions_dir` param and removes
files for the deleted session and its children
- `prune_sessions()` accepts optional `sessions_dir` param and removes
files for all pruned sessions after the DB transaction
- Wire up CLI `hermes sessions delete` and `hermes sessions prune` to
pass `sessions_dir`
- File cleanup is best-effort (OSError silenced) so DB operations are
never blocked by filesystem issues
- Fully backward-compatible: `sessions_dir=None` (default) preserves
existing behavior
Extends the existing channel_skill_bindings mechanism (previously
Discord-only) to Slack, so a channel or DM can auto-load one or more
skills at session start without relying on the model's skill selector
for every short reply.
Motivation: Mats's German flashcards DM pushes a cron-driven card
5x/day; he responds with one-word guesses like 'work'. Previously each
reply required the main agent to decide whether to load german-flashcards
(full opus turn just to pick a skill). With the binding configured per
Slack channel, the skill is injected at session start and grading runs
directly.
Changes:
- Extract resolve_channel_skills() from DiscordAdapter._resolve_channel_skills
into gateway.platforms.base (now shared across adapters).
- DiscordAdapter._resolve_channel_skills delegates to the shared helper
(behavior preserved — existing test suite still passes unchanged).
- SlackAdapter: resolve channel_skill_bindings on each message and attach
auto_skill to MessageEvent. gateway/run.py already handles auto-skill
injection on new sessions; this just wires Slack through it.
- gateway/config.py: accept channel_skill_bindings in slack: block of
config.yaml (was Discord-only).
- Tests: new tests/gateway/test_slack_channel_skills.py with 11 cases
covering DM/thread/parent resolution, single-vs-list skills, dedup,
malformed entries. Discord suite unchanged.
- Docs: add 'Per-Channel Skill Bindings' section to Slack user guide.
Config example:
slack:
channel_skill_bindings:
- id: "D0ATH9TQ0G6"
skills: ["german-flashcards"]
Enter while the agent is busy can now inject the typed text via /steer —
arriving at the agent after the next tool call — instead of interrupting
(current default) or queueing for the next turn.
Changes:
- cli.py: keybinding honors busy_input_mode='steer' by calling
agent.steer(text) on the UI thread (thread-safe), with automatic
fallback to 'queue' when the agent is missing, steer() is unavailable,
images are attached, or steer() rejects the payload. /busy accepts
'steer' as a fourth argument alongside queue/interrupt/status.
- gateway/run.py: busy-message handler and the PRIORITY running-agent
path both route through running_agent.steer() when the mode is 'steer',
with the same fallback-to-queue safety net. Ack wording tells users
their message was steered into the current run. Restart-drain queueing
now also activates for 'steer' so messages aren't lost across restarts.
- agent/onboarding.py: first-touch hint has a steer branch for both
CLI and gateway.
- hermes_cli/commands.py: /busy args_hint updated to include steer,
and 'steer' is registered as a subcommand (completions).
- hermes_cli/web_server.py: dashboard select widget offers steer.
- hermes_cli/config.py, cli-config.yaml.example, hermes_cli/tips.py:
inline docs updated.
- website/docs/user-guide/cli.md + messaging/index.md: documented.
- Tests: steer set/status path for /busy; onboarding hints;
_load_busy_input_mode accepts steer; busy-session ack exercises
steer success + two fallback-to-queue branches.
Requested on X by @CodingAcct.
Default is unchanged (interrupt).
MCP stdio servers are spawned via the SDK's stdio_client, which on
Linux uses start_new_session=True (setsid). When a cron job is
cancelled mid-way (timeout, agent finish, exception), the subprocess
often escapes the SDK's teardown and survives as a session leader.
Because setsid() detaches the child from the gateway's process group
/ cgroup tree, systemd does not reap it on service restart either —
so every cron tick that touches an MCP tool leaks a dangling server
process.
Fix:
* tools/mcp_tool.py — _run_stdio now wraps the whole stdio+session
context in try/finally. On any exit path (clean, exception,
cancellation), PIDs still alive are moved from the active
_stdio_pids set into a new _orphan_stdio_pids set. Orphan
detection is done via os.kill(pid, 0) — a cheap liveness probe
that never signals the target.
* tools/mcp_tool.py — _kill_orphaned_mcp_children gains an
include_active=False flag. Default behaviour now only reaps the
orphan set so concurrent sessions (other parallel cron jobs or
live user chats) are never disrupted. The existing shutdown path
passes include_active=True to keep the previous "kill everything"
semantics after the MCP loop is stopped.
* cron/scheduler.py — the cleanup hook is moved from run_job()'s
finally (which would race with parallel siblings after #13021)
into tick() after the ThreadPoolExecutor has joined every future.
At that point there are no in-flight sessions from this tick, so
sweeping the orphan set is always safe.
Net effect: zero regression for healthy sessions, and orphan MCP
servers no longer accumulate between gateway restarts.
Made-with: Cursor
Multiple overlapping Slack attachment improvements:
1. Upload retry with backoff on transient errors (429, 5xx, connection
reset, rate_limited, service unavailable). New _is_retryable_upload_error
helper covers three upload paths: _upload_file, send_video,
send_document. Up to 3 attempts with 1.5s * attempt backoff.
2. Thread participation tracking: successful file uploads now add the
thread_ts to _bot_message_ts, mirroring how text replies are tracked.
This lets follow-up thread messages auto-trigger the bot (same
engagement rules as replied threads).
3. Thread metadata preservation in the image redirect-guard fallback
(send_image → send text fallback) and in two gateway.run.py send
paths (image + document fallback calls).
4. HTML response rejection in _download_slack_file_bytes. Parallels
the existing check in _download_slack_file. Guards against Slack
returning a sign-in / redirect page as document bytes when scopes
are missing, so the agent doesn't get HTML-as-a-PDF.
5. File lifecycle event acks (file_shared / file_created / file_change).
These events arrive around snippet uploads. Acking them silences the
slack_bolt 'Unhandled request' 404 warnings without changing behavior.
6. Post-loop message type classification so a mixed image+document upload
classifies as PHOTO (or VOICE if no image), falling back to DOCUMENT.
Previously, the per-file classification in the inbound loop could be
overwritten unpredictably.
7. Expanded text-inject whitelist in inbound document handling to cover
.csv, .json, .xml, .yaml, .yml, .toml, .ini, .cfg (up to 100KB) so
snippets and config files are directly visible to the agent, not just
cached as opaque uploads. Paired with new MIME entries in
SUPPORTED_DOCUMENT_TYPES in base.py.
Squashed from two commits in #11819 so the single commit carries the
contributor's GitHub attribution (the original commits were authored
under a local dev hostname).
- stringWidth: true LRU on cache hit (touch-on-read via delete+set) so
hot strings stay resident under long sessions; was insertion-order
FIFO before
- virtualHeights: include todos, panel sections, and intro version in
messageHeightKey so height-cache reuse correctly invalidates when
todo content / panel sections change
- virtualHeights: estimate trail+todos rows at todos.length+2 (or 2
collapsed) instead of the generic ~1-line fallback, so initial
virtualization offsets are closer to reality
- useInputHandlers: clearTimeout on unmount for scrollIdleTimer so
pending relaxStreaming() never fires after teardown
- render-node-to-output: drop unused declined.noHint counter from
scrollFastPathStats; it was always 0 (the "hint missing" branch is
outside the diagnostics block)
- perfPane / hermes-ink.d.ts: follow the noHint removal
- wheelAccel: replace ~/claude-code path comment with generic
attribution that doesn't reference a developer-local checkout
TodoPanel now renders as a child of the most recent user message's
virtualized row container, so it visually belongs to that prompt and
follows it during scroll. Falls back gracefully when no user message
exists yet (panel just doesn't render).
Adds an `evictInkCaches(level)` API that prunes the four hot module-level
caches (`widthCache`, `wrapCache`, `sliceCache`, `lineWidthCache`) with
either a half-keep LRU pass or a full clear. Wired into:
- memoryMonitor: half-prune on 'high', full drop on 'critical', before
the heap dump / auto-restart path. Gives long sessions a shot at
recovering RSS instead of hard-exiting.
- useSessionLifecycle.resetSession: half-prune so a /new session starts
with a half-warm pool and the prior session can resume cheaply.
Also: lineWidthCache now uses LRU half-eviction on overflow instead of a
full `cache.clear()`, matching the other three caches.
Comparison vs claude-code: both forks now share the same `prevScreen`
blit + dirty-cascade machinery in render-node-to-output. Their smoothness
came from sibling-memo discipline (every chrome pane memo'd so dirty
cascade doesn't disable transcript blit) — already in place in our
appLayout.tsx (TranscriptPane / ComposerPane / StatusRulePane all memo'd).
Alt-screen is not the cause; both use it. The remaining gap was per-row
CPU on width/wrap/slice, which the previous commit closed.
CPU profile (Apr 2026, real-user scroll on 11k-line session) showed three
hot loops in the per-frame render path:
Output.get() per-frame walk: 24% total
└─ sliceAnsi(line, from, to) per write: 18% total
stringWidth(line) chain (cached + JS): 14% total
All three were re-doing identical work every frame: same string → same
clipped slice → same width.
Fixes:
1. Memoize stringWidth (8k-entry LRU) for non-ASCII strings; ASCII fast-path
skips the cache (inline scan beats Map.get for short ASCII, the >90%
case). String.charCodeAt scan up to 64 chars is cheaper than the regex
fallback.
2. Memoize wrapText (4k-entry LRU keyed by maxWidth|wrapType|text) — wrapAnsi
is pure and the same content reflows identically every frame.
3. Memoize sliceAnsi (4k-entry LRU keyed by start|end|str) for the
end-defined hot path used by Output.get().
4. Skip the slice entirely in Output.get() when the line already fits the
clip box (startsBefore=false && endsAfter=false). Most transcript lines
never exceed their container width, and tokenizing them just to slice
(line, 0, width) was pure overhead. This single fast-path drops
sliceAnsi from 18% → ~0% in the profile.
Also tighten virtualization constants (MAX_MOUNTED 260→120, OVERSCAN 40→20,
SLIDE_STEP 25→12) and cap historical-message render at 800 chars / 16
lines via HISTORY_RENDER_MAX_*; messages inside the FULL_RENDER_TAIL_ITEMS
window still render in full so reading-zone behavior is unchanged.
Validation, real-user CPU profile, page-up scroll on 11k-line session:
Output.get() self-time: 24% → 0.3%
sliceAnsi total: 18% → not in top 25
stringWidth family: 14% → ~3%
idle: 60.7% → 77.3%
Frame timings (synthetic page-up profile harness):
dur p95: ~10ms → 4.87ms
dur p99: 25ms+ → 12.80ms
yoga p99: ~20ms → 1.87ms
The remaining CPU in the profile is Yoga layoutNode + React commit,
which is the irreducible work for this UI tree size.
Ports openclaw/openclaw#72038 to hermes-agent.
Telegram's `editMessageText` preserves the original message timestamp,
so a long-running streamed reply (reasoning models that take 60+ seconds
to finish) would keep the first-token timestamp even after completion.
Users can't tell how long a task actually took.
When a preview message has been visible for >= 60s (configurable via
`streaming.fresh_final_after_seconds`), finalize by sending a fresh
message instead of editing in place, then best-effort delete the stale
preview. Short previews still edit in place (the existing fast path).
Implementation notes adapted from OpenClaw's TypeScript original:
- `StreamConsumerConfig` gains `fresh_final_after_seconds` (default 0 =
legacy edit-in-place). Gateway-level `StreamingConfig` defaults to 60.
- `GatewayStreamConsumer` tracks `_message_created_ts` at first-send and
checks it in `_send_or_edit` on `finalize=True`. New helpers
`_should_send_fresh_final` + `_try_fresh_final`.
- `BasePlatformAdapter` gains optional `delete_message(chat_id, message_id)`
returning False by default. `TelegramAdapter` implements it via
`_bot.delete_message`.
- `gateway/run.py` only enables fresh-final for `Platform.TELEGRAM`;
other platforms ignore the setting (they don't have the stale-edit
timestamp problem or edit-then-read works cheaply).
- Fallback to normal edit on any fresh-send failure — no user-visible
regression if Telegram rate-limits a send or the message is gone.
Tests: 15 new cases in tests/gateway/test_stream_consumer_fresh_final.py
covering short/long previews, config plumbing, delete-support absent,
send-failure fallback, __no_edit__ sentinel safety, and StreamingConfig
round-trip.
Co-authored-by: Hermes Agent <agent@nousresearch.com>
Adds a corner-overlay FPS readout gated on HERMES_TUI_FPS, fed by
ink's onFrame callback (so it's the REAL render rate, not a timer).
Displays fps, last-frame duration, and total frame count, colored by
threshold (green ≥50, yellow ≥30, red below).
Implementation:
* lib/fpsStore.ts — nanostore atom updated from a trackFrame()
sink. Ring buffer of last 30 frame timestamps; fps = 29/elapsed.
trackFrame is undefined when SHOW_FPS is off so ink's onFrame
short-circuits at the optional chain.
* components/fpsOverlay.tsx — tiny <Text> subscriber; returns null
when SHOW_FPS is off (React skips the subtree entirely).
* entry.tsx — composes onFrame from logFrameEvent (dev-perf) and
trackFrame (fps) so both flags can coexist. When both are off,
onFrame is undefined and ink never attaches the handler.
* appLayout.tsx — mounts the overlay as a flex-shrink=0 right-
aligned Box below the composer, conditional on SHOW_FPS.
Usage:
HERMES_TUI_FPS=1 hermes --tui
# bottom right: " 62.3fps · 0.8ms · #1234" (green/yellow/red)
Intended as a user-facing diagnostic during the scroll-perf tuning
pass — watch the counter drop while holding PageUp to see where
frames go silent, without having to run scripts/profile-tui.py in a
side terminal.
126 files post-compile with React Compiler; 352 tests still pass.
Replaces the static WHEEL_SCROLL_STEP=1 multiplier on wheel events
with an adaptive accel state machine that infers user intent from
inter-event timing.
Algorithm ported straight from claude-code's
src/components/ScrollKeybindingHandler.tsx. All tuning constants,
the native/xterm.js path split, the encoder-bounce detection, the
trackpad-burst signature → all theirs. This file is a mechanical
port into our module structure.
What it does:
precision click (>500ms gap) 1 row/event (deliberate scan)
sustained mouse (40-200ms) 2-6 rows (decay curve)
detected wheel bounce ramps to 15 (sticky wheel-mode)
trackpad flick (5+ <5ms) 1 row/event (burst detect)
direction reversal reset to base
Two implementation paths:
* native terminals (ghostty, iTerm2, Kitty, WezTerm) — linear
window-ramp + optional wheel-mode curve triggered by detected
encoder bounce. SGR proportional reporting handled via the
burst-count guard.
* xterm.js (VS Code / Cursor / browser terminals) — pure
exponential-decay curve with fractional carry. Events arrive
1-per-notch with no pre-amplification, so the curve is more
aggressive.
Selected at construction via isXtermJs() from @hermes/ink (now
exported). Per-user tune via HERMES_TUI_SCROLL_SPEED (alias
CLAUDE_CODE_SCROLL_SPEED for portability).
13 unit tests covering direction flip/bounce/reversal, idle
disengage, trackpad-burst disengage, frac invariants, and the
native vs xterm.js branches.
Profiled under --rate 30 (stress test) and --rate 10 (realistic
sustained scroll): accel ramps to cap=6 at 30Hz burst, decays to
1-3 rows at sparse 10Hz clicks. Perf is comparable to baseline
because accel IS multiplying step — the win is perceptual (fast
flicks cover distance, slow clicks keep precision), not raw fps.
Companion to the earlier WHEEL_SCROLL_STEP=1 change: that set the
base; this modulates around it.
Was user-local in ~/.hermes/skills/. Ported into skills/software-development/
so other Hermes users get it and so the related_skills links from
node-inspect-debugger and python-debugpy resolve in-repo.
Frontmatter upgraded to match repo convention (version/author/license/
metadata.hermes.{tags,related_skills}, description rewritten as "Use when ...").
Body expanded with debugging-tactics section pointing at the two new
debugger skills, and additional common-issues / pitfalls entries.
Adds a gate so we can A/B test whether bypassing the alt-screen +
viewport constraint lets the terminal's native scrollback beat our
virtualization on scroll perf.
Result: definitively NO. Inline mode is 40x worse on every metric
that moves, because AlternateScreen is what constrains the ScrollBox
to the viewport height. Without it, the ScrollBox grows to contain
every child of the transcript and every frame re-renders all 1100
messages.
Profile under hold-wheel_up (1106-msg session, 30Hz for 6s):
metric fullscreen inline delta
patches_total 28,864 1,111,574 +3751%
writeBytes_total 42 KB 1.6 MB +3881%
fps_throughput 15.8 fps 1.75 fps -89%
frames 179 18 -90%
gap_p50_ms 17 (~60fps) 726 (~1fps) +4170%
yoga_p99 34 ms 405 ms +1083%
renderer_p99 14 ms 169 ms +1062%
flickers 0 5 offscreen —
This is actually the cleanest data we've gotten so far:
* AlternateScreen is LOAD-BEARING for perf — its viewport height
constraint is what lets useVirtualHistory's culling work. No
constraint → ScrollBox grows unbounded → every fiber mounts.
* The outer terminal (Cursor's xterm.js) parsed 1.6 MB of ANSI in
under 10 seconds with drain p99 = 8.83 ms and 0 backpressure
frames. Our terminal-write hypothesis from last session was
wrong: the bottleneck is React + Yoga, not the wire.
* Doing proper inline mode (non-virtualized transcript in
scrollback, composer pinned below) is not a flag flip — it's a
different UI architecture. Leaving this flag in so anyone
re-running the experiment gets the same numbers, but not
building the architecture until we're sure the perf win is
worth the UX loss (it probably isn't — the fullscreen + virt
path is the one we should optimize, not replace).
Keeping the flag as an experiment gate. Flip HERMES_TUI_INLINE=1
and run scripts/profile-tui.py --compare to reproduce.
Two new skills under skills/software-development/ for real breakpoint-driven
debugging from the terminal:
- node-inspect-debugger: node --inspect / --inspect-brk, node inspect REPL,
CDP scripting via chrome-remote-interface, attaching to running Node
processes (SIGUSR1), ui-tui-specific recipes, Vitest under debugger,
CPU profiles + heap snapshots.
- python-debugpy: pdb quick reference, breakpoint() workflow, pytest --pdb
(with xdist caveat for scripts/run_tests.sh), post-mortem, debugpy for
remote/attach, remote-pdb as the agent-friendly alternative to DAP,
recipes for tui_gateway/_SlashWorker/subprocess debugging.
Before: change code → build → run profile → manually compare to
mental model of last run. After: `--loop` watches ui-tui/src and
packages/hermes-ink/src for .ts(x) changes, rebuilds on change,
re-runs the same scenario, prints a side-by-side A/B diff against
the previous iteration — so each edit's impact is quantified
instantly. Ctrl+C to stop.
Also added:
--save LABEL saves metrics snapshot to /tmp/perf-<LABEL>.json
--compare LABEL diffs the current run vs that snapshot
--extra-flag X pass-through to node dist/entry.js (prepping for
--no-fullscreen below)
key_metrics() flattens a full run into scalar numbers across
frames, React commits, and per-phase timings. format_diff() prints
a table with ↑/↓ markers denoting regressions vs improvements based
on whether the metric is lower-is-better (p99, max, patches, drain)
or higher-is-better (fps, gaps_under_16ms).
Run-to-run noise on static code is ~5-15% on most metrics — big
signal (>30% change on renderer_p99 / fps) cuts through cleanly.
Useful both for validating a single fix and for detecting subtle
regressions during the wheel-accel port.
Usage during the next perf session:
# one-shot with a baseline for later comparison
scripts/profile-tui.py --seconds 6 --hold wheel_up --save pre-accel
# after porting the wheel handler
scripts/profile-tui.py --seconds 6 --hold wheel_up --compare pre-accel
# continuous iteration
scripts/profile-tui.py --seconds 6 --hold wheel_up --loop
Adds four fields to FrameEvent.phases and the matching profile
summary:
optimizedPatches post-optimize patch count (what's actually
written to stdout; the .patches field is
pre-optimize)
writeBytes UTF-8 byte count of the write this frame
backpressure true when Node's stdout.write returned false
(Writable buffer full — outer terminal can't
keep up)
prevFrameDrainMs end-to-end drain time of the PREVIOUS frame's
write, captured from stdout.write's 2-arg
callback. Reported on the next frame so the
measurement reflects "time until OS flushed
the bytes to the terminal fd", not "time until
queued in Node".
writeDiffToTerminal() now returns { bytes, backpressure } and
accepts an optional onDrain callback. Only attached on TTY with
diff; piped/non-TTY stdout bypasses flow control so the callback
would fire synchronously anyway.
Initial measurements under hold-wheel_up against 1106-msg session
(30Hz for 6s):
patches total 28,888
optimized total 16,700 (ratio 0.58 — optimizer cuts ~42%)
writeBytes 42 KB / 10s = 4.2 KB/s throughput
drainMs p50 0.14 ms terminal accepts bytes instantly
drainMs p99 0.85 ms
backpressure 0% of frames
This rules out the terminal-parse hypothesis — Cursor's xterm.js
drains our output in sub-millisecond time at only 4 KB/s. The
remaining lag has to be in the render pipeline, not the wire.
Profile output now includes the bytes+drain+backpressure lines to
keep this visible on every subsequent iteration.
Profiled with scripts/profile-tui.py under hold-PageUp + hold-wheel.
The placeholder → microtask-upgrade pattern did not reduce renderer
p99 (63ms → 63ms) or max (96ms → 142ms, slightly worse). Each fresh
row still pays the Md cost — just on a follow-up commit instead of
inline — and the follow-up commit shows up as a second heavy frame
a few ms later.
The real bottlenecks turned out to be:
1. wheel step too large (fixed in 7ca16eea)
2. outer terminal ANSI parse throughput (diagnosing next)
3. React commit frequency during hold-scroll (needs coalescing)
None of which DeferredMd addresses. Clearing the complexity so the
next experiments land on a simpler substrate.
User observation: "it doesn't scroll line by line/row by row."
Was right. Two places hardcoded big deltas:
1. WHEEL_SCROLL_STEP = 6 (config/limits.ts)
Each wheel event scrolled 6 rows. A mechanical wheel notch emits
3-5 events → 18-30 rows per click, which visually teleports past
content instead of smooth-scrolling it. Drop to 1. Trackpads
emit 50-100 events per flick — at step=1 that's still a fast flick
(a whole viewport in one flick) but each intermediate frame is
visible. Porting claude-code's wheel accel state machine is the
right next step if this feels sluggish on precision scrolls.
2. pageUp/pageDown = viewport - 2 (useInputHandlers.ts)
Full-viewport jumps replace the entire screen — no visual
continuity, can't scan content — AND land right at Ink's fast-path
threshold (`delta < innerHeight`), which disqualifies the DECSTBM
blit on every press. Half-viewport keeps 50% continuity AND
drops well under the threshold. Two presses still cover the same
total distance.
Profiled against the 1106-msg session, holding the key at 30Hz for
6s:
wheel_up (step 6 → 1):
frames 142 → 163 (+15%)
throughput 10.7 → 15.8 fps (+48%)
patches tot 53018→ 36562 (-31%)
gap p50 5ms → 16ms (actual rendering ~60fps now)
<16ms frames 93 → 76
16-33ms 82 → 76
hitches 3 → 1
pageUp (viewport-2 → viewport/2):
throughput 10.7 → 9.5 fps (same ballpark — smaller delta × same
event rate = less total scroll)
Ink's proportional drain caps at `innerHeight - 1` per frame to keep
the DECSTBM fast path firing. With these smaller deltas every event
comfortably fits under that cap, so fast-path hit rate goes up and
patch volume per frame drops — the measured 31% reduction in total
patches-sent correlates with users perceiving smoother scrolling
because the outer terminal (VS Code / xterm.js / tmux) isn't drowning
in ANSI between paints.
Tests/type-check/build clean; 352 tests pass.
Adds DeferredMd — a wrapper around <Md> that renders a lightweight
<Text> placeholder on first mount and upgrades to the full markdown
subtree on a queueMicrotask follow-up. Rationale: fresh MessageLine
mounts during PageUp hold run our markdown tokenizer + syntax
highlighter synchronously, producing the 63-112ms renderer spikes
profiled earlier. A plain <Text> placeholder only needs Yoga to wrap
the pre-stripped string (no tokenizer, no highlight), then the Md
subtree builds in a follow-up React commit.
Upgrade cache: once a (theme, compact, text) tuple has been upgraded,
a WeakMap-keyed Set remembers it so remounts (scroll-out then
scroll-back) mount straight into <Md> — no placeholder round-trip.
WeakMap on theme means palette swaps re-upgrade naturally.
Honesty note: profiling under hold-PageUp showed this didn't reduce
renderer p99 measurably — the upgrade commit just pays the Md cost on
a follow-up frame instead of inline. The bigger bottleneck turned out
to be React commit frequency (3.5 commits/sec during 30Hz scroll
input, with 200ms+ silent gaps between commits dominating perceived
FPS), which this change doesn't address. Keeping the deferred path
anyway because:
1. It's correct and tested — no regressions across 352 tests
2. Defensive for pathological fresh-mount cases (giant code blocks,
wide tables) that aren't in the current profile fixture
3. Pairs naturally with useVirtualHistory's useDeferredValue to keep
React's concurrent scheduler able to interrupt upgrade commits
If the follow-up perf investigation (terminal write throughput / patch
volume / commit frequency) shows DeferredMd is net-neutral-or-worse in
practice, this can be reverted with a one-line swap back to <Md> in
messageLine.tsx:115.
Companion to the streaming 2-column fix in 7242361a — these two
touched messageLine.tsx together so they land as a pair.
StreamingMd returned <><Md/><Md/></> — a bare Fragment with two <Md>
children. Each <Md> returns a <Box flexDirection="column">, but its
parent in messageLine.tsx (line 169) is `<Box width={...}>` with no
flexDirection, which Ink defaults to 'row'. So during streaming the
two column boxes rendered side-by-side, producing the visible "tokens
jumble into two columns until it fixes itself" bug — the "fix" was
message.complete flipping isStreaming→false, which swaps the
StreamingMd subtree for a single DeferredMd/Md child (no siblings → row
direction is harmless).
Wrap the two <Md> siblings in a flexDirection="column" Box so they
stack. Localized fix so the non-streaming path (single-child, works
fine in a row parent) is untouched.
Reported by user:
> "tokens streaming... going into 2 columns randomly and jumbling
> together until it fixes itself"
No test changes — findStableBoundary tests still pass (the layout
change is parent-structural, not in the boundary logic). Build clean,
tsc clean, 352 tests pass.
Adds scrollFastPathStats counters to render-node-to-output.ts: captures
every time a ScrollBox's DECSTBM scroll hint is generated, records
whether the fast path took it (blit+shift from prevScreen) or declined,
and why. Exposed through hermes-ink's public exports and snapshotted on
every FrameEvent so the profiler harness can correlate decline reasons
with the actual patch/renderer cost per frame.
This is pure observation — no behaviour change. Preparing for the
virtual-history rewrite: the hypothesis was that our topSpacer/
bottomSpacer scheme disqualifies every scroll via heightDelta
mismatch, but the data shows the fast path is actually taken on most
scrolls (19/23 over a 6s PageUp hold through 1100 messages) — the
remaining steady-state renderer cost is Yoga tree traversal, not
the per-frame full redraw I initially suspected.
Declines that do happen correlate with React commits that changed the
mounted range mid-scroll (heightDelta=±3 to ±35). Those are the rarer
cases the virtualization rewrite still needs to address.
No test diffs — instrumentation-only. Build verified: `tsc --noEmit`
plus the full `npm run build` compiler post-pass pass cleanly.
Extends HERMES_DEV_PERF to capture the complete render pipeline, not
just React commits. Adds scripts/profile-tui.py to drive repeatable
hold-PageUp stress tests against a real long session.
perfPane.tsx:
Wires ink's onFrame callback (already plumbed through the fork) into
the same perf.log as the React.Profiler samples. Captures per-phase
timing (yoga calculateLayout, renderNodeToOutput, screen diff, patch
optimize, stdout write) plus yoga counters (visited/measured/cache-
Hits/live) and patch counts per frame. Events are tagged
{src: 'react'|'frame'} so jq can split them. logFrameEvent is
undefined when HERMES_DEV_PERF is unset, so ink doesn't even attach
the callback.
entry.tsx:
Passes logFrameEvent into render().
types/hermes-ink.d.ts:
Declares FrameEvent + onFrame on RenderOptions so the ui-tui side
type-checks against the plumbed-through ink option.
scripts/profile-tui.py:
New harness. Launches the built TUI under a PTY with the longest
session in state.db resumed, holds PageUp/PageDown/etc at a
configurable Hz for N seconds, then parses perf.log and prints
per-phase p50/p95/p99/max plus yoga-counter summaries. Zero deps
beyond stdlib. Exit 2 if nothing was captured (wiring broken).
Initial findings (1106-msg session, 6s PageUp hold at 30Hz):
- Steady state: 10 fps; renderer phase p99=63ms, write p99=0.2ms
- 4/107 heavy frames (>=16ms), all dominated by renderNodeToOutput
- One pathological 97ms frame with yoga measuring 70,415 text cells
and Yoga visiting 225k nodes — the cold-unmeasured-region hit
- Ink's scroll fast-path (DECSTBM blit from prevScreen) is
disqualified because our spacer-based virtual history doesn't
keep heightDelta in sync with scroll.delta, so every PageUp step
falls through to a full 2000-4800 patch re-render instead of ~40
Split in-flight assistant text at the last stable block boundary so only
the unclosed tail re-tokenizes per stream delta. Previously the full
text was rendered as plain <Text> during streaming and only flipped to
<Md> at message.complete — cheap per delta but loses live markdown
formatting.
New StreamingMd component holds a monotonically-growing stablePrefix
in a ref (idempotent under StrictMode double-render), renders it as
one <Md> that memoizes across deltas, and renders the unstable suffix
as a second <Md> that re-parses on each delta. Cost per delta drops
from O(total length) to O(unstable length).
findStableBoundary walks back to the last "\n\n" outside an open
fenced code block — splitting inside an open fence would orphan the
opener and break highlighting in the prefix.
Adapted from claude-code's src/components/Markdown.tsx:186 but built
on our line-based tokenizer instead of marked.lexer. 9 new tests cover
fence balance, boundary walk, and empty input.
Part of the --tui perf audit (see audit #7).
Slack's modern composer sends messages with a 'blocks' array that
contains rich_text elements. When a user forwards or quotes another
message, the quoted content shows up in the rich_text_quote children
of that array — and is NOT included in the plain 'text' field. The
agent saw only the lossy plain text and was blind to forwarded /
quoted content. Same story for link unfurl previews (Notion, docs,
GitHub, etc.) which Slack puts in the 'attachments' array.
Two fixes in the inbound handler:
1. _extract_text_from_slack_blocks walks rich_text / rich_text_quote /
rich_text_list / rich_text_preformatted trees and renders readable
text ('> quoted', '• bullet', code fences), dedupes against the
plain text field, and appends the extracted content so the agent
sees everything.
2. Link unfurl / attachment preview extraction reads title, url,
body, and footer from the 'attachments' array and appends a
'📎 [title](url)\n body\n _footer_' section per preview.
Skips is_msg_unfurl to avoid echoing our own Slack replies back.
Routing is careful not to trust augmented text: mention gating
(is_mentioned) and slash-command detection both run against the
original 'text' field, so forwarded content containing '<@bot>' or
'/deploy' in a quote can't trick the bot into responding in a
channel it shouldn't or classifying a normal message as a command.
Adjustment from original PR: dropped _serialize_slack_blocks_for_agent,
which inlined a redacted JSON dump of non-rich_text blocks (section,
accessory, actions, etc.) — the agent would see the raw Block Kit
structure for UI-heavy alerts. It added up to 6000 characters to the
prompt context on every qualifying message with no opt-out. The
rich_text extraction and attachment unfurls cover the common bug-fix
case (quoted/forwarded content + link previews) without the prefill
tax. If a user needs block inspection later, it can return as a
config opt-in.
Also updates the Slack platform notes in session.py to accurately
describe what the gateway inlines.
After #14798 made cron honor per-platform `hermes tools` config, the
`_DEFAULT_OFF_TOOLSETS` filter silently stripped `homeassistant` from
cron jobs for users who'd been relying on the previous blanket toolset.
Norbert's HA cron reports regressed as a result.
The HA toolset is already runtime-gated by its `check_fn` (requires
HASS_TOKEN to register any tools). When HASS_TOKEN is set the user has
explicitly opted in — `_DEFAULT_OFF_TOOLSETS` adds nothing in that case,
so stop double-gating and restore HA for cron / cli / other platforms
without an explicit saved toolset list.
moa and rl stay off by default (original #14798 goal preserved).
Fixes HA cron regression reported by Norbert.
HindsightEmbedded.close() delegates to its sync client.close(). When Hermes
created/used that client on the shared async loop, closing it from the main
thread raises 'attached to a different loop' before aiohttp releases the
session — so the ClientSession / TCPConnector leak past provider teardown.
Close the embedded inner async client on the shared loop first via
_run_sync(inner_client.aclose()), then let the wrapper's sync close()
do its daemon/UI bookkeeping.
Salvage of #14605: test placement rebased — appended TestShutdown class
after TestSharedEventLoopLifecycle (which landed on main after the PR was
written). Original author attribution preserved.
Translate Slack attachment failures into actionable user-facing notices
instead of generic download errors. When a scope/auth/permission issue
breaks attachment processing, the user sees:
[Slack attachment notice]
- Slack attachment access failed for photo.jpg. Missing scope:
files:read. Update the Slack app scopes/settings and reinstall
the app to the workspace.
Two helpers do the translation:
_describe_slack_api_error — handles SlackApiError responses
(missing_scope, invalid_auth, file_not_found, access_denied, etc.)
_describe_slack_download_failure — handles httpx.HTTPStatusError
(401/403/404) and Slack-returns-HTML-sign-in fallbacks
Wired into three existing call sites:
- the Slack Connect files.info path (PR #11111) so scope errors
surface instead of being logged as generic "files.info failed"
- the image, audio, and document download paths so 401/403 and
HTML-body responses translate into actionable notices
Adjustment from original PR: dropped _probe_slack_file_access_issue,
the proactive pre-download files.info probe. It added one extra
Slack API call per attachment even on healthy ones, and overlapped
with the existing files.info call from PR #11111. The post-failure
translation path covers the same user-facing diagnostic value
without the per-message tax.
Also documents files:read scope more prominently in the Slack setup
guide and troubleshooting table.
Contributed back from https://github.com/xinbenlv/zn-hermes-agent.
Closes#7015.
Co-authored-by: xinbenlv <zzn+pa@zzn.im>
Background review fork now inherits session_id, credential_pool, and
status_callback from the parent (added in #16099 after this PR was
written). Extend the bare-agent helper so the regression test keeps
reaching the cleanup assertions instead of failing in the runtime
resolver.
Signed-off-by: Teknium <8425893+teknium1@users.noreply.github.com>
Temporary background review agents can initialize Hindsight-backed memory clients, but close() alone skips provider teardown. Shut the memory provider down before closing so aiohttp sessions do not leak at process exit.
Made-with: Cursor
Slack Connect channels return file objects with file_access="check_file_info"
and no url_private_download field (see
https://docs.slack.dev/reference/objects/file-object/#slack_connect_files).
These stub objects must be resolved via files.info before download can
proceed. Without this the agent silently skips attachments posted in
Slack Connect channels.
Call files.info on every file whose file_access is check_file_info,
replace the stub with the full file object, and let the existing
download path continue. Warn and skip on files.info failures.
Closes#11095.
The Slack thread-context fetcher used to drop every message with a
bot_id, which silently erased the thread parent whenever a cron job (or
any other bot) had posted it. As a result, replies to a cron-posted
summary lost all context and the agent answered as if from a blank
thread.
Changes:
1. gateway/platforms/slack.py::_fetch_thread_context
- Keep the thread parent even when it was posted by a bot
(e.g. cron summaries, third-party integrations).
- Only skip *our own* prior bot replies to avoid circular context,
matching the per-workspace bot user id via _team_bot_user_ids so
multi-workspace deployments stay correct.
- Keep non-self bot children (useful third-party context).
2. gateway/platforms/slack.py::_handle_slack_message
- Populate MessageEvent.reply_to_text for thread replies (parity
with Telegram/Discord/Feishu/WeCom). gateway.run uses this field
to inject a [Replying to: "..."] prefix when the parent is not
already in the session history, which is exactly the scenario
triggered by cron-generated thread parents.
- New helper _fetch_thread_parent_text reuses the existing thread-
context cache (and its 60s TTL) to avoid duplicate
conversations.replies calls; falls back to a cheap limit=1 fetch
when the cache is cold.
Tests:
- Updated TestSlackThreadContext::test_skips_bot_messages to reflect
the new behaviour (self-bot child dropped, third-party bot kept).
- Added:
* test_fetch_thread_context_includes_bot_parent
* test_fetch_thread_context_excludes_self_bot_replies
* test_fetch_thread_context_multi_workspace
* test_fetch_thread_context_current_ts_excluded (regression guard)
* test_fetch_thread_parent_text_from_cache
* test_slack_reply_to_text_set_on_thread_reply
* test_slack_reply_to_text_none_for_top_level_message
Full Slack suite: 176 passed (was 169).
Slack's chat.postMessage API rejects user IDs (U...) and workspace
IDs (W...) — they are not valid conversation IDs. Posting to them
fails because the API requires a channel ID (C/G/D). To DM a user,
the sender must first call conversations.open to obtain a D... ID.
Tighten _SLACK_TARGET_RE from [CGDUW] to [CGD] so the send path rejects
U/W values as explicit targets and instead falls through to channel-
name resolution (where they'll fail with a clear 'could not resolve'
error rather than silently getting stuck in a retry loop on the API).
Flip the corresponding regression test to assert U/W values are not
explicit. Matches the narrower regex briandevans proposed in #15939.
Co-authored-by: briandevans <brian@bde.io>
send_message(target='slack:<channel_id>') failed with "Could not
resolve" because _parse_target_ref had no Slack branch — Slack's
uppercase alphanumeric IDs fell through to channel-name resolution,
which only matched by name. As a fallback, the agent would retry with
bare target='slack' and post to the home channel instead.
Three fixes:
- _parse_target_ref recognizes Slack IDs (C/G/D/U/W prefix) as
explicit targets so the name-resolver is bypassed entirely.
- resolve_channel_name tries a case-sensitive raw-ID match before
the existing name match, so any platform's IDs resolve cleanly.
- _build_slack now actually calls users.conversations against each
workspace's AsyncWebClient (paginated), instead of only returning
session-history entries. This populates the directory with public
and private channels the bot has joined, so action='list' shows
them and they can also be addressed by name. Errors from one
workspace don't block others.
build_channel_directory becomes async (Slack web calls require it).
The two async-context callers in gateway/run.py are awaited; the
cron ticker thread call bridges via asyncio.run_coroutine_threadsafe.
Slack bot needs channels:read and groups:read scopes for full
enumeration; missing scopes degrade gracefully per-workspace.
addressing #15927
Removes deepseek/deepseek-v4-pro and deepseek/deepseek-v4-flash from
OPENROUTER_MODELS and _PROVIDER_MODELS['nous'], then regenerates
website/static/api/model-catalog.json so the hosted picker JSON drops
them too. Direct-API deepseek provider support is unchanged.
load_gateway_config() has a side effect: when config.yaml contains
platform-gating keys (slack.require_mention, slack.strict_mention,
slack.free_response_channels, slack.allow_bots, slack.reactions, plus
analogous keys for discord/telegram/whatsapp/dingtalk/matrix), it calls
os.environ[KEY] = ... to bridge them to env-var form.
monkeypatch.delenv doesn't track direct os.environ mutations made
inside the test body, so tests that call load_gateway_config() leak
those env vars into later tests on the same xdist worker. The failure
mode is flaky seed-dependent: test_top_level_message_requires_mention_
even_with_session (and siblings in TestThreadReplyHandling) pass when
SLACK_REQUIRE_MENTION is unset but fail when a leaked value of 'false'
is present.
Add the gating env vars to _HERMES_BEHAVIORAL_VARS so the hermetic
autouse fixture blanks them on every test setup, closing the leak
regardless of which test sets them.
Extends the strict_mention feature so an @mention in strict mode no
longer persistently tags the thread as 'mentioned'. Without this, the
thread's first mention would permanently auto-trigger the bot on every
subsequent message — which is exactly what strict_mention is designed
to prevent. Closes the agent-to-agent ack loop hole hhhonzik identified
in #14117.
Co-authored-by: hhhonzik <me@janstepanovsky.cz>
Adds a strict_mention config option that, when enabled, requires an
explicit @-mention on every message in channel threads. Disables the
'once mentioned, forever in the thread' and session-presence auto-triggers.
- New _slack_strict_mention() helper (config.extra + SLACK_STRICT_MENTION env)
- Bridged top-level slack.strict_mention yaml to SLACK_STRICT_MENTION env,
matching require_mention/allow_bots bridging
- Unit tests for the helper + config bridge
* fix(install): add /usr/local/bin PATH guard for RHEL root non-login shells
The FHS-layout branch assumed /usr/local/bin is on PATH for every
standard shell. That holds for login shells (via /etc/profile's
pathmunge) but breaks on RHEL/CentOS/Rocky/Alma 8+ root in non-login
interactive shells (su, sudo -s, tmux panes, some web terminals) —
/etc/bashrc does not add /usr/local/bin and /root/.bash_profile
doesn't either. Result: hermes command links to /usr/local/bin/hermes
but the user has to type the absolute path each time.
Probe a fresh 'bash -i -c' (non-login interactive, matching the user
scenario) after symlinking. If hermes isn't resolvable, append an
idempotent PATH guard to /root/.bashrc and /root/.bash_profile, same
grep pattern already used by the ~/.local/bin branch below. No change
on distros where /usr/local/bin is already inherited.
* fix(update): repair RHEL root PATH on hermes update
Existing RHEL/CentOS/Rocky/Alma root installs won't be repaired by the
install.sh fix alone because 'hermes update' is an in-place git pull, not
a rerun of install.sh. Port the same probe + idempotent .bashrc write
into cmd_update so affected users get fixed automatically on next update.
_ensure_fhs_path_guard() runs after 'Update complete!':
- Linux + root + FHS-layout install (command at /usr/local/bin/hermes) only
- Probe: env -i bash -i -c 'command -v hermes' — fresh non-login interactive
shell, same scenario the user reports
- On failure, append PATH guard to /root/.bashrc and /root/.bash_profile,
skipping if any uncommented PATH line already mentions /usr/local/bin
- Silent no-op on macOS, non-root, legacy layout, or shells that already
resolve hermes
Top-level channel messages arrive at _resolve_thread_ts with
metadata.thread_id set to the message's own ts, because the inbound
handler in _handle_message_event uses 'event.ts' as a session-keying
fallback when event.thread_ts is absent. That made metadata alone
insufficient to distinguish a real thread reply from a top-level
message, so reply_in_thread=false only took effect in DMs.
Use reply_to (== incoming message_id == ts for top-level messages) as
the tiebreaker: when metadata.thread_id == reply_to the 'thread' is the
synthetic session-keying fallback, not a real parent, so we reply
directly in the channel. Real thread replies (reply_to != thread_id)
still resolve to the parent thread and preserve conversation context.
Closes#9268.
Parameterize the test helpers in test_status_command.py to accept a
Platform and add two regression tests ensuring the first-run home-channel
onboarding uses '/hermes sethome' on Slack and '/sethome' everywhere else.
Co-authored-by: sgaofen <135070653+sgaofen@users.noreply.github.com>
Slack's adapter registers a single parent slash command /hermes and
dispatches subcommands via slack_subcommand_map(). Bare /sethome is
not a registered command on Slack and fails with 'app did not
respond', logging 'Unhandled request' in slack_bolt.AsyncApp.
Show /hermes sethome in the first-run onboarding hint when the
source platform is Slack; keep /sethome for Telegram, Discord,
Matrix, Mattermost, and other platforms that register it directly.
Fixes#14632
Repeated /queue commands now each produce a full agent turn, in order,
with no merging. Previously the second /queue overwrote the first
because the handler wrote directly into the adapter's single-slot
_pending_messages dict.
- GatewayRunner grows a _queued_events overflow buffer (dict of list).
- /queue puts new items in the adapter's next-up slot when free,
otherwise appends to the overflow. After each run's drain consumes
the slot, the next overflow item is promoted so the recursive run
picks it up.
- /new and /reset clear the overflow.
- /status now reports queue depth when non-zero.
- Ack message shows the depth once it exceeds 1.
Helpers (_enqueue_fifo, _promote_queued_event, _queue_depth) use the
getattr default-fallback pattern so existing tests that build bare
GatewayRunner instances via object.__new__ keep working.
Before: delegate_task children each allocated their own terminal
sandbox keyed by child task_id. Starting extra containers (or Modal
sandboxes / Daytona workspaces) is expensive, and the subagent's work
is invisible to the parent — files written by the child in its
container don't exist in the parent's when the subagent returns.
After: a single `_resolve_container_task_id` helper maps any
tool-call task_id to "default" UNLESS an env override is registered
for it. The parent agent and all delegate_task children therefore
share one long-lived sandbox — installed packages, cwd, /workspace
files, and /tmp scratch carry over freely between them.
RL and benchmark environments (TerminalBench2, HermesSweEnv, ...)
opt in to isolation via `register_task_env_overrides(task_id, {...})`;
those task_ids survive the collapse and get their own sandbox,
preserving the per-task Docker image behavior these benchmarks rely on.
file_state / active-subagents registry / TUI events still key off the
original child task_id, so the 'subagent wrote a file the parent read'
warning and UI per-subagent panels keep working.
Tradeoff: parallel delegate_task children (tasks=[...]) now share one
bash/container. Concurrent cd, env-var mutations, and writes to the
same path will collide. If that bites a specific workflow, the
subagent can opt back into isolation via register_task_env_overrides.
Applied at four lookup sites:
- tools/terminal_tool.py terminal_tool() and get_active_env()
- tools/file_tools.py _get_file_ops() and _get_live_tracking_cwd()
- tools/code_execution_tool.py _get_or_create_environment()
Docs: website/docs/user-guide/configuration.md updated to reflect the
shared-container reality and document the RL/benchmark carve-out.
Tests: tests/tools/test_shared_container_task_id.py (9 cases).
Every command in COMMAND_REGISTRY (/btw, /stop, /model, /help, /new,
/bg, /reset, ...) is now a first-class Slack slash command instead of
a /hermes <subcommand>. Users get the same autocomplete-driven slash
picker experience Slack users expect and that Discord and Telegram
already provide.
Previously Slack registered ONE native slash (/hermes) and split on
the first word, so typing /btw in Slack's composer got 'couldn't find
an app for /btw' because the workspace manifest never declared it.
Changes
- hermes_cli/commands.py: slack_native_slashes() + slack_app_manifest()
generate a Slack manifest from the registry (canonical names +
aliases + plugin commands), clamped to Slack's 50-slash cap with
/hermes reserved as the catch-all.
- gateway/platforms/slack.py: single regex matcher dispatches every
registered slash to _handle_slash_command, which dispatches on
command['command']. Legacy /hermes <subcommand> keeps working for
backward compat with older workspace manifests.
- hermes_cli/slack_cli.py + hermes_cli/main.py: new 'hermes slack
manifest' command prints/writes a full manifest (display info,
OAuth scopes, event subs, socket mode, slash commands) ready to
paste into 'Create from manifest' or Features → App Manifest.
- hermes_cli/setup.py: _setup_slack() now writes the manifest up-front
and points users at the 'From an app manifest' flow; also offers
to refresh the manifest on reconfigure for picking up new commands.
- Tests: 14 new tests covering native-slash dispatch (/btw, /stop,
/model), legacy /hermes <sub> compat, manifest structure, and
telegram<->slack parity (every Telegram command must also register
as a Slack slash). Existing /hermes-registration test updated to
assert the new regex matches /hermes, /btw, /stop, /model, /help.
- Docs: slack.md gains a 'Slash Commands' section + Option A manifest
flow in Step 1; cli-commands.md documents 'hermes slack manifest'.
Users pick up the new slashes by running 'hermes slack manifest --write'
and pasting into Features → App Manifest → Edit in their Slack app
config, then Save (Slack prompts for reinstall if scopes changed).
The Docker terminal-backend docs said 'each session starts a long-lived
container', implying a fresh container per chat session. That hasn't been
true for a while: for the top-level agent, task_id defaults to 'default'
and the container is cached in _active_environments for the lifetime of
the Hermes process. /new, /reset, and switching sessions all reuse the
same container. Only delegate_task subagents and RL rollouts get isolated
containers keyed by their own task_id.
skills/feeds/ only contained a category-marker DESCRIPTION.md with no
actual skills in it. Removing the directory and the 'feeds' -> 'Feeds'
display-label mapping in website/scripts/extract-skills.py (the only
other reference in the repo).
* fix(tui): call maybe_auto_title for TUI sessions (#15961)
The maybe_auto_title() helper is called from cli.py and gateway/run.py
but was never wired into tui_gateway/server.py, so every session started
via 'hermes --tui' landed in state.db with an empty title. Evidence from
the issue reporter: 0/154 TUI sessions titled vs 91/383 CLI.
Mirror the CLI/Gateway pattern: after emitting message.complete, when the
turn finished cleanly, fire-and-forget title generation using the session
key, user prompt, agent response, and current history.
Fixes#15949.
Co-authored-by: math0r-be <math0r-be@github.com>
* chore(release): map math0r-be placeholder email in AUTHOR_MAP
---------
Co-authored-by: math0r-be <math0r-be@github.com>
* fix(/branch): redirect session_log_file and expose branch sessions in list
Two bugs when using /branch:
1. cli.py _handle_branch_command updated agent.session_id but not
agent.session_log_file, so all messages written after branching
landed in the original session's JSON file and the branch never
got its own session_{id}.json on disk.
Fix: mirror the compression-split path (run_agent.py:7579) and
update session_log_file immediately after changing session_id.
2. hermes_state.py list_sessions_rich filtered out every session
with parent_session_id IS NOT NULL to hide sub-agent runs and
compression continuations. Branch sessions share this column, so
they became invisible to `hermes sessions list` and `sessions browse`.
Fix: also include branch children — those whose parent ended with
end_reason='branched' AND whose started_at >= parent.ended_at
(the same timing condition that get_compression_tip uses to
distinguish continuations from live-spawned subagents).
Fixes#14854
Co-Authored-By: Octopus <liyuan851277048@icloud.com>
* chore(release): map octo-patch placeholder email in AUTHOR_MAP
---------
Co-authored-by: octo-patch <octo-patch@github.com>
Co-authored-by: Octopus <liyuan851277048@icloud.com>
When a cloud browser provider (Browserbase / Browser-Use / Firecrawl) is
configured, browser_navigate now transparently spawns a local Chromium
sidecar for URLs whose host resolves to a private/loopback/LAN address
(localhost, 127.0.0.1, 192.168.x.x, 10.x.x.x, *.local, *.lan, *.internal,
::1, 169.254.x.x). Public URLs continue to use the cloud provider in the
same conversation.
Previously, setting BROWSERBASE_API_KEY / cloud_provider: browserbase
pinned the whole tool to cloud for the process — localhost URLs were
either SSRF-blocked (default) or sent to Browserbase (where they 404'd
because the cloud can't reach your LAN). Users who wanted 'cloud for
public, local for localhost' had no way to express it short of toggling
providers mid-session.
Implementation uses a composite session key scheme: the bare task_id
serves the cloud session, and a '{task_id}::local' sidecar serves the
local Chromium. _last_active_session_key[task_id] tracks which of the
two served the most recent nav so snapshot/click/fill/etc. hit the
correct one. cleanup_browser(bare_task_id) reaps both.
Feature is on by default. Opt out via:
browser:
auto_local_for_private_urls: false
The cloud provider never sees private URLs. Post-redirect SSRF guard
is preserved: redirects from public onto private addresses still block.
'hermes skills list' now shows every skill's enabled/disabled status
and accepts --enabled-only to filter down to what will actually load
for the active profile:
hermes -p dario skills list --enabled-only
Previously the command was a flat catalog — it did not apply
skills.disabled from config.yaml, so there was no way to see the
live skill set for a profile without reading config by hand.
Profile switching already works via -p (swaps HERMES_HOME); this
just surfaces the result visibly.
Changes:
- hermes_cli/skills_hub.py: do_list adds a Status column and an
enabled_only filter; summary reports enabled/disabled split
- hermes_cli/main.py: --enabled-only flag on 'skills list'
- /skills list slash command accepts --enabled-only too
- tests: 4 new (status column, disabled marking, enabled-only
hiding, no platform leakage into get_disabled_skill_names);
existing fixtures updated to accept skip_disabled kwarg
Reported by @mochizukimr on X.
* feat(cli,tui): surface /queue, /bg, /steer in agent-running placeholder
While the agent loop is running, the input placeholder previously only
hinted at Enter-to-interrupt. Surface the full set of busy-time actions
(interrupt via new message, /queue, /bg, /steer) so users discover them
without hunting through docs or Teknium's tweets.
- cli.py: "msg=interrupt · /queue · /bg · /steer · Ctrl+C cancel"
- ui-tui/src/components/appLayout.tsx: same string (was "Ctrl+C to interrupt…")
* revert tui placeholder change (cli-only per review)
Address Copilot review findings:
1. Gate _last_activity_desc on interrupt_depth == 0 alongside _last_activity_ts.
Both fields are semantically paired — desc describes the activity *at* ts.
Updating desc without ts made get_activity_summary() report "starting new
turn (cached)" for 20+ minutes while the timestamp showed the true stale
duration, producing misleading diagnostic output.
2. Monkeypatch gateway.run.time.time to a fixed epoch in tests that assert
on _last_activity_ts values. Real time.time() comparisons were latently
flaky under slow CI or NTP adjustments. _FAKE_NOW = 10_000.0 is used
as the reference; assertions are now exact equality rather than >=.
3. Add test_fresh_turn_resets_desc and test_interrupt_turn_preserves_desc to
directly cover the gated desc behaviour introduced by (1).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
_last_activity_ts was unconditionally reset to time.time() on every
_agent_cache hit. For interrupt-recursive _run_agent calls
(_interrupt_depth > 0) this silently reset the inactivity watchdog's
idle clock on each re-entry, preventing the 30-min timeout from ever
firing when a turn got stuck in an interrupt loop. A stuck session
would emit "Still working... iteration 0/60, starting new turn (cached)"
heartbeats indefinitely instead of timing out.
Gate the reset on _interrupt_depth == 0 only. Fresh external turns
still receive the reset so a session idle for 29 min doesn't trip the
watchdog before the new turn makes its first API call (#9051).
The per-turn reset logic is extracted into a static helper
_init_cached_agent_for_turn() to make it directly testable.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Follow-up to #6616 covering the remaining user-injected prompt markers that
the original PR did not touch (reporter's second comment on #6576 explicitly
flagged these). Azure OpenAI Default/DefaultV2 content filters treat any
bracketed [SYSTEM: ...] as prompt-injection and reject with HTTP 400.
Remaining call sites renamed:
- cli.py: background-process notifications (watch_disabled, watch_match,
completion), MCP reload notice (4 live + 1 docstring)
- gateway/run.py: same notification paths + auto-loaded skill banner +
MCP reload notice (5 live + 1 docstring)
- tools/process_registry.py: comment reference
Not renamed:
- environments/hermes_base_env.py '[SYSTEM]\n{content}' — RL training
trajectory rendering only, never sent to Azure, part of a symmetric
[USER]/[ASSISTANT]/[TOOL] scheme.
AUTHOR_MAP: buraysandro9@gmail.com -> ygd58.
Azure OpenAI content filters (Default/DefaultV2) treat bracketed
[SYSTEM: ...] meta-instructions as prompt-injection attempts and
reject requests with HTTP 400.
Replacing [SYSTEM: with [IMPORTANT: preserves the same semantic
meaning for the model while bypassing the Azure heuristic.
Fixes#6576
Follow-up to cherry-picked PR #15920:
- agent/credential_pool.py: hoist 'from hermes_cli.config import get_env_value'
to module top instead of inline try/except in each seed site (3 sites).
No import cycle — hermes_cli/config.py doesn't depend on agent.credential_pool.
- hermes_cli/auth.py: same hoist for the _resolve_api_key_provider_secret loop.
- tests/tools/test_credential_pool_env_fallback.py: replace smoke-only tests
with real .env file I/O. Each test writes a temp ~/.hermes/.env, verifies
_seed_from_env / _resolve_api_key_provider_secret read from it, and asserts
the full priority chain: os.environ > .env > credential_pool. Uses
'deepseek' as the test provider since 'openai' isn't in PROVIDER_REGISTRY
and _seed_from_env's generic path requires a real pconfig lookup.
_resolve_api_key_provider_secret() and _seed_from_env() only checked
os.environ for provider API keys. When keys exist in ~/.hermes/.env but
are not loaded into the process environment (e.g. ACP adapter entry
point, post-session-start .env edits, or non-CLI entry points), the
resolution returns an empty string, causing HTTP 401 failures.
Changes:
- credential_pool._seed_from_env: use get_env_value() which checks both
os.environ and ~/.hermes/.env file, preventing _prune_stale_seeded_entries
from removing valid entries whose env var isn't in os.environ
- credential_pool._seed_from_env: same fix for openrouter and
base_url_env_var resolution
- auth._resolve_api_key_provider_secret: use get_env_value() instead of
os.getenv(), and add credential_pool fallback when env resolution fails
Fixes#15914
The background memory/skill review (_spawn_background_review) has always
forked a new AIAgent passing only model and provider, then relied on
AIAgent.__init__ to re-resolve credentials from env vars. This works for
users with keys in ~/.hermes/.env but silently falls back to env-var
auto-resolution in all cases, which fails for OAuth-only providers,
session-scoped creds, and credential-pool setups where auth can't be
reconstructed from env.
This used to be invisible -- failures were swallowed via logger.debug().
PR 8a2506af4 (Apr 24) surfaced auxiliary failures to the user, which
made the stale bug visible as:
"Auxiliary background review failed: No LLM provider configured"
Fix: pass api_key, base_url, api_mode, and credential_pool from the
parent's live runtime into the fork -- matching how every other
auxiliary path (compression, memory flush, vision, session search)
already inherits the parent's credentials via _current_main_runtime().
The chown/chmod block on config.yaml was added in b24d239ce to keep the
file readable by the hermes runtime user, but it sat in the post-gosu
'running as hermes' section of the entrypoint. That meant:
1. Default `docker run <image>` — container starts as root, entrypoint
drops to hermes via gosu, then non-root hermes tries to chown the
file to hermes. Works by coincidence because the file was just
created by root during volume setup and gosu target == target owner.
2. `docker run -u $(id -u):$(id -g) <image>` (#15865) — container
starts as the caller's UID. The root block is skipped entirely, we
land in the hermes section as some arbitrary non-root user, and
chown to 'hermes' fails with 'Operation not permitted'. Script
aborts under `set -e`.
Move the chown/chmod into the root block (before the gosu exec) where
it actually has privilege, and guard with `2>/dev/null || true` so
rootless Podman (where even in-container root lacks host-side chown
rights) doesn't abort either.
Closes#15865
Salvage PR #15883 cherry-picked FocusFlow Dev's commit; release-notes
CI needs the AUTHOR_MAP entry to attribute to the PR author's GitHub
login rather than a placeholder.
New `hermes kanban` CLI subcommand + `/kanban` slash command + skills for
worker and orchestrator profiles. SQLite-backed task board
(~/.hermes/kanban.db) shared across all profiles on the host. Zero
changes to run_agent.py, no new core tools, no tool-schema bloat.
Motivation: delegate_task is a function call — sync fork/join, anonymous
subagent, no resumability, no human-in-the-loop. Kanban is the durable
shape needed for research triage, scheduled ops, digital twins,
engineering pipelines, and fleet work. They coexist (workers may call
delegate_task internally).
What this adds
- hermes_cli/kanban_db.py — schema, CAS claim, dependency resolution,
dispatcher, workspace resolution, worker-context builder.
- hermes_cli/kanban.py — 15-verb CLI surface and shared run_slash()
entry point used by both CLI and gateway.
- skills/devops/kanban-worker — how a profile should work a claimed task.
- skills/devops/kanban-orchestrator — "you are a dispatcher, not a
worker" template with anti-temptation rules.
- /kanban slash command wired into cli.py and gateway/run.py. Bypasses
the running-agent guard (board writes don't touch agent state), so
/kanban unblock can free a stuck worker mid-conversation.
- Design spec at docs/hermes-kanban-v1-spec.pdf — comparative analysis
vs Cline Kanban, Paperclip, NanoClaw, Gemini Enterprise; 8 patterns;
4 user stories; implementation plan; concurrency correctness.
- Docs: website/docs/user-guide/features/kanban.md, CLI reference
updated, sidebar entry added.
Architecture highlights
- Three planes: control (user + gateway), state (board + dispatcher),
execution (pool of profile processes).
- Every worker is a full OS process, spawned as `hermes -p <profile>`.
No in-process subagent swarms — solves NanoClaw's SDK-lifecycle
failure class.
- Atomic claim via SQLite CAS in a BEGIN IMMEDIATE transaction; stale
claims reclaimed 15 min after their TTL expires.
- Tenant namespacing via one nullable column — one specialist fleet
can serve many businesses with data isolation by workspace path.
Tests: 60 targeted tests (schema, CAS atomicity, dependency resolution,
dispatcher, workspace kinds, tenancy, CLI + slash surface). All pass
hermetic via scripts/run_tests.sh.
Follow-up to PR #16053 (/btw as /background alias). Cleans up the
plumbing added exclusively for the old ephemeral /btw handler and
repairs a broken btw bypass that landed between my refactor and this
follow-up.
run_agent.py:
- Remove persist_session kwarg, instance attr, and _persist_session
short-circuit. Only /btw ever passed persist_session=False; with
/btw gone the default (always persist) is the only behavior anyone
ever wanted.
gateway/run.py:
- Remove the unreachable 'if _cmd_def_inner.name == "btw"' block
(PR #16059). Canonical name for a /btw message is 'background' after
alias resolution — the comparison could never be true, and it called
_handle_btw_command which no longer exists. The /background branch
above it already dispatches /btw correctly.
tests/gateway/test_running_agent_session_toggles.py:
- Fix test_btw_dispatches_mid_run to mock _handle_background_command
(the real dispatch target for /btw) instead of the deleted
_handle_btw_command.
/btw spawns a parallel ephemeral side-question task (self-guarded against
concurrent /btw on the same chat) — exactly like /background. But it was
missing from the running-agent bypass list in _handle_message(), so it
fell through to the catch-all and returned:
⏳ Agent is running — /btw can't run mid-turn. Wait for the current
response or /stop first.
That's the opposite of what /btw is for — asking a side question while
the main turn is still working. Add the bypass next to /background and a
regression test covering the mid-turn dispatch path.
Reported by @IuriiTiunov on Telegram.
The ephemeral no-tools side-question variant of /btw confused users who
expected 'by-the-way' to mean 'run this off to the side with tools' —
they'd type /btw and get a toolless agent that couldn't do the work.
/bg worked because it was /background with full tools.
Collapse the two: /btw and /bg both alias to /background. One command,
one behavior, no more gotchas about which variant has tools.
Removed:
- _handle_btw_command in cli.py and gateway/run.py
- _run_btw_task + _active_btw_tasks state in gateway/run.py
- prompt.btw JSON-RPC method + btw.complete event in tui_gateway
- BtwStartResponse type + btw.complete case in ui-tui
- Standalone /btw slash tree registration in Discord
- Standalone btw CommandDef in hermes_cli/commands.py
Updated:
- background CommandDef aliases: (bg,) -> (bg, btw)
- TUI session.ts: local btw handler merged into background
- Docs and tips updated to describe /btw as a /background alias
PR #16046 added /busy and /verbose hints to the classic CLI and the
gateway runner but skipped the Ink TUI (and therefore the dashboard
/chat page, which embeds the TUI via PTY). This extends the same
latch to the TUI with TUI-native wording.
The TUI's busy-input model is not the /busy knob from the CLI —
single Enter while busy auto-queues, double Enter on an empty line
interrupts. The new busy-input hint teaches THAT gesture instead of
telling the user to flip a config that does not apply.
Changes:
- agent/onboarding.py — add busy_input_hint_tui() + tool_progress_hint_tui()
- tui_gateway/server.py — onboarding.claim JSON-RPC (Ink triggers busy
hint on enqueue) + _maybe_emit_onboarding_hint helper hooked into
_on_tool_complete for the 30s/tool_progress=all path. Same
config.yaml latch so each hint fires at most once per install across
CLI, gateway, and TUI combined.
- ui-tui/src/gatewayTypes.ts — OnboardingClaimResponse + onboarding.hint event
- ui-tui/src/app/createGatewayEventHandler.ts — render the hint event as sys()
- ui-tui/src/app/useSubmission.ts — claim busy_input_prompt on first
busy enqueue
- tests/agent/test_onboarding.py — +3 cases for TUI hint shape
- tests/tui_gateway/test_protocol.py — +4 cases for onboarding.claim
- website/docs/user-guide/tui.md — new 'Interrupting and queueing'
section explaining the TUI's double-Enter model and the hints
Validation:
scripts/run_tests.sh tests/agent/test_onboarding.py \
tests/tui_gateway/test_protocol.py \
tests/gateway/test_busy_session_ack.py
-> 66 passed
npm --prefix ui-tui run type-check -> clean
npm --prefix ui-tui run lint -> clean
npm --prefix ui-tui run build -> clean
Manage the fallback_providers chain from the CLI instead of hand-editing
config.yaml. The picker reuses select_provider_and_model() from 'hermes
model' — same provider list, same credential prompts, same model picker.
hermes fallback [list] Show the current chain (primary + fallbacks)
hermes fallback add Run the model picker, append selection to chain
hermes fallback remove Pick an entry to delete (arrow-key menu)
hermes fallback clear Remove all entries (with confirmation)
'add' snapshots config['model'] before calling the picker, extracts the
user's selection from the post-picker state, then restores the primary
and appends {provider, model, base_url?, api_mode?} to fallback_providers.
Auth store's active_provider is snapshot/restored too so OAuth-provider
fallbacks don't silently deactivate the user's primary. Duplicates and
self-as-fallback are rejected. Legacy single-dict 'fallback_model' entries
are auto-migrated to the list format on first write.
Instead of a blocking first-run questionnaire, show a one-time hint the first
time the user hits each behavior fork:
1. First message while the agent is working — appends a hint to the busy-ack
explaining the /busy queue vs /busy interrupt knob, phrased to match the
mode that was just applied (don't tell a queue-mode user to switch to
queue).
2. First tool that runs for >= 30s in the noisiest progress mode
(tool_progress: all) — prints a hint about /verbose to cycle display
modes (all -> new -> off -> verbose). Gated on /verbose actually being
usable on the surface: always shown on CLI; on gateway only shown when
display.tool_progress_command is enabled.
Each hint is latched in config.yaml under onboarding.seen.<flag>, so it
fires exactly once per install across CLI, gateway, and cron, then never
again. Users can wipe the section to re-see hints.
New:
- agent/onboarding.py — is_seen / mark_seen / hint strings, shared by
both CLI and gateway.
- onboarding.seen in DEFAULT_CONFIG (hermes_cli/config.py) and in
load_cli_config defaults (cli.py). No _config_version bump — deep
merge handles new keys.
Wired:
- gateway/run.py: _handle_active_session_busy_message appends the hint
after building the ack. progress_callback tracks tool.completed
duration and queues the tool-progress hint into the progress bubble.
- cli.py: CLI input loop appends the busy-input hint on the first busy
Enter; _on_tool_progress appends the tool-progress hint on the first
>=30s tool completion. In-memory CLI_CONFIG is also updated so
subsequent fires in the same process are suppressed immediately.
All writes go through atomic_yaml_write and are wrapped in try/except
so onboarding can never break the input/busy-ack paths.
The base adapter's auto-TTS path fired on any voice message unless the
chat had explicitly run /voice off — it never read voice.auto_tts from
config.yaml, so users who set auto_tts: false still got audio replies.
Gate the base adapter on a three-layer decision instead:
1. chat in _auto_tts_enabled_chats (explicit /voice on|tts) → fire
2. chat in _auto_tts_disabled_chats (explicit /voice off) → suppress
3. else → voice.auto_tts global default
Runner now pushes voice.auto_tts onto the adapter as _auto_tts_default
and mirrors /voice on|tts chats into _auto_tts_enabled_chats via the
existing _sync_voice_mode_state_to_adapter path. /voice off still wins.
Closes#16007.
Users who run `hermes setup` get `cli-config.yaml.example` copied verbatim
(including comments) to ~/.hermes/config.yaml. But several display settings
had thin comments that didn't enumerate the valid options, so users couldn't
tell from reading their config what values each key accepts.
- busy_input_mode: widen from 'CLI' to 'CLI and gateway platforms';
note /stop as gateway equivalent of Ctrl+C; add /busy_input_mode runtime hint
- compact, interim_assistant_messages, bell_on_complete, show_reasoning,
streaming: add true/false option lines showing effect of each value
- skin: refresh the built-in skin list (was missing daylight, warm-lightmode,
poseidon, sisyphus, charizard — 5 of 9 built-ins undocumented)
When the LLM response carries N parallel tool calls, the agent fires
N tool.started events back-to-back before its interrupt check runs.
A user sending /stop mid-batch would see the '⚡ Interrupting current
task' ack followed by a trail of 🔍 web_search bubbles for the remaining
events in the batch — making the interrupt feel ignored.
progress_callback and the drain loop in send_progress_messages now
check agent.is_interrupted (via agent_holder[0], the existing
cross-scope handle). Events that arrive after interrupt are dropped
at both the queueing and rendering stages. The '⚡ Interrupting'
message is sent through a separate adapter path and is unaffected.
Follow-up on #16020 salvage. Three corrections:
1. Truth signal for /copy
Before: success was 'OSC 52 sequence was emitted to stdout'. That's
false on local Linux inside tmux (emitSequence=false), so /copy kept
printing 'clipboard copy failed' to users whose xclip/wl-copy had
already succeeded fire-and-forget.
Fix: setClipboard() now returns { sequence, success } where success =
native-fired OR tmux-buffer-loaded OR osc52-emitted. copyNative()
returns a boolean telling setClipboard whether a native attempt was
made. /copy only shows 'failed' when literally no path was taken.
2. Dashboard keybinding
Before: Ctrl+C for copy on non-Mac (Ctrl+Shift+C for paste).
That swallows SIGINT when a stale selection is present and breaks
the xterm/gnome-terminal/konsole/Windows-Terminal convention where
Ctrl+C in a terminal emulator is always SIGINT. The real bug was
that clipboard writes lost user-gesture through OSC-52 round-trips,
which the direct writeText already fixes.
Fix: revert copyModifier to Ctrl+Shift+C on non-Mac. Direct
writeText in the keydown handler preserves user gesture. term.write
Escape replaced with term.clearSelection() (works without relying
on TUI input mode).
3. Error toast text
Before: 'see HERMES_TUI_DEBUG_CLIPBOARD' — tells users how to
debug but not how to fix.
Fix: point users at HERMES_TUI_FORCE_OSC52=1 first (the actual
escape hatch), mention the debug var second.
- Dashboard copy: direct Clipboard API on Ctrl+C/Cmd+C (user gesture);
send Escape to TUI to clear selection; Ctrl+Shift+C kept as fallback.
- TUI /copy: copySelection() async; only reports success if OSC52 emitted.
- Add HERMES_TUI_FORCE_OSC52 env var to override native-tool detection.
- Fixes "copied N chars" false-positive when clipboard backend absent.
Changes:
web/src/pages/ChatPage.tsx — direct navigator.clipboard.writeText
ui-tui/packages/hermes-ink/src/ink/ink.tsx — async copySelection
ui-tui/packages/hermes-ink/src/ink/termio/osc.ts — HERMES_TUI_FORCE_OSC52
ui-tui/src/app/slash/commands/core.ts — async /copy with honest feedback
Problem: Ctrl+C in Hermes TUI shows 'copied' but clipboard often empty.
Root causes:
- Native Linux tools (xclip, wl-copy) require DISPLAY/WAYLAND_DISPLAY; in
headless Docker/SSH they fail or hang.
- OSC 52 fallback requires terminal emulator support; when absent, sequence
is dropped silently.
- Dashboard OSC 52 → Clipboard API path fails due to missing user gesture;
errors were silently caught.
- User feedback 'copied selection' was shown unconditionally, regardless of
success.
Solution implemented:
- Short-circuit Linux native clipboard probing when no display server is
present (no DISPLAY and no WAYLAND_DISPLAY). Avoids futile attempts and
timeouts.
- Add HERMES_TUI_DEBUG_CLIPBOARD env var (1/true). When set, TUI logs to
stderr which clipboard path is used, probe results on Linux, and whether
OSC 52 was emitted. Greatly improves diagnosability.
- Improve dashboard clipboard error handling: replace empty catch blocks
with console.warn messages for OSC 52 decode/Write failures and direct
copy/paste errors. Makes browser permission/user-gesture failures visible
in DevTools.
- Add comprehensive clipboard troubleshooting documentation to README and
AGENTS, covering OSC 52 verification, tmux config, Docker/headless
constraints, env vars, dashboard caveats, and fallback strategies.
Technical details:
- in ui-tui/packages/hermes-ink/src/ink/termio/osc.ts:
- Early return on Linux if both DISPLAY and WAYLAND_DISPLAY unset.
- Refactor probe sequence to async with 500ms timeout,
caching result; subsequent copies use cached tool immediately.
- Emit debug logs when HERMES_TUI_DEBUG_CLIPBOARD=1.
- in ink.tsx: log when OSC 52 not emitted (native
or tmux path in use) in debug mode.
- : OSC 52 handler and Ctrl+Shift+C handler now
log warnings to console on Clipboard API rejection with error message.
- Documentation: new 'Clipboard Troubleshooting' section in README; new
'Clipboard environment variables and pitfalls' subsection in AGENTS.md
(Known Pitfalls).
Tests: full ui-tui test suite (292 tests) passes; clipboard and OSC tests
unaffected. No breaking changes.
Files changed:
- ui-tui/packages/hermes-ink/src/ink/termio/osc.ts
- ui-tui/packages/hermes-ink/src/ink/ink.tsx
- web/src/pages/ChatPage.tsx
- README.md
- AGENTS.md
- CHANGELOG.md (new)
OpenRouter and Nous Portal curated picker lists now resolve via a JSON
manifest served by the docs site, falling back to the in-repo snapshot
when unreachable. Lets us update model lists without shipping a release.
Live URL: https://hermes-agent.nousresearch.com/docs/api/model-catalog.json
(source at website/static/api/model-catalog.json; auto-deploys via the
existing deploy-site.yml GitHub Pages pipeline on every merge to main).
Schema (v1) carries id + optional description + free-form metadata at
manifest, provider, and model levels. Pricing and context length stay
live-fetched via existing machinery (/v1/models endpoints, models.dev).
Config (new model_catalog section, default enabled):
model_catalog.url master manifest URL
model_catalog.ttl_hours disk cache TTL (default 24h)
model_catalog.providers.<name>.url optional per-provider override
Fetch pipeline: in-process cache -> disk cache (fresh < TTL) -> HTTP
fetch -> disk-cache-on-failure fallback -> in-repo snapshot as last
resort. Never raises to callers; at worst returns the bundled list.
Changes:
- website/static/api/model-catalog.json initial manifest (35 OR + 31 Nous)
- scripts/build_model_catalog.py regenerator from in-repo lists
- hermes_cli/model_catalog.py fetch + validate + cache module
- hermes_cli/models.py fetch_openrouter_models() +
new get_curated_nous_model_ids()
- hermes_cli/main.py, hermes_cli/auth.py Nous flows use the helper
- hermes_cli/config.py model_catalog defaults
- website/docs/reference/model-catalog.md + sidebars.ts
- tests/hermes_cli/test_model_catalog.py 21 tests (validation, fetch
success/failure, accessors,
disabled, overrides, integration)
Stop pre-stripping the path from the configured MCP server URL before
constructing OAuthClientProvider. The MCP SDK strips the path itself via
OAuthContext.get_authorization_base_url() for authorization-server
discovery, but uses the full server_url through
resource_url_from_server_url() + check_resource_allowed() to validate
against the server's RFC 9728 Protected Resource Metadata.
For servers whose PRM advertises a path-scoped resource (e.g. Notion's
https://mcp.notion.com/mcp), our _parse_base_url() collapsed the URL to
the origin, so check_resource_allowed() saw requested='/' vs
configured='/mcp/' and refused the token. Fixes OAuth against Notion MCP
(and any other path-scoped resource).
Closes#16015.
`_apply_model_switch_result` (the interactive `/model` picker's
confirmation path) printed `ModelInfo.context_window` straight from
models.dev, which reports the vendor-wide value (1.05M for gpt-5.5 on
openai). ChatGPT Codex OAuth caps the same slug at 272K, so the picker
showed 1M while the runtime (compressor, gateway `/model`, typed
`/model <name>`) correctly used 272K — the classic 'sometimes 1M,
sometimes 272K' mismatch on a single model.
Both display paths now go through `resolve_display_context_length()`,
matching the fix that `_handle_model_switch` received earlier.
Also bump the stale last-resort fallback in DEFAULT_CONTEXT_LENGTHS
(`gpt-5.5: 400000 -> 1050000`) to match the real OpenAI API value; the
272K Codex cap is already enforced via the Codex-OAuth branch, so the
fallback now reflects what every non-Codex probe-miss should see.
Tests: adds `test_apply_model_switch_result_context.py` with three
scenarios (Codex cap wins, OpenRouter shows 1.05M, resolver-empty falls
back to ModelInfo). Updates the existing non-Codex fallback test to
assert 1.05M (the correct value).
## Validation
| path | before | after |
|-------------------------------|-----------|-----------|
| picker -> gpt-5.5 on Codex | 1,050,000 | 272,000 |
| picker -> gpt-5.5 on OpenAI | 1,050,000 | 1,050,000 |
| picker -> gpt-5.5 on OpenRouter | 1,050,000 | 1,050,000 |
| typed /model gpt-5.5 on Codex | 272,000 | 272,000 |
#14934 added deepseek-v4-pro / deepseek-v4-flash to the DeepSeek native
provider but the context-window lookup still falls back to the existing
"deepseek" substring entry (128K). DeepSeek V4 ships with a 1M context
window, so any caller relying on get_model_context_length() for
pre-flight token budgeting (compression, context warnings) under-counts
by ~8x.
Add explicit lowercase entries for the four DeepSeek model ids that
ship 1M context:
- deepseek-v4-pro
- deepseek-v4-flash
- deepseek-chat (legacy alias, server-side maps to v4-flash non-thinking)
- deepseek-reasoner (legacy alias, server-side maps to v4-flash thinking)
Longest-key-first substring matching means these explicit entries also
cover the vendor-prefixed forms (deepseek/deepseek-v4-pro on OpenRouter
and Nous Portal) without regressing the existing 128K fallback for
older / unknown DeepSeek model ids on custom endpoints.
Source: https://api-docs.deepseek.com/zh-cn/quick_start/pricing
The background skill-review prompt (spawned after N user turns) now instructs
the reviewer to SURVEY existing skills first, identify the CLASS of task, and
PREFER updating/generalizing an existing skill over creating a new narrow one.
This reduces near-duplicate skill accumulation at the source. Catches the
common failure mode where repeated tasks of the same class each spawn their
own specific skill ("fix-my-tauri-error", "fix-my-electron-error") instead
of a single class-level skill ("desktop-app-build-troubleshooting").
Applied to both _SKILL_REVIEW_PROMPT and the **Skills** half of
_COMBINED_REVIEW_PROMPT. Memory-only review prompt unchanged.
Groundwork for the Curator feature (issue #7816) — the creation-side fix.
Curator handles the retirement/consolidation side in a follow-up PR.
Tests assert the behavioral instructions are present (survey, class, update-
over-create, overlap-flagging, opt-out clause) rather than snapshotting the
full prompt text.
Nous Portal multiplexes multiple upstream providers (DeepSeek, Kimi,
MiMo, Hermes) behind one endpoint. Before this fix, any 429 on any of
those models recorded a cross-session file breaker that blocked EVERY
model on Nous for the cooldown window -- even though the caller's
own RPM/RPH/TPM/TPH buckets were healthy. Users hit a DeepSeek V4 Pro
capacity error, restarted, switched to Kimi 2.6, and still got
'Nous Portal rate limit active -- resets in 46m 53s'.
Nous already emits the full x-ratelimit-* header suite on every
response (captured by rate_limit_tracker into agent._rate_limit_state).
We now gate the breaker on that data: trip it only when either the
429's own headers or the last-known-good state show a bucket with
remaining == 0 AND a reset window >= 60s. Upstream-capacity 429s
(healthy buckets everywhere, but upstream out of capacity) fall
through to normal retry/fallback and the breaker is never written.
Note: the in-memory 'restart TUI/gateway to clear' workaround
circulated in Discord does NOT work -- the breaker is file-backed at
~/.hermes/rate_limits/nous.json. The workaround for users still
affected by a bad state file is to delete it.
Reported in Discord by CrazyDok1 and KYSIV (Apr 2026).
Plugin hooks fired after a tool dispatch now receive an integer
duration_ms kwarg measuring how long the tool's registry.dispatch()
call took (time.monotonic() before/after). Inspired by Claude Code
2.1.119 which added the same field to PostToolUse hook inputs.
Wire points:
- model_tools.py: measure dispatch latency, pass duration_ms to
invoke_hook("post_tool_call", ...) and invoke_hook("transform_tool_result", ...)
- hermes_cli/hooks.py: include duration_ms in the synthetic payload
used by 'hermes hooks test' and 'hermes hooks doctor' so shell-hook
authors see the same shape at development time as runtime
- shell hooks (agent/shell_hooks.py): no code change needed;
_serialize_payload already surfaces non-top-level kwargs under
payload['extra'], so duration_ms lands at extra.duration_ms for
shell-hook scripts
Plugin authors can now build latency dashboards, per-tool SLO alerts,
and regression canaries without having to wrap every tool manually.
Test: tests/test_model_tools.py::test_post_tool_call_receives_non_negative_integer_duration_ms
E2E: real PluginManager + dispatch monkey-patched with a 50ms sleep,
hook callback observes duration_ms=50 (int).
Refs: https://code.claude.com/docs/en/changelog (2.1.119, Apr 23 2026)
Adds a floor below --yolo: a tiny set of commands so catastrophic they
should never run via the agent, regardless of --yolo, gateway /yolo,
approvals.mode=off, or cron approve mode. Opting into yolo is trusting
the agent with your files and services — not trusting it to wipe the
disk or power the box off.
The list is deliberately small (12 patterns), covering only
unrecoverable ops:
- rm -rf targeting /, /home, /etc, /usr, /var, /boot, /bin, /sbin,
/lib, ~, $HOME
- mkfs (any variant)
- dd + redirection to raw block devices (/dev/sd*, /dev/nvme*, etc.)
- fork bomb
- kill -1 / kill -9 -1
- shutdown, reboot, halt, poweroff, init 0/6, telinit 0/6,
systemctl poweroff/reboot/halt/kexec
Recoverable-but-costly commands (git reset --hard, rm -rf /tmp/x,
chmod -R 777, curl | sh) stay in DANGEROUS_PATTERNS where yolo can
still pass them through — that's what yolo is for.
Container backends (docker/singularity/modal/daytona) continue to
bypass both hardline and dangerous checks, since nothing they do can
touch the host.
Inspired by Mercury Agent's permission-hardened blocklist.
Bare `hermes setup` on a returning user now drops straight into the
full reconfigure wizard — every prompt shows the current value as its
default, press Enter to keep or type a new value to change it. The
returning-user menu is gone.
Behavior:
- First-time user: first-time wizard (unchanged)
- Returning user, bare command: full reconfigure wizard (new default)
- Returning user, `--quick`: only prompt for missing/unset items
- Returning user, one section: `hermes setup model|terminal|gateway|tools|agent`
- `--reconfigure`: preserved as backwards-compat alias (no-op since it's now default)
The section functions already used current values as prompt defaults —
this change just removes the extra click to get to them.
The 'Quick Setup - configure missing items only' menu option is now
exposed as the explicit `--quick` flag; it's the narrow case of
filling in missing config (e.g. after a partial OpenClaw migration or
when a required API key got cleared).
Inspired by Mercury Agent's `mercury doctor` UX.
Also removes:
- RETURNING_USER_MENU_SECTION_KEYS (orphaned constant)
- Two returning-user menu tests in test_setup_noninteractive.py
(guarding behavior that no longer exists — covered by
test_setup_reconfigure.py instead)
- New website/docs/guides/azure-foundry.md covering both OpenAI-style
and Anthropic-style endpoints, auto-detection behaviour, gpt-5.x
routing, /v1 stripping, api-version query forwarding, and the
provider: anthropic + Azure URL alternative setup.
- environment-variables.md picks up AZURE_FOUNDRY_API_KEY,
AZURE_FOUNDRY_BASE_URL, AZURE_ANTHROPIC_KEY.
- cli-commands.md includes azure-foundry in the provider choices list.
- configuration.md lists azure-foundry among auxiliary-task providers.
- sidebars.ts wires the new guide into the Guides section.
- scripts/release.py AUTHOR_MAP entries for TechPrototyper,
HangGlidersRule (noreply), and pein892 so the contributor-attribution
CI check does not reject the salvage.
The azure-foundry wizard now probes the endpoint before asking the user
to pick anything by hand:
1. URL path sniff — endpoints ending in /anthropic are Azure Foundry
Claude routes and skip to anthropic_messages.
2. GET <base>/models probe — if the endpoint returns an OpenAI-shaped
model list, we switch to chat_completions and prefill the picker
with the returned deployment/model IDs.
3. Anthropic Messages probe — fallback for endpoints that don't expose
/models but do speak the Anthropic Messages shape.
4. Manual fallback — private endpoints / custom routes still work;
the user picks API mode + types a deployment name.
Context length for the selected model is resolved through the existing
agent.model_metadata.get_model_context_length chain (models.dev,
provider metadata, hardcoded family fallbacks) and stored in
model.context_length when a non-default value is found.
Also refactors runtime_provider so Azure Foundry resolution is reused
between the explicit-credentials path and the default top-level path —
previously the /v1 strip for Anthropic-style Azure only ran when the
caller passed explicit_* args, which meant config-driven sessions
hit a double-/v1 URL.
New module hermes_cli/azure_detect.py with 19 unit tests covering:
- path sniff, model ID extraction, probe fallbacks
- HTTP error handling (URLError, HTTPError)
- context-length lookup passthrough
- DEFAULT_FALLBACK_CONTEXT rejection
New runtime tests cover:
- OpenAI-style Azure Foundry
- Anthropic-style Azure Foundry with /v1 stripping
- Missing base_url / API key raising AuthError
Rationale: Microsoft confirms there's no pure-API-key endpoint to list
Azure deployments (that requires ARM management auth). The v1 Azure
OpenAI endpoint does expose /models with the resource's available
model catalog, which is good enough for picker prefill in the common
case. Users on private/gated endpoints fall through to manual entry.
Azure OpenAI exposes an OpenAI-compatible endpoint at
`{resource}.openai.azure.com/openai/v1` that accepts the standard
`openai` Python client. Two issues prevented gpt-5.x models from working:
1. `_max_tokens_param()` only sent `max_completion_tokens` for
`api.openai.com` URLs. Azure also requires `max_completion_tokens`
for gpt-5.x models.
2. The `codex_responses` upgrade gate unconditionally upgraded gpt-5.x
to Responses API. Azure does NOT support the Responses API — it serves
gpt-5.x on the regular `/chat/completions` path, causing a 404.
Fix: add `_is_azure_openai_url()` that matches `openai.azure.com` URLs.
- `_max_tokens_param()` now returns `max_completion_tokens` for Azure.
- The `codex_responses` upgrade gate skips Azure so gpt-5.x stays on
`chat_completions` where Azure actually serves it.
- The fallback-provider api_mode picker also recognises Azure and stays
on chat_completions.
- Tests cover max_tokens routing, api_mode behaviour, and URL detection.
gpt-4.x models on Azure are unaffected (already used chat_completions +
max_tokens, which Azure accepts for those models).
Salvage of PR #10086 — rewritten against current main where the
codex_responses upgrade gate gained copilot-acp / explicit-api_mode
exclusions.
Azure OpenAI requires an `api-version` query parameter on every request.
When users include it in the base_url (e.g. `?api-version=2025-04-01-preview`),
the OpenAI SDK silently drops it during URL construction, causing 404 errors.
Extract query params from base_url and pass them via `default_query` so the
SDK appends them to every request. This is a generic solution that works for
any custom endpoint requiring query parameters, not just Azure.
No-op for URLs without query params — fully backward compatible.
Add support for Azure Foundry as a new inference provider. Azure Foundry
endpoints can use either OpenAI-style (/v1/chat/completions) or
Anthropic-style (/v1/messages) API formats.
Changes:
- Add azure-foundry to PROVIDER_REGISTRY (auth.py)
- Add azure-foundry overlay in HERMES_OVERLAYS (providers.py)
- Add empty model list for azure-foundry (models.py)
- Add _model_flow_azure_foundry() interactive setup (main.py)
- Add azure-foundry runtime resolution with api_mode support (runtime_provider.py)
- Add AZURE_FOUNDRY_API_KEY and AZURE_FOUNDRY_BASE_URL env vars (config.py)
Usage:
hermes model -> More providers -> Azure Foundry
The setup wizard prompts for:
- Endpoint URL
- API format (OpenAI or Anthropic-style)
- API key
- Model name
Configuration is saved to config.yaml (model.provider, model.base_url,
model.api_mode, model.default) and ~/.hermes/.env (AZURE_FOUNDRY_API_KEY).
Fixes#15779. Custom-provider per-model context_length (`custom_providers[].models.<id>.context_length`) is now honored across every resolution path, not just agent startup. Also adds 256K as the top probe tier and default fallback.
## What changed
New helper `hermes_cli.config.get_custom_provider_context_length()` — single source of truth for the per-model override lookup, with trailing-slash-insensitive base-url matching.
`agent.model_metadata.get_model_context_length()` gains an optional `custom_providers=` kwarg (step 0b — runs after explicit `config_context_length` but before every other probe).
Wired through five call sites that previously either duplicated the lookup or ignored it entirely:
- `run_agent.py` startup — refactored to use the new helper (dedups legacy inline loop, keeps invalid-value warning)
- `AIAgent.switch_model()` — re-reads custom_providers from live config on every /model switch
- `hermes_cli.model_switch.resolve_display_context_length()` — new `custom_providers=` kwarg
- `gateway/run.py` /model confirmation (picker callback + text path)
- `gateway/run.py` `_format_session_info` (/info)
## Context probe tiers
`CONTEXT_PROBE_TIERS = [256_000, 128_000, 64_000, 32_000, 16_000, 8_000]` — was `[128_000, ...]`. `DEFAULT_FALLBACK_CONTEXT` follows tier[0], so unknown models now default to 256K. The stale `128000` literal in the OpenRouter metadata-miss path is replaced with `DEFAULT_FALLBACK_CONTEXT` for consistency.
## Repro (from #15779)
```yaml
custom_providers:
- name: my-custom-endpoint
base_url: https://example.invalid/v1
model: gpt-5.5
models:
gpt-5.5:
context_length: 1050000
```
`/model gpt-5.5 --provider custom:my-custom-endpoint` → previously "Context: 128,000", now "Context: 1,050,000".
## Tests
- `tests/hermes_cli/test_custom_provider_context_length.py` — new file, 19 tests covering the helper, step-0b integration, and the 256K tier invariants
- `tests/hermes_cli/test_model_switch_context_display.py` — added regression tests for #15779 through the display resolver
- `tests/gateway/test_session_info.py` — updated default-fallback assertion (128K → 256K)
- `tests/agent/test_model_metadata.py` — updated tier assertions for the new top tier
task.cancel() can't preempt the run_in_executor thread running
run_conversation(), so we rely on agent.interrupt() to wake the loop.
Without a timeout, a slow/unresponsive interrupt blocks the HTTP
response indefinitely. Wrap the await in wait_for(shield(task), 5.0)
and log a warning on timeout.
Also tidy one extra space in the module docstring's /stop entry.
Add ability to interrupt a running agent via the runs API. Previously
/v1/runs could start a run and subscribe to events, but there was no
way to cancel it. The new endpoint stores agent and task references
during execution, calls agent.interrupt() to stop LLM calls, then
cancels the asyncio task.
Includes 15 tests covering start, events, and stop scenarios.
- resolveEditor() now returns argv (string[]) so EDITOR='code --wait'
and VISUAL='emacsclient -t' tokenize correctly into spawnSync's
separate command + args. Previously the whole string was passed as
argv[0] and would ENOENT.
- Skip the POSIX X_OK PATH walk on Windows; return ['notepad.exe']
there since fs.constants.X_OK is not meaningful and PATHEXT-based
resolution would need its own implementation.
- Surface openEditor() rejections via actions.sys instead of letting
them become unhandled promise rejections in the useInput callback.
- Hotkey docs/comment now say Cmd/Ctrl+G to match isAction()'s
platform-action-modifier behavior (Cmd on macOS, Ctrl elsewhere).
When the user interrupts a long-running task, prompt_toolkit tries to
flush stdout during emergency shutdown. If stdout is in a broken state
(redirected to /dev/null, pipe closed, terminal gone), the flush raises
`OSError: [Errno 5] Input/output error` which propagates unhandled and
crashes the CLI.
Two defense layers:
1. `_suppress_closed_loop_errors`: add `OSError` with `errno.EIO` to
the asyncio exception handler, matching the existing pattern for
`RuntimeError("Event loop is closed")` and `KeyError("is not
registered")`.
2. Outer `except (KeyError, OSError)` block: add `errno.EIO` check
before the existing string-match guards, silently suppressing the
error instead of printing a misleading stdin-related message.
Fixes#13710.
- editor.ts: collapse two private helpers into one flatMap-driven lookup,
keep `isExecutable` as the only named primitive, document the fallback
chain with prompt_toolkit parity
- editor.test.ts: hoist the `exe` helper out of `describe`, drop the
empty afterEach + dead mkdir branch, materialize expected paths before
the resolveEditor call so argument evaluation order doesn't bite
- useComposerState.openEditor: rmSync the mkdtemp dir (was leaking),
early-return on bad exit / empty buffer, run cleanup in finally
- useInputHandlers: cheap `ch.toLowerCase() === 'g'` guard before the
modifier check
- hermes-ink/screen.ts: pick up `npm run fix` import-sort cleanup so
lint passes
Base CLI's editor UX was better because prompt_toolkit picks the system
editor first, then friendly terminal editors before vi. Do not override
that with a vim-first chain.
Keep the CLI on prompt_toolkit's picker and only set tempfile_suffix='.md'
to avoid the complex-tempfile EEXIST path. Update the TUI resolver to
match prompt_toolkit's fallback order: $VISUAL, $EDITOR, editor, nano,
pico, vi, emacs.
Setting buffer.tempfile = 'prompt.md' pushed prompt_toolkit into its
complex-tempfile path, which creates a temp dir and then calls
os.makedirs() on that same path when no subdirectory is present. That
raises EEXIST before the editor can launch.
Keep prompt_toolkit on the simple tempfile path with .md suffix, and
make the editor fallback chain explicit on both surfaces:
$VISUAL -> $EDITOR -> nvim -> vim -> vi -> nano.
The cherry-picked approach serialized the UI-shaped transcript on the Node
side, producing a third JSON format alongside cli.py save_conversation and
tui_gateway session.save. Simpler to call the existing session.save method,
which already writes the canonical agent history (raw OpenAI messages +
model) to an absolute-path file.
- /save still short-circuits before the slash worker
- Empty transcript -> 'no conversation yet'
- No active session -> 'no active session - nothing to save'
- Otherwise: rpc('session.save', {session_id}) and echo back the file path
- Tests updated to assert RPC contract; new test covers the no-sid case
prompt_toolkit's default editor list is: $VISUAL, $EDITOR, /usr/bin/editor,
/usr/bin/nano, /usr/bin/pico, /usr/bin/vi, /usr/bin/emacs — so when
neither env var is set, the base CLI launched nano. The TUI fell back
to a literal 'vi'. Same Ctrl+G keystroke, two different editors.
Pick the same chain on both surfaces:
$VISUAL → $EDITOR → vim → vi → nano
CLI: override input_area.buffer._open_file_in_editor on the TextArea
once at app build time. Local to that buffer; doesn't touch
os.environ or affect other subprocesses.
TUI: extract resolveEditor() into ui-tui/src/lib/editor.ts. PATH walk
with accessSync(X_OK), no shelling out. Six-line unit test verifies
the priority order and the multi-entry PATH walk.
The raw-template lookup added in PR #15817 went through
`get_compatible_custom_providers(read_raw_config())`, which calls
`_normalize_custom_provider_entry` → `urlparse(base_url)`. Any
entry whose `base_url` is itself an env-ref (`${NEURALWATT_API_BASE}`)
was dropped as 'not a valid URL', so `api_key_ref` stayed empty and the
resolved secret was still written to `model.api_key` — the exact case
the original Discord report described.
Replace the normalizer-gated lookup with a direct read of
`raw['custom_providers']` and `raw['providers']`, indexed by name
(case-insensitive, optionally qualified by model) so the loaded
(expanded) entry can be matched regardless of how `base_url` is
written.
Add an integration regression test driving the real
`select_provider_and_model` entry point with the Discord-reported
NeuralWatt config (`${VAR}` in both `base_url` and `api_key`).
This test fails on the PR-only fix and passes with the broadened
lookup.
Base CLI was handing prompt_toolkit's Buffer.open_in_editor() a default
config — Buffer.tempfile_suffix and .tempfile both empty — so it
created /tmp/tmpXXXXXX with no extension. nano/vim/helix all key
syntax highlighting off the file extension, so the buffer rendered
plain.
The TUI already writes to <mkdtemp>/prompt.md and gets full markdown
highlighting + a sensible title bar. Set buffer.tempfile = 'prompt.md'
on the TextArea so prompt_toolkit's complex-tempfile path produces
<mkdtemp>/prompt.md to match. shutil.rmtree cleanup is built-in.
When switching models on a custom endpoint (ollama-launch):
- Same-provider switches no longer re-resolve credentials (fixes base_url
being lost for 'custom' provider on subsequent switches)
- Named providers (ollama-launch) are resolved via user_providers so
switch_model can find their base_url from config
- Models not in the /v1/models probe but present in the user's saved
provider config are accepted with a warning instead of rejected
- CLI /model and TUI /model both pass user_providers/custom_providers
to switch_model so the config model list is available for validation
Closes#15088
Same problem as the TUI: Cursor and VSCode bind Ctrl+G to "Find Next"
at the editor level, so the keystroke never reaches the terminal and
the prompt_toolkit-driven Hermes CLI sees nothing.
Register ('escape', 'g') alongside the existing 'c-g' on the same
handler so the editor handoff works inside Cursor/VSCode too. The
filter (no clarify/approval/sudo/secret prompt active) is unchanged.
VSCode and Cursor bind Ctrl+G to "Find Next" at the editor level, so
the keystroke never reaches the embedded terminal — Ctrl+G to open
\$EDITOR was effectively dead inside those IDEs.
Alt+G is unbound in both editors and reaches the TUI cleanly as
`\x1bg` → `key.meta && ch === 'g'` after parse-keypress. Accept it
alongside the existing isAction(key, ch, 'g') check, and document the
fallback in README + the hotkeys panel.
The Ctrl+G handler was toggling the alt-screen by hand
(`\x1b[?1049l` ... `\x1b[?1049h`) without releasing stdin or kitty
keyboard mode, so the launched editor would lose keystrokes (Ink kept
swallowing them) and editors that don't speak CSI-u (e.g. nano) would
print "Unknown sequence" for every Ctrl-key.
Switch to `withInkSuspended` from @hermes/ink, the same helper
`/setup` already uses. It pauses Ink, removes stdin listeners, drops
raw mode, disables kitty/modifyOtherKeys + mouse + focus reporting,
runs the editor, then restores everything with a full repaint.
Previously _copy_reasoning_content_for_api only padded reasoning_content
when the assistant message had tool_calls. DeepSeek V4 thinking mode
requires the field on every assistant turn, including plain text replies
without tool_calls.
- Remove the 'source_msg.get("tool_calls") and' guard
- Update test: plain assistant turns now get padded for DeepSeek/Kimi
Fixes#15213
- webhooks.md: adds a Video Tutorial section under the intro with a
responsive YouTube iframe (WNYe5mD4fY8).
- configuration.md: adds a Video Tutorial subsection under Auxiliary
Models with a responsive YouTube iframe (NoF-YajElIM).
Both use a 16:9 aspect-ratio wrapper so the embeds scale cleanly on
mobile. Verified with `npm run build` — MDX parses clean, no new
warnings or broken links introduced.
Adds a 'Video Guide' section pointing at the walkthrough of a Hermes agent
abliterating Gemma with OBLITERATUS, so the agent can surface it when the
user wants a visual overview before running the workflow.
> The Curator release — Hermes Agent now maintains itself. An autonomous background Curator grades, prunes, and consolidates your skill library on its own schedule. The self-improvement loop that reviews what to save got a substantial upgrade. Four new inference providers, a 18th messaging platform, a 19th via Teams plugin, native Spotify + Google Meet integrations, ComfyUI and TouchDesigner-MCP moved from optional to bundled-by-default, and a ~57% cut to visible TUI cold start.
---
## ✨ Highlights
- **Autonomous Curator** — `hermes curator` runs as a background agent on the gateway's cron ticker (7-day cycle default). It grades your skill library, consolidates related skills, prunes dead ones, and writes per-run reports to `logs/curator/run.json` + `REPORT.md`. Archived skills are classified consolidated-vs-pruned via model + heuristic. Defense-in-depth gates protect bundled/hub skills from mutation. Unified under `auxiliary.curator` — pick the curator's model in `hermes model`, manage it from the dashboard. `hermes curator status` ranks skills by usage (most-used / least-used). ([#17277](https://github.com/NousResearch/hermes-agent/pull/17277), [#17307](https://github.com/NousResearch/hermes-agent/pull/17307), [#17941](https://github.com/NousResearch/hermes-agent/pull/17941), [#17868](https://github.com/NousResearch/hermes-agent/pull/17868), [#18033](https://github.com/NousResearch/hermes-agent/pull/18033))
- **Self-improvement loop — substantially upgraded** — The background review fork (the core of Hermes' self-improvement: after each turn it decides what memories/skills to save or update) is now class-first (rubric-based rather than free-form), active-update biased (prefers the skill the agent just loaded), handles `references/`/`templates/` sub-files, and properly inherits the parent's live runtime (provider, model, credentials actually propagate). Restricted to memory + skills toolsets so it can't sprawl. Memory providers shut down cleanly. Prior-turn tool messages excluded from the summary so the fork sees a clean context. ([#16026](https://github.com/NousResearch/hermes-agent/pull/16026), [#17213](https://github.com/NousResearch/hermes-agent/pull/17213), [#16099](https://github.com/NousResearch/hermes-agent/pull/16099), [#16569](https://github.com/NousResearch/hermes-agent/pull/16569), [#16204](https://github.com/NousResearch/hermes-agent/pull/16204), [#15057](https://github.com/NousResearch/hermes-agent/pull/15057))
- **Skill integrations — major expansion** — **ComfyUI v5** with official CLI + REST + hardware-gated local install, moved from optional to **built-in by default** ([#17610](https://github.com/NousResearch/hermes-agent/pull/17610), [#17631](https://github.com/NousResearch/hermes-agent/pull/17631), [#17734](https://github.com/NousResearch/hermes-agent/pull/17734)). **TouchDesigner-MCP** bundled by default, expanded with GLSL, post-FX, audio, geometry, and 9 new reference docs ([#16753](https://github.com/NousResearch/hermes-agent/pull/16753), [#16624](https://github.com/NousResearch/hermes-agent/pull/16624), [#16768](https://github.com/NousResearch/hermes-agent/pull/16768) — @kshitijk4poor + @SHL0MS). **Humanizer** skill ports a text-cleaner that strips AI-isms ([#16787](https://github.com/NousResearch/hermes-agent/pull/16787)). **claude-design** HTML artifact skill + design-md (Google DESIGN.md spec) + airtable salvage + `skill_manage` edits in `external_dirs` + direct-URL skill install + `/reload-skills` slash command. ([#16358](https://github.com/NousResearch/hermes-agent/pull/16358), [#14876](https://github.com/NousResearch/hermes-agent/pull/14876), [#16291](https://github.com/NousResearch/hermes-agent/pull/16291), [#17512](https://github.com/NousResearch/hermes-agent/pull/17512), [#16323](https://github.com/NousResearch/hermes-agent/pull/16323), [#17744](https://github.com/NousResearch/hermes-agent/pull/17744))
- **LM Studio — first-class provider** — upgraded from a custom-endpoint alias to a full-blown native provider: dedicated auth, `hermes doctor` checks, reasoning transport, live `/models` listing. (Salvage of @kshitijk4poor's #17061.) ([#17102](https://github.com/NousResearch/hermes-agent/pull/17102))
- **Four more new inference providers** — **GMI Cloud** (first-class, salvage of #11955 — @isaachuangGMICLOUD), **Azure AI Foundry** with auto-detection, **MiniMax OAuth** with PKCE browser flow (salvage #15203), **Tencent Tokenhub** (salvage of #16860). ([#16663](https://github.com/NousResearch/hermes-agent/pull/16663), [#15845](https://github.com/NousResearch/hermes-agent/pull/15845), [#17524](https://github.com/NousResearch/hermes-agent/pull/17524), [#16960](https://github.com/NousResearch/hermes-agent/pull/16960))
- **Pluggable gateway platforms + Microsoft Teams** — the gateway is now a plugin host. Drop-in messaging adapters live outside the core, and Microsoft Teams is the first plugin-shipped platform. (Salvage of #17664.) ([#17751](https://github.com/NousResearch/hermes-agent/pull/17751), [#17828](https://github.com/NousResearch/hermes-agent/pull/17828))
- **Tencent 元宝 (Yuanbao) — 18th messaging platform** — native gateway adapter with text + media delivery. ([#16298](https://github.com/NousResearch/hermes-agent/pull/16298), [#17424](https://github.com/NousResearch/hermes-agent/pull/17424))
- **Models dashboard tab + in-browser model config** — rich per-model analytics, switch main + auxiliary models from the dashboard. ([#17745](https://github.com/NousResearch/hermes-agent/pull/17745), [#17802](https://github.com/NousResearch/hermes-agent/pull/17802))
- **Remote model catalog manifest** — OpenRouter + Nous Portal model catalogs are now pulled from a remote manifest so new models show up without a release. ([#16033](https://github.com/NousResearch/hermes-agent/pull/16033))
- **Native multimodal image routing** — images now route based on the model's actual vision capability rather than provider defaults. ([#16506](https://github.com/NousResearch/hermes-agent/pull/16506))
- **Gateway media parity** — native multi-image sending across Telegram, Discord, Slack, Mattermost, Email, and Signal; centralized audio routing with FLAC support + Telegram document fallback. ([#17909](https://github.com/NousResearch/hermes-agent/pull/17909), [#17833](https://github.com/NousResearch/hermes-agent/pull/17833))
- **TUI catches up to (and past) the classic CLI** — LaTeX rendering (@austinpickett), `/reload` .env hot-reload, pluggable busy-indicator styles (@OutThisLife, #13610), opt-in auto-resume of last session, expanded light-terminal auto-detection, session delete from `/resume` picker with `d`, modified mouse-wheel line scroll, and a `/mouse` toggle that kills ConPTY's phantom mouse injection (@kevin-ho). ([#17175](https://github.com/NousResearch/hermes-agent/pull/17175), [#17286](https://github.com/NousResearch/hermes-agent/pull/17286), [#17150](https://github.com/NousResearch/hermes-agent/pull/17150), [#17130](https://github.com/NousResearch/hermes-agent/pull/17130), [#17113](https://github.com/NousResearch/hermes-agent/pull/17113), [#17668](https://github.com/NousResearch/hermes-agent/pull/17668), [#17669](https://github.com/NousResearch/hermes-agent/pull/17669), [#15488](https://github.com/NousResearch/hermes-agent/pull/15488))
- **Observability + achievements plugins** — bundled Langfuse observability plugin (salvage #16845) + bundled hermes-achievements plugin that scans full session history. ([#16917](https://github.com/NousResearch/hermes-agent/pull/16917), [#17754](https://github.com/NousResearch/hermes-agent/pull/17754))
- **TTS provider registry + Piper local TTS** — pluggable `tts.providers.<name>` registry; Piper ships as a native local TTS provider. (Closes #8508.) ([#17843](https://github.com/NousResearch/hermes-agent/pull/17843), [#17885](https://github.com/NousResearch/hermes-agent/pull/17885))
- **Vercel Sandbox backend** — Vercel sandboxes as an execute_code/terminal backend (@kshitijk4poor). ([#17445](https://github.com/NousResearch/hermes-agent/pull/17445))
- **Secret redaction off by default** — default flipped to off. Prevents the long-standing patch-corruption incidents where fake secret-shaped substrings mangled tool outputs. Opt in via `redaction.enabled: true` when you need it. ([#16794](https://github.com/NousResearch/hermes-agent/pull/16794))
- **Configurable prompt cache TTL** — `prompt_caching.cache_ttl` (5m default, 1h opt-in — cost savings for bursty sessions that keep cache warm). Salvage of #12659. ([#15065](https://github.com/NousResearch/hermes-agent/pull/15065))
---
## 🧠 Autonomous Curator & Self-Improvement Loop
### Curator — autonomous skill maintenance
- **`hermes curator` as a background agent** — runs on the gateway's cron ticker, 7-day cycle by default, umbrella-first prompt, inherits parent config, unbounded iterations ([#17277](https://github.com/NousResearch/hermes-agent/pull/17277) — issue #7816)
- **Per-run reports** — `logs/curator/run.json` + `REPORT.md` per cycle ([#17307](https://github.com/NousResearch/hermes-agent/pull/17307))
- **Consolidated vs pruned classification** — archived skills split with model + heuristic ([#17941](https://github.com/NousResearch/hermes-agent/pull/17941))
- **`hermes curator status`** — ranks skills by usage, shows most-used and least-used ([#18033](https://github.com/NousResearch/hermes-agent/pull/18033))
- **Unified under `auxiliary.curator`** — pick the model in `hermes model`, configure from the dashboard ([#17868](https://github.com/NousResearch/hermes-agent/pull/17868))
- **Documentation** — dedicated curator feature page on the docs site ([#17563](https://github.com/NousResearch/hermes-agent/pull/17563))
- Fix: seed defaults on update, create `logs/curator/` directory, defer fire import ([#17927](https://github.com/NousResearch/hermes-agent/pull/17927))
- Fix: scan nested archive subdirs in `restore_skill` (@0xDevNinja) ([#17951](https://github.com/NousResearch/hermes-agent/pull/17951))
- Fix: use actual skill activity in curator status (@y0shua1ee) ([#17953](https://github.com/NousResearch/hermes-agent/pull/17953))
- Fix: `skill_manage` refuses writes on pinned skills; pinning now blocks curator writes ([#17562](https://github.com/NousResearch/hermes-agent/pull/17562), [#17578](https://github.com/NousResearch/hermes-agent/pull/17578))
- **Class-first skill-review prompt** — rubric-based grading rather than free-form "should this update" ([#16026](https://github.com/NousResearch/hermes-agent/pull/16026))
- **Active-update bias** — prefers updating skills the agent just loaded, handles `references/` + `templates/` sub-files ([#17213](https://github.com/NousResearch/hermes-agent/pull/17213))
- **Fork inherits parent's live runtime** — provider, model, credentials actually propagate now ([#16099](https://github.com/NousResearch/hermes-agent/pull/16099))
- **Scoped toolsets** — review fork restricted to memory + skills (no shell, no web) ([#16569](https://github.com/NousResearch/hermes-agent/pull/16569))
### Skill integrations — newly bundled or promoted
- **ComfyUI v5** — official CLI + REST + hardware-gated local install; **moved from optional to built-in** ([#17610](https://github.com/NousResearch/hermes-agent/pull/17610), [#17631](https://github.com/NousResearch/hermes-agent/pull/17631), [#17734](https://github.com/NousResearch/hermes-agent/pull/17734), [#17612](https://github.com/NousResearch/hermes-agent/pull/17612))
- **TouchDesigner-MCP** — **bundled by default** ([#16753](https://github.com/NousResearch/hermes-agent/pull/16753) — @kshitijk4poor), expanded with GLSL, post-FX, audio, geometry references ([#16624](https://github.com/NousResearch/hermes-agent/pull/16624)), 9 new reference docs ([#16768](https://github.com/NousResearch/hermes-agent/pull/16768) — @SHL0MS)
- **Humanizer** — strips AI-isms from text ([#16787](https://github.com/NousResearch/hermes-agent/pull/16787))
- **claude-design** — HTML artifact skill with disambiguation from other design skills ([#16358](https://github.com/NousResearch/hermes-agent/pull/16358))
- Point agent at `hermes-agent` skill + docs site for Hermes questions ([#16535](https://github.com/NousResearch/hermes-agent/pull/16535))
---
## 🏗️ Core Agent & Architecture
### Provider & Model Support
#### New providers
- **GMI Cloud** — first-class API-key provider on par with Arcee/Kilocode/Xiaomi (salvage of #11955 — @isaachuangGMICLOUD) ([#16663](https://github.com/NousResearch/hermes-agent/pull/16663))
- **Azure AI Foundry** — auto-detection, full wiring ([#15845](https://github.com/NousResearch/hermes-agent/pull/15845))
- **LM Studio** — upgraded from custom-endpoint alias to first-class provider: dedicated auth, doctor checks, reasoning transport, live `/models` (salvage of #17061 — @kshitijk4poor) ([#17102](https://github.com/NousResearch/hermes-agent/pull/17102))
- **MiniMax OAuth** — PKCE browser flow with full OAuth integration (salvage #15203) ([#17524](https://github.com/NousResearch/hermes-agent/pull/17524))
- **Tencent Tokenhub** — new provider (salvage of #16860) ([#16960](https://github.com/NousResearch/hermes-agent/pull/16960))
#### Model catalog
- **Remote model catalog manifest** — OpenRouter + Nous Portal catalogs pulled from remote manifest so new models show up without a release ([#16033](https://github.com/NousResearch/hermes-agent/pull/16033))
-`openai/gpt-5.5` and `gpt-5.5-pro` added to OpenRouter + Nous Portal ([#15343](https://github.com/NousResearch/hermes-agent/pull/15343))
-`deepseek-v4-pro` and `deepseek-v4-flash` added ([#14934](https://github.com/NousResearch/hermes-agent/pull/14934))
-`qwen3.6-plus` added to Alibaba-supported models ([#16896](https://github.com/NousResearch/hermes-agent/pull/16896))
- Gemini free-tier keys blocked at setup with 429 guidance surfacing ([#15100](https://github.com/NousResearch/hermes-agent/pull/15100))
-`/fast` whitelist broadened to all OpenAI + Anthropic models ([#16883](https://github.com/NousResearch/hermes-agent/pull/16883))
-`auxiliary.extra_body.reasoning` translates into Codex Responses API ([#17004](https://github.com/NousResearch/hermes-agent/pull/17004))
-`hermes fallback` command for managing fallback providers ([#16052](https://github.com/NousResearch/hermes-agent/pull/16052))
### Agent Loop & Conversation
- **Native multimodal image routing** — based on model vision capability, not provider defaults ([#16506](https://github.com/NousResearch/hermes-agent/pull/16506))
- **Delegate `child_timeout_seconds` default bumped to 600s** ([#14809](https://github.com/NousResearch/hermes-agent/pull/14809))
- **Diagnostic dump when subagent times out with 0 API calls** ([#15105](https://github.com/NousResearch/hermes-agent/pull/15105))
- **Gateway busts cached agent on compression/context_length config edits** ([#17008](https://github.com/NousResearch/hermes-agent/pull/17008))
- **Opt-in runtime-metadata footer on final replies** ([#17026](https://github.com/NousResearch/hermes-agent/pull/17026))
- Fix: retry on `json.JSONDecodeError` instead of treating as local validation error ([#15107](https://github.com/NousResearch/hermes-agent/pull/15107))
- Fix: handle unescaped control chars in `tool_call.arguments` ([#15356](https://github.com/NousResearch/hermes-agent/pull/15356))
- Fix: persist streamed `reasoning_content` on assistant turns (#16844) ([#16892](https://github.com/NousResearch/hermes-agent/pull/16892))
- Fix: cancel coroutine on timeout so worker thread exits; full traceback on tool failure ([#17428](https://github.com/NousResearch/hermes-agent/pull/17428))
- Fix: rename `[SYSTEM:` → `[IMPORTANT:` in all user-injected markers (dodges Azure content filter) ([#16114](https://github.com/NousResearch/hermes-agent/pull/16114))
### Compression
- **Retry summary on main model for unknown errors before giving up** ([#16774](https://github.com/NousResearch/hermes-agent/pull/16774))
- **Notify users when configured aux model fails even if main-model fallback recovers** ([#16775](https://github.com/NousResearch/hermes-agent/pull/16775))
-`/compress` wrapped in `_busy_command` to block input during compression ([#15388](https://github.com/NousResearch/hermes-agent/pull/15388))
- Fix: reserve system + tools headroom when aux binds threshold ([#15631](https://github.com/NousResearch/hermes-agent/pull/15631))
- Fix: use text-char sum for multimodal token estimation in `_find_tail_cut_by_tokens` ([#16369](https://github.com/NousResearch/hermes-agent/pull/16369))
### Session, Memory & State
- **Trigram FTS5 index for CJK search, replace LIKE fallback** (@alt-glitch) ([#16651](https://github.com/NousResearch/hermes-agent/pull/16651))
- **Index `tool_name` + `tool_calls` in FTS5, with repair + migration** (salvages #16866) ([#16914](https://github.com/NousResearch/hermes-agent/pull/16914))
- **Checkpoints: auto-prune orphan and stale shadow repos at startup** ([#16303](https://github.com/NousResearch/hermes-agent/pull/16303))
- **Memory providers notified on mid-process session_id rotation** (#6672) ([#17409](https://github.com/NousResearch/hermes-agent/pull/17409))
- Fix: quote underscored terms in FTS5 query sanitization ([#16915](https://github.com/NousResearch/hermes-agent/pull/16915))
- Fix: generalize unsupported-parameter detector and harden `max_tokens` retry ([#15633](https://github.com/NousResearch/hermes-agent/pull/15633))
---
## 📱 Messaging Platforms (Gateway)
### New Platforms
- **Microsoft Teams (19th platform)** — as a plugin, + xdist collision guard ([#17828](https://github.com/NousResearch/hermes-agent/pull/17828))
- **Yuanbao (Tencent 元宝, 18th platform)** — native adapter with text + media delivery ([#16298](https://github.com/NousResearch/hermes-agent/pull/16298), [#17424](https://github.com/NousResearch/hermes-agent/pull/17424), [#16880](https://github.com/NousResearch/hermes-agent/pull/16880))
### Pluggable Gateway Platforms
- **Drop-in messaging adapters** — the gateway is now a plugin host for platforms (salvage of #17664) ([#17751](https://github.com/NousResearch/hermes-agent/pull/17751))
### Telegram
- **Chat allowlists for groups and forums** (@web3blind) ([#15027](https://github.com/NousResearch/hermes-agent/pull/15027))
- **Send fresh finals for stale preview streams** (port openclaw#72038) ([#16261](https://github.com/NousResearch/hermes-agent/pull/16261))
- **Auto-spawn local Chromium for LAN/localhost URLs** when cloud provider is configured ([#16136](https://github.com/NousResearch/hermes-agent/pull/16136))
### Execute code / Terminal
- **Vercel Sandbox backend** for `execute_code` / terminal (@kshitijk4poor) ([#17445](https://github.com/NousResearch/hermes-agent/pull/17445))
- **Collapse subagent `task_id`s to shared container** ([#16177](https://github.com/NousResearch/hermes-agent/pull/16177))
- **Docker: run container as host user** to avoid root-owned bind mounts (@benbarclay) ([#17305](https://github.com/NousResearch/hermes-agent/pull/17305))
- Fix: close file descriptor in `LocalEnvironment._update_cwd` ([#17300](https://github.com/NousResearch/hermes-agent/pull/17300))
- Fix: SSH — prevent tar from overwriting remote home dir permissions ([#17898](https://github.com/NousResearch/hermes-agent/pull/17898), [#17867](https://github.com/NousResearch/hermes-agent/pull/17867))
### Image generation
- See Provider section for updates; no new image providers this window.
### TTS / Voice
- **Pluggable TTS provider registry** under `tts.providers.<name>` ([#17843](https://github.com/NousResearch/hermes-agent/pull/17843))
- **Piper** as native local TTS provider (closes #8508) ([#17885](https://github.com/NousResearch/hermes-agent/pull/17885))
- **Voice mode CLI parity in the TUI** — VAD loop + TTS + crash forensics ([#14810](https://github.com/NousResearch/hermes-agent/pull/14810))
- Fix: vision — use HERMES_HOME-based cache dir instead of cwd ([#17719](https://github.com/NousResearch/hermes-agent/pull/17719))
### Cron
- **Honor `hermes tools` config for the cron platform** ([#14798](https://github.com/NousResearch/hermes-agent/pull/14798))
- **Delete sessions from `/resume` picker with `d`** (@OutThisLife) ([#17668](https://github.com/NousResearch/hermes-agent/pull/17668))
- **Line-by-line scroll on modified mouse wheel** (@OutThisLife) ([#17669](https://github.com/NousResearch/hermes-agent/pull/17669))
- **Delete queued message while editing with ctrl-x / cancel with esc** (@OutThisLife) ([#16707](https://github.com/NousResearch/hermes-agent/pull/16707))
- **Per-section visibility for the details accordion** (@OutThisLife) ([#14968](https://github.com/NousResearch/hermes-agent/pull/14968))
- **Surface `/queue`, `/bg`, `/steer` in agent-running placeholder** ([#16118](https://github.com/NousResearch/hermes-agent/pull/16118))
### Setup / onboarding
- **Auto-reconfigure on existing installs** ([#15879](https://github.com/NousResearch/hermes-agent/pull/15879))
- **Contextual first-touch hints for `/busy` and `/verbose`** ([#16046](https://github.com/NousResearch/hermes-agent/pull/16046))
- **Cost-saving tips from the April 30 tip-of-the-day** ([#17841](https://github.com/NousResearch/hermes-agent/pull/17841))
- **Hyperlink startup banner title to the latest GitHub Release** ([#14945](https://github.com/NousResearch/hermes-agent/pull/14945))
### Update / backup
- **Snapshot pairing data before `git pull`** ([#16383](https://github.com/NousResearch/hermes-agent/pull/16383))
- **Auto-backup HERMES_HOME before `hermes update`** (opt-in, off by default) ([#16539](https://github.com/NousResearch/hermes-agent/pull/16539), [#16566](https://github.com/NousResearch/hermes-agent/pull/16566))
- **Exclude `checkpoints/` from backups** ([#16572](https://github.com/NousResearch/hermes-agent/pull/16572))
- **Exclude SQLite WAL/SHM/journal sidecars from backups** ([#16576](https://github.com/NousResearch/hermes-agent/pull/16576))
- **Installer FHS layout for root installs on Linux** ([#15608](https://github.com/NousResearch/hermes-agent/pull/15608))
- Fix: kill stale dashboards instead of warning ([#17832](https://github.com/NousResearch/hermes-agent/pull/17832))
- Fix: show correct update status on nix-built hermes ([#17550](https://github.com/NousResearch/hermes-agent/pull/17550))
### Slash-command housekeeping
- Refactor: drop `/provider`, `/plan` handler, and clean up slash registry ([#15047](https://github.com/NousResearch/hermes-agent/pull/15047))
- **Lazily seed virtual history heights** ([#16523](https://github.com/NousResearch/hermes-agent/pull/16523))
---
## 🔒 Security & Reliability
- **Secret redaction off by default** — stops corrupting patches / API payloads with fake-key substitutions. Opt in via `redaction.enabled: true` ([#16794](https://github.com/NousResearch/hermes-agent/pull/16794))
- **`[SYSTEM:` → `[IMPORTANT:`** in all user-injected markers (Azure content filter dodge) ([#16114](https://github.com/NousResearch/hermes-agent/pull/16114))
- **Hardline blocklist for unrecoverable commands** ([#15878](https://github.com/NousResearch/hermes-agent/pull/15878))
- **Canonical `mask_secret` helper; fix status.py DIM drift** ([#17207](https://github.com/NousResearch/hermes-agent/pull/17207))
- **Sweep expired paste.rs uploads on a real timer** ([#16431](https://github.com/NousResearch/hermes-agent/pull/16431))
- **Preserve symlinks during atomic file writes** ([#16980](https://github.com/NousResearch/hermes-agent/pull/16980))
- **Probe `/dev/tty` by opening it, not bare existence** ([#17024](https://github.com/NousResearch/hermes-agent/pull/17024))
---
## 🐛 Notable Bug Fixes
This window includes 360 `fix:` PRs. Selected highlights from across the stack:
- **Background review fork inherits parent's live runtime** — provider/model/creds now propagate correctly ([#16099](https://github.com/NousResearch/hermes-agent/pull/16099))
- **Background reviewer scoped to memory + skills toolsets** — no more accidental web/shell escapes ([#16569](https://github.com/NousResearch/hermes-agent/pull/16569))
- **Compression recovery** — retry on main before giving up; notify user when aux fails ([#16774](https://github.com/NousResearch/hermes-agent/pull/16774), [#16775](https://github.com/NousResearch/hermes-agent/pull/16775))
- **`croniter` promoted to a core dependency** ([#17577](https://github.com/NousResearch/hermes-agent/pull/17577))
- **Discord tool `limit` parameter coerced to int** before `min()` call ([#16319](https://github.com/NousResearch/hermes-agent/pull/16319))
The vast majority of the 360 fixes landed in the streaming/compression/tool-calling paths across all providers — DeepSeek, Kimi, Moonshot, GLM, Qwen, MiniMax, Gemini, Anthropic, OpenAI — alongside TUI polish (resize, scroll, sticky-prompt) and gateway platform-specific edge cases.
---
## 🧪 Testing & CI
- Hermetic test parity (`scripts/run_tests.sh`) held across this window
- **Microsoft Teams xdist collision guard** — prevents worker collisions when Teams platform tests run in parallel ([#17828](https://github.com/NousResearch/hermes-agent/pull/17828))
- Chore: remove unused imports and dead locals (ruff F401, F841) ([#17010](https://github.com/NousResearch/hermes-agent/pull/17010))
---
## 📚 Documentation
- **Curator feature page** added to docs site ([#17563](https://github.com/NousResearch/hermes-agent/pull/17563))
- **Document pin also blocking `skill_manage` writes** ([#17578](https://github.com/NousResearch/hermes-agent/pull/17578))
- **Direct-URL skill install documented** across features, reference, guide, and `hermes-agent` skill ([#16355](https://github.com/NousResearch/hermes-agent/pull/16355))
- **Hooks tutorial — build a BOOT.md startup checklist** (replaces the removed built-in hook) ([#17202](https://github.com/NousResearch/hermes-agent/pull/17202))
- **ComfyUI docs: ask local vs cloud FIRST before hardware check** ([#17612](https://github.com/NousResearch/hermes-agent/pull/17612))
- **Obliteratus skill: link YouTube video guide in SKILL.md** ([#15808](https://github.com/NousResearch/hermes-agent/pull/15808))
- Per-skill docs pages generated for bundled + optional skills; ASCII art code blocks auto-wrapped ([#14929](https://github.com/NousResearch/hermes-agent/pull/14929), [#16497](https://github.com/NousResearch/hermes-agent/pull/16497))
---
## ⚖️ Removed / Reverted
- **Kanban multi-profile collaboration board** — landed in #16081, reverted in ([#16098](https://github.com/NousResearch/hermes-agent/pull/16098)) while the design is reworked
- **computer-use cua-driver** — 3 preparatory PRs landed then were reverted in ([#16927](https://github.com/NousResearch/hermes-agent/pull/16927))
- **BOOT.md built-in hook** removed ([#17093](https://github.com/NousResearch/hermes-agent/pull/17093)); the hooks tutorial ([#17202](https://github.com/NousResearch/hermes-agent/pull/17202)) shows how to build the same workflow yourself with a shell hook
Salvaged or co-authored work from **@isaachuangGMICLOUD** (GMI Cloud), earlier upstream PRs from the original author of each salvage chain, and a long tail of one-shot fixes, documentation nudges, and skill contributions from the community.
### All Contributors (alphabetical, excluding @teknium1)
f'[SYSTEM: The user has invoked the "{skill_name}" skill, indicating they want you to follow its instructions. The full skill content is loaded below.]',
f'[IMPORTANT: The user has invoked the "{skill_name}" skill, indicating they want you to follow its instructions. The full skill content is loaded below.]',
Some files were not shown because too many files have changed in this diff
Show More
Reference in New Issue
Block a user
Blocking a user prevents them from interacting with repositories, such as opening or commenting on pull requests or issues. Learn more about blocking a user.
