PR #30136 review item O6: test_container_restart.py used fixed
`time.sleep(8)` calls after `docker restart` to wait for the
cont-init reconciler to finish. Fixed sleeps are slow when the
event happens fast and false-fail when the event happens slow.
Replace with two polling helpers:
* `_wait_for_path(container, path, kind='f' | 'd', deadline_s=...)`
— generic `test -f/-d` poller. Returns True on success, False on
timeout; callers assert with a clear message.
* `_wait_for_reconcile_log_mention(container, profile, ...)` — the
reconciler's per-profile log line is the canonical signal that
the cont-init reconcile has finished for that profile. Poll on
it instead of a sleep that hopes 8 seconds is enough.
The fixture-level setup wait is similarly migrated: it now polls
for `profile=default` in the boot log (every container always
gets a default-slot entry per item I1) and raises a clear timeout
error from the fixture if the container never finishes cont-init —
much better diagnostics than a mid-test KeyError.
The remaining `time.sleep()` calls are all internal interval_s
between probe attempts; no fixed wait points left.
PR #30136 review noted the asymmetry: `register_profile_gateway`
used tmp_dir + rename to publish a new service slot atomically,
but the boot-time reconciler wrote files into the slot directly.
Same underlying concern (a concurrent s6-svscan rescan could
observe a half-populated directory), different code path.
Rewrite `container_boot._register_service` to mirror the manager:
build everything in `<scandir>/gateway-<profile>.tmp/`, then
`Path.replace` into place. If a previous interrupted run left a
`.tmp` sibling, it's cleaned up before the new build starts. If
the target already exists, it's removed before the rename so
`Path.replace` doesn't error on a non-empty target (Linux `rename`
overwrites empty targets only).
Three new tests: atomic publication leaves no .tmp leftovers,
overwriting an existing slot still leaves no .tmp leftovers, and
a stale .tmp from an interrupted run is cleaned up automatically.
PR #30136 review noted: container-boot.log was append-only with no
rotation. On a long-lived container with frequent restarts and
many profiles it would grow unboundedly (~80 B per profile per
reconcile pass).
Add a soft cap: when the file size hits 256 KiB (`_LOG_ROTATE_BYTES`,
≈3000 reconcile lines, ≈1 year of daily reboots × 5 profiles), the
current file is renamed to `container-boot.log.1` (replacing any
existing one) before new entries are appended. Worst case is two
files at ~512 KiB — well within visibility limits for grep/cat.
Rotation is intentionally simple (no logrotate or s6-log machinery
for one append-only file). Failures during rotation are logged via
the module logger and treated as non-fatal — we keep appending to
the existing file rather than dropping the reconcile entry. Three
new unit tests cover above-threshold rotation, below-threshold
non-rotation, and overwrite of an existing .1 file.
PR #30136 review caught: `_allocate_gateway_port()` in profiles.py
computed a SHA-256-derived port that was threaded through
`register_profile_gateway(profile, port=N)` →
`_render_run_script(profile, port, extra_env)` → and then **ignored**.
The rendered run script picked the bind port from the profile's
config.yaml (`[gateway] port = …`), never from the allocator. So
the entire allocator + parameter chain was dead code.
Remove:
* `hermes_cli.profiles._allocate_gateway_port` (deterministic
SHA-256 → [9200, 9800) — never used).
* `port` kwarg from `ServiceManager.register_profile_gateway`
(Protocol + Mixin + S6 implementation).
* `port` positional arg from `_render_run_script(profile, port,
extra_env)` — now `_render_run_script(profile, extra_env)`.
* The pass-through call in `profiles._maybe_register_gateway_service`.
config.yaml is now the single source of truth for gateway port
selection — matches reality and reduces the API surface. Three
explanatory comments in service_manager.py / profiles.py document
the retirement so future readers don't reach for the allocator and
find a ghost.
Tests: drop the three `_allocate_gateway_port` tests; update
fakes' signatures throughout test_service_manager.py and
test_profiles_s6_hooks.py to match the new no-port API.
PR #30136 review caught a false positive: when HERMES_DASHBOARD was
unset, the dashboard run script did `exec sleep infinity`, so
`s6-svstat /run/service/dashboard` reported the slot as 'up'.
`hermes doctor` and any other s6-svstat-based health check saw the
dashboard as supervised-running even though no dashboard process
existed.
Add cont-init.d/03-dashboard-toggle: writes a `down` marker file
into `/run/service/dashboard/` when HERMES_DASHBOARD is falsy,
removes any leftover marker when it's truthy. s6-supervise honors
`down` by not starting the service, so s6-svstat reports 'down' —
matching reality.
The run script's HERMES_DASHBOARD case-statement stays in place as
a belt-and-suspenders guard, so the two layers can never disagree.
Two new integration tests lock the behavior: slot reports down
when unset; slot reports up when set to 1.
PR #30136 review caught: `S6ServiceManager.start/stop/restart` called
`subprocess.run(check=True)` on `s6-svc`, so any failure surfaced as
a raw `CalledProcessError` traceback. The two cases operators
actually hit are:
1. The service slot doesn't exist — most commonly because the user
typed a profile name wrong (`hermes -p typo gateway start`).
2. s6-svc itself fails — most commonly EACCES on the supervise
control FIFO when running unprivileged.
Both deserve named errors with actionable messages, not stacktraces.
Changes:
* Add `S6Error` base + two concrete errors in `hermes_cli.service_manager`:
- `GatewayNotRegisteredError(profile)` — carries the unprefixed
profile name; message: `no such gateway 'typo': register it
with `hermes profile create typo` first, or pass an existing
profile name via `-p <name>``.
- `S6CommandError(service, action, returncode, stderr)` — carries
the s6-svc rc and stderr; message: `s6-svc start on
'gateway-coder' failed (rc=111): <stderr>`.
* Factor lifecycle dispatch through `_run_svc(flag, label, name)`:
pre-checks that the service directory exists (raises
GatewayNotRegisteredError before invoking s6-svc), then runs
s6-svc and translates any CalledProcessError into S6CommandError.
* `_dispatch_via_service_manager_if_s6` in `hermes_cli.gateway`
catches both errors and prints `✗ <message>` + `sys.exit(1)`
instead of letting the exception bubble. The dispatch path that
used to dump a traceback at the user now gives an actionable
one-liner.
Tests: 6 new tests for the error types and their CLI rendering;
existing lifecycle test pre-seeds the slot directory before calling
`mgr.start` etc.
PR #30136 review caught: `hermes gateway start` (no `-p`) inside
the container resolves `_profile_suffix() == ""` → service name
`gateway-default`, but no such slot was ever registered. The Phase 4
profile-create hook only fired on `hermes profile create <name>`,
and the root profile (which lives at the top of $HERMES_HOME, not
under `profiles/`) was never one of those. So bare `hermes gateway
start` landed on `s6-svc -u /run/service/gateway-default` →
uncaught `CalledProcessError` → traceback to the user.
Changes:
1. `reconcile_profile_gateways` now always registers a
`gateway-default` slot before iterating named profiles. Its
prior state is read from `$HERMES_HOME/gateway_state.json`
(sibling to the profile root, not under `profiles/`); stale
runtime files there are swept the same way. Auto-up only if the
prior state was `running` — same rule as named profiles.
2. `S6ServiceManager._render_run_script` special-cases
`profile == "default"` to emit `hermes gateway run` with NO
`-p` flag. Passing `-p default` would resolve to
`$HERMES_HOME/profiles/default/` — a different profile that
almost certainly doesn't exist. The empty profile-suffix
convention is the dispatcher's contract and the run script has
to match.
3. A user-created `profiles/default/` collides with the reserved
root-profile slot; the reconciler now skips it with a warning
rather than producing two registrations of the same service name.
Action-list ordering is stable: `default` first, then named
profiles in directory order. Boot-log readers can rely on this.
Tests: 8 new dedicated default-slot tests plus updates to every
existing test that asserted against the action list (via the new
`_named_actions` helper that drops the always-present default
entry).
PR #30136 review caught that `hermes gateway stop --all` and
`... restart --all` were broken under s6. The Phase 4 dispatcher was
gated on `not stop_all` (and the symmetric restart_all), so `--all`
fell through to `kill_gateway_processes(all_profiles=True)`. pkill
SIGTERMed every gateway, s6-supervise observed the crashes, and
restarted every gateway ~1s later — net effect: `--all` *kicked*
gateways instead of *stopping* them.
Add `_dispatch_all_via_service_manager_if_s6(action)` that iterates
`mgr.list_profile_gateways()` and routes stop/restart through each
service slot. s6's `want up`/`want down` flips correctly, so a
stop persists. Partial failures are surfaced per-profile with a
running success count; the host pkill path is only reached when s6
isn't in play.
`start --all` isn't a CLI surface — the helper rejects it and
returns False (host code path can take over).
PR #30136 review surfaced two issues, both rooted in the same audit gap:
docker integration tests were running as root, not the unprivileged
`hermes` user (UID 10000) that the runtime actually uses via
`s6-setuidgid hermes`. Anything that probed PID-1 state or wrote to
the s6 control surface worked as root in the tests but was inert in
production.
Fixes:
1. `_s6_running()` previously called `Path("/proc/1/exe").resolve()`,
which is root-only readable. For UID 10000 the symlink yields
PermissionError, `resolve()` silently returns the unresolved path,
and `exe.name == "exe"` — so detection always returned False, the
service-manager runtime-registration path was inert, and every
`hermes profile create` / `hermes -p X gateway start` silently
skipped the s6 hook. Replace with `/proc/1/comm` (world-readable)
+ `/run/s6/basedir` (s6-overlay-specific) — both required, fail
closed.
2. `02-reconcile-profiles` now also chowns `/run/service/.s6-svscan/`
{control,lock} to hermes so `s6-svscanctl -a/-an` works without
root. Previously the directory chown stopped at `/run/service`
and the FIFO inside stayed root-owned, so `register_profile_gateway`
from hermes failed at the rescan-trigger step with EACCES — the
wrapper in profiles.py caught the exception and printed a swallowed
warning, so profile creation appeared to succeed while the slot
was rolled back.
Audit changes to flush this class of bug next time:
- Add `docker_exec` / `docker_exec_sh` helpers to `tests/docker/conftest.py`
that default to `-u hermes`. The module docstring explains why and
flags `user="root"` as opt-in only for tests that explicitly need
root (none currently do).
- Refactor every `docker exec` call in tests/docker/ through the new
helpers (test_dashboard.py, test_zombie_reaping.py, test_profile_gateway.py,
test_container_restart.py, test_s6_profile_gateway_integration.py).
- Add 5 unit tests covering `_s6_running` under various probe states
(both signals present; comm wrong; basedir missing; PermissionError
on /proc/1/comm; missing /proc — non-Linux). The PermissionError
test is the explicit regression guard for the original bug.
Known follow-up: the per-service `supervise/control` FIFO inside each
`/run/service/gateway-<profile>/supervise/` is created root-owned by
s6-supervise (which runs as root because s6-svscan is PID 1). `s6-svc
-u/-d/-t` from the hermes user will get EACCES on those. The audit
under `-u hermes` will reveal this in lifecycle tests — surfacing the
issue cleanly so it can be fixed in a focused follow-up (likely via a
small SUID helper or a polling chown loop in cont-init.d). The
detection + svscanctl fixes here are independent and complete on
their own.
The s6-overlay migration replaced every runtime use of gosu with
s6-setuidgid (in stage2-hook.sh, main-wrapper.sh, per-service run
scripts, and cont-init.d hooks), but the gosu binary itself was still
being copied into the image from tianon/gosu, and several comments
across the repo still pointed to it.
Image changes:
- Drop the FROM tianon/gosu:1.19-trixie AS gosu_source stage
- Drop the COPY --from=gosu_source /gosu /usr/local/bin/ layer
- Net: one fewer base-image pull, ~12-15 MB layer eliminated
Documentation/comment refresh (no behavior change):
- Dockerfile: update root-user rationale comment + cont-init.d comment
- docker/main-wrapper.sh: drop "pre-s6 contract (gosu drop)" reference
- docker-compose.yml: update UID/GID remap comment
- .hadolint.yaml: update DL3002 ignore rationale
- website/docs/user-guide/docker.md: privilege-drop helper is s6-setuidgid now
- hermes_cli/config.py: docker_run_as_host_user docstring
tools/environments/docker.py runs *arbitrary user images* via the
terminal backend, not the bundled Hermes image. It still needs SETUID/
SETGID caps so user images that use gosu/su/s6-setuidgid all work.
Renamed the cap-list constant _GOSU_CAP_ARGS → _PRIVDROP_CAP_ARGS and
updated comments to list s6-setuidgid alongside the others as examples.
The matching test (test_security_args_include_setuid_setgid_for_gosu_drop
→ test_security_args_include_setuid_setgid_for_privdrop) was renamed
and its docstring updated; behavior is unchanged.
Verification:
- hadolint clean against .hadolint.yaml
- shellcheck clean against all docker/ shell scripts
- Image rebuilt successfully (sha 1a090924ccea)
- Docker harness: 19 passed in 41.87s (every Phase 0 test + Phase 4
per-profile-gateway lifecycle + container-restart reconciliation)
- tests/tools/test_docker_environment.py: 23 passed (rename did not
break test discovery; pre-existing unrelated mock warning)
The plan document (docs/plans/2026-05-07-s6-overlay-dynamic-subagent-gateways.md)
intentionally retains its historical references to gosu — it describes
the pre-s6 entrypoint as background for understanding the migration.
Phase 4 of the s6-overlay supervision plan. Activates the Phase 3
S6ServiceManager by hooking it into the profile lifecycle and the
`hermes gateway start/stop/restart` dispatcher, and adds a cont-
init.d-time reconciliation pass that survives `docker restart`.
Task 4.0 — container-boot reconciliation:
/run/service/ is tmpfs, so every `docker restart` wipes every
per-profile gateway slot. /etc/cont-init.d/02-reconcile-profiles
invokes hermes_cli.container_boot.reconcile_profile_gateways() on
every boot, which walks $HERMES_HOME/profiles/<name>/, reads each
gateway_state.json, recreates the s6 service slot, and auto-starts
only those whose last state was 'running'. Other states
(stopped, starting, startup_failed, missing) register the slot
in the down state — avoiding crash-loops across restarts for a
gateway that was broken last boot. Per-profile outcome is recorded
to $HERMES_HOME/logs/container-boot.log.
Implementation: hermes_cli/container_boot.py + 12 unit tests.
Profile-marker is SOUL.md, not config.yaml, because `hermes profile
create` only seeds SOUL.md by default (config.yaml comes from
`hermes setup`).
Task 4.1 / 4.2 — profile create/delete hooks:
hermes_cli/profiles.py::create_profile now calls
_maybe_register_gateway_service(<canon>) at the end, which routes
through ServiceManager.register_profile_gateway when running on s6
and no-ops on host backends. delete_profile mirrors with
_maybe_unregister_gateway_service. _allocate_gateway_port produces
a deterministic SHA-256-derived port in [9200, 9800).
Task 4.3 — gateway dispatch + remove rejection arms:
_dispatch_via_service_manager_if_s6(action) intercepts
start/stop/restart at the top of each subcommand and routes them
through S6ServiceManager.{start,stop,restart}. The pre-Phase-4
`elif is_container():` rejection arms are kept as fallback for
pre-s6 containers / unsupported runtimes, but only ever fire when
detect_service_manager() != 's6'. install/uninstall under s6
print informational guidance pointing users at profile create/delete.
Removed the two xfail(strict=True) markers from
tests/docker/test_profile_gateway.py — both tests now pass strictly.
Task 4.4 — status reporting:
get_gateway_runtime_snapshot() reports
Manager: 's6 (container supervisor)' inside an s6 container instead
of 'docker (foreground)'.
Plan-vs-reality drift fixed in this commit:
- Plan's S6ServiceManager._render_run_script used
`gateway start --foreground --port {port}` — invented args; the
real CLI is `gateway run`. Switched accordingly. port arg
retained for API parity but now documented as 'currently ignored'.
- Plan's reconciler keyed on config.yaml; switched to SOUL.md
(config.yaml is created by hermes setup, not by hermes profile
create, so the original gate caught nothing).
- The plan's _dispatch helper used _profile_arg() which returns
'--profile <name>' (i.e. with the flag prefix). Switched to
_profile_suffix() which returns the bare name.
- Architecture B's docker exec doesn't get /command on PATH or
the venv on PATH; Dockerfile's runtime PATH now includes
/opt/hermes/.venv/bin so 'docker exec <c> hermes ...' works
without sourcing the venv.
- stage2-hook now chowns $HERMES_HOME/profiles to hermes on every
boot, not just on the UID-remap path. Without this, files created
by docker-exec-as-root accumulate and the next reconciler run
fails with PermissionError reading SOUL.md.
Test harness:
19 passed, 0 xfailed (the two pre-Phase-4 xfail targets flip to
passing). 78 unit tests across service_manager + container_boot +
profiles_s6_hooks + gateway_s6_dispatch. Hadolint + shellcheck
pass cleanly.
Refs: docs/plans/2026-05-07-s6-overlay-dynamic-subagent-gateways.md
Phase 3 of the s6-overlay supervision plan. Implements the runtime-
registration surface from D4 — only the s6 backend supports
register_profile_gateway / unregister_profile_gateway /
list_profile_gateways; host backends continue to raise
NotImplementedError. No caller yet (Phase 4 wires in the profile
create/delete hooks).
Key implementation notes:
- Service directory shape: /run/service/gateway-<profile>/{type,run,log/run}.
Atomic register: write to gateway-<profile>.tmp, fsync via
os.rename. Cleanup on rescan failure.
- Run script uses #!/command/with-contenv sh so HERMES_HOME and any
extra_env arrive at exec time. The hermes -p <profile> gateway
start --foreground --port <port> command is wrapped in
s6-setuidgid hermes for the per-service privilege drop (OQ2-A).
- Log script (OQ8-C): persists via s6-log to
${HERMES_HOME}/logs/gateways/<profile>/. CRITICAL — HERMES_HOME is
a runtime env-var expansion in the rendered script, NOT a Python
f-string substitution. Negative-asserted in
test_s6_register_creates_service_dir_and_triggers_scan so
regressions are caught.
- PATH gotcha: /command/ is only on PATH for processes spawned by
the supervision tree (services, cont-init.d). `docker exec` and
profile-create hooks don't get it. S6ServiceManager calls all
s6-* binaries via absolute path through the new _S6_BIN_DIR
constant so callers don't have to fix up env vars.
- validate_profile_name rejects path-traversal, leading-dash (s6
would parse as a flag), uppercase, whitespace, and names >251
chars (s6-svscan default name_max).
Test coverage:
- 13 new unit tests in tests/hermes_cli/test_service_manager.py
(kind detection, run-script content, env quoting, register
rollback on rescan failure, unregister idempotence, list filter,
lifecycle dispatch, svstat parsing). Total: 36 passing.
- 2 new in-container integration tests in
tests/docker/test_s6_profile_gateway_integration.py validating
end-to-end registration against a real s6 supervision tree.
Docker harness: 14 passed, 2 xfailed (Phase 4 target unchanged).
Refs: docs/plans/2026-05-07-s6-overlay-dynamic-subagent-gateways.md
BREAKING CHANGE: the container ENTRYPOINT is now /init (s6-overlay)
instead of /usr/bin/tini. Main hermes runs as the container CMD with
TTY inherited (preserving --tui), dashboard runs as a supervised s6-rc
service (HERMES_DASHBOARD=1 starts it; crashes auto-restart), and the
ground is laid for per-profile gateway supervision (Phase 3+4).
All five pre-s6 docker run invocation patterns continue to work
identically — verified by the Phase 0 docker harness:
docker run <image> → `hermes` with no args
docker run <image> chat -q "..." → `hermes chat -q ...` passthrough
docker run <image> sleep infinity → `sleep infinity` direct
docker run <image> bash → interactive bash
docker run -it <image> --tui → interactive Ink TUI
Phase 2 harness result: 12 passed, 2 xfailed (Phase 4 target). Hadolint
+ shellcheck pass cleanly.
Architecture pivot from plan v3 (documented in main-hermes/run header):
the plan called for main hermes to be an s6-supervised service, but
two real s6-overlay v3 mechanics blocked that — cont-init.d scripts
receive no arguments (CMD args are not visible to stage2-hook), and
`/run/s6/basedir/bin/halt` after writing the exit code did not
propagate the desired exit code (container exits 143). We use the
s6-overlay-native CMD pattern instead: main-wrapper.sh is the
container's main program (ENTRYPOINT prepends it so leading-dash
args like --version aren't intercepted by /init), exec's the final
program with stdin/stdout/stderr inherited, and the program's exit
code becomes the container exit code. main-hermes is now a no-op
`sleep infinity` slot kept for future supervised-gateway-container
modes. This trades "supervised restart of main hermes" for arg-
parity with the pre-s6 contract — main hermes was already unsupervised
under tini, so we lose nothing functional. Dashboard supervision is
the only new guarantee added by this phase.
Files added:
docker/main-wrapper.sh # arg routing + s6-setuidgid drop
docker/stage2-hook.sh # gosu-equivalent + chown + seed
docker/s6-rc.d/main-hermes/{type,run,dependencies.d/base}
docker/s6-rc.d/dashboard/{type,run,dependencies.d/base}
docker/s6-rc.d/user/contents.d/{main-hermes,dashboard}
Files changed:
Dockerfile: tini → s6-overlay install + ENTRYPOINT flip + service wiring
docker/entrypoint.sh: thin shim to stage2-hook.sh for back-compat
tests/docker/test_dashboard.py: add test_dashboard_restarts_after_crash
Refs: docs/plans/2026-05-07-s6-overlay-dynamic-subagent-gateways.md
Phase 1 of the s6-overlay supervision plan. Pure-refactor addition:
introduces the abstract interface (with runtime_checkable Protocol),
detect_service_manager(), validate_profile_name(), and thin
SystemdServiceManager / LaunchdServiceManager / WindowsServiceManager
wrappers around the existing systemd_* / launchd_* / gateway_windows.*
module-level functions. No host call site was modified — host code
continues to use the existing functions directly; the protocol is for
new backend-agnostic code (Phase 4 profile create/delete hooks and the
Phase 4 s6 dispatch path in 'hermes gateway start/stop/restart').
WindowsServiceManager.install() forwards the v3 kwargs (start_now,
start_on_login, elevated_handoff) added in PRs #28169-adjacent so
non-Windows callers — there aren't any today — can opt in.
The s6 backend lands in Phase 3; until then get_service_manager()
raises a clear error if invoked on a host that detects as 's6'.
Two pre-existing baseline issues found while running the Phase 0 harness
against the tini image that need fixing before later phases can use the
harness as a behavior-parity oracle:
1. The autouse `_enforce_test_timeout` fixture in tests/conftest.py
hard-coded a 30s SIGALRM, which preempted any `pytest.mark.timeout`
marker (already honored by pytest-timeout). Honor the marker if
present; fall back to 30s otherwise. Docker harness tests carry a
180s marker applied at collection time in tests/docker/conftest.py.
2. test_dashboard_port_override polled via `ss -tlnp` / `netstat -tln`
— neither is installed in the Hermes image, so the probe trivially
failed even when the dashboard was bound. The dashboard also takes
8-15s to bind on cold image; the 5s sleep was insufficient. Replace
with a poll loop reading /proc/net/tcp directly (port 9120 = 0x23A0,
state 0A = LISTEN). Bump probe deadline to 60s and switch
test_dashboard_opt_in_starts to a similar poll for pgrep so we don't
regress to the same race.
Result: 11 passed, 2 xfailed (Phase 4 target) on tini image. Harness
now ready to serve as Phase 2's behavior-parity oracle.
The agent-test suite default is 30s; docker test_no_args (the dashboard
spin-up, the container restart) routinely take 60-90s. Without this
they intermittently fail in CI with TimeoutError.
Tasks 0.2-0.6 of the s6-overlay supervision plan. Locks the
user-visible behavior we must preserve through the Phase 2 init-
system swap:
- test_main_invocation.py (Task 0.2): docker run <image> with no
args, chat subcommand passthrough, bare executable passthrough,
bash pattern, exit-code propagation
- test_tui_passthrough.py (Task 0.3): TTY allocation via docker -t
using the host's script(1) for a PTY
- test_dashboard.py (Task 0.4): HERMES_DASHBOARD=1 opt-in,
HERMES_DASHBOARD_PORT override
- test_profile_gateway.py (Task 0.5): per-profile gateway
start/stop and profile-delete-stops-gateway. Both marked
xfail(strict=True) because the current tini image refuses
gateway lifecycle commands inside the container; Phase 4
Task 4.3 flips them to passing.
- test_zombie_reaping.py (Task 0.6): PID 1 reaps orphaned
zombies. tini does this today; s6-overlay's /init must
continue to.
Refs: docs/plans/2026-05-07-s6-overlay-dynamic-subagent-gateways.md
Task 0.1 of the s6-overlay supervision plan. Establishes the test
infrastructure for tests/docker/: skip-on-missing-Docker collection
hook, session-scoped image-build fixture (overridable via the
HERMES_TEST_IMAGE env var for faster local iteration), and a
container_name fixture that ensures cleanup on test exit.
Refs: docs/plans/2026-05-07-s6-overlay-dynamic-subagent-gateways.md
* feat(tui): make display.mouse_tracking pick which DEC modes to enable
Previously the boolean flag was all-or-nothing across modes 1000+1002+1003+1006.
Inside tmux, mode 1003 (any-motion) makes every mouse cross of the prompt row
fire a clipboard probe that surfaces as "No image in clipboard" — sometimes
dozens in a row. Disabling tracking entirely killed scroll-wheel scrolling too,
since tmux's own scrollback is preempted by the alt-screen TUI.
`display.mouse_tracking` (and `/mouse <preset>`) now accepts `off | wheel |
buttons | all` in addition to the legacy booleans. `wheel` is 1000+1006:
scroll wheel + click only, no drag, no hover — the tmux-friendly subset.
`buttons` adds 1002 for drag-to-select. `all` (= legacy `true`) keeps the
hover-driven UI (scrollbar paginate-on-hover, link mouseenter, etc.).
* fix(tui): repaint + sync mouse mode when display.mouse_tracking changes
Two interacting bugs left the TUI blank when `display.mouse_tracking`
switched at runtime (config edit, /mouse <preset>):
1. AlternateScreen's effect re-runs on every `mouseTracking` change,
tearing down and re-entering the alt screen. After re-entry, ink's
frame buffers are reset by `resetFramesForAltScreen()` but nothing
schedules the follow-up render — the alt screen sits blank until
some other state change happens to trigger one. Add a
`scheduleRender()` in `setAltScreenActive`'s active=true branch so
the freshly-entered alt screen gets a full repaint immediately.
2. `setAltScreenActive` early-returns when `active` hasn't changed,
which silently drops a `mouseTracking` change if the cleanup→setup
pair somehow leaves `altScreenActive` already true. Call
`setAltScreenMouseTracking` explicitly from the AlternateScreen
effect so the in-memory mode and terminal DECSET sequence stay in
sync regardless of how `setAltScreenActive` resolved (the call is a
no-op when the mode is unchanged).
* fix(tui): address copilot review #4341269705
- tui_gateway/server.py: drop the never-referenced _MOUSE_TRACKING_MODES
frozenset (comment #3284802434). _MOUSE_TRACKING_ALIASES already
centralizes the canonical preset set via its values; the separate
constant added no behavior.
- tests/test_tui_gateway_server.py: update the existing
test_config_mouse_uses_documented_key_with_legacy_fallback to assert
the new preset strings ('all'/'off' instead of 'on'/'off',
display.mouse_tracking persisted as 'all' instead of True) and add
test_config_mouse_accepts_preset_strings_and_aliases covering /mouse
set with wheel/click/unknown (comment #3284802453). The on/off legacy
config.set return shape was an implementation detail of the boolean
flag, not a stable API — the slash command, gateway help text, and
docs all advertise the preset values now.
- ui-tui/packages/hermes-ink/src/ink/ink.tsx: schedule a render at the
end of reenterAltScreen() (comment #3284802461). Mirrors the same fix
in setAltScreenActive() from ece0a2f4c — without it, SIGCONT/resize
self-heal/stdin-gap re-entry leaves the alt screen blank because
every caller returns early after invoking us.
* fix(tui): address copilot review #4341308478 round 2
- ui-tui/src/config/env.ts (comment #3284837577): the precedence
comment was misleading. Actual behavior on origin/main is
HERMES_TUI_MOUSE_TRACKING (explicit override) > Termux default >
HERMES_TUI_DISABLE_MOUSE legacy kill-switch. This is preserved from
main; the only change here was the wrong comment that claimed
DISABLE_MOUSE kept kill-switch semantics. Rewrote the comment block
to document the actual precedence ladder.
- tui_gateway/server.py /mouse set (comment #3284837607): replaced
'str(value or "").strip().lower()' with the explicit None idiom
already used for /indicator, so programmatic callers can pass 0 /
False and have them route through _MOUSE_TRACKING_ALIASES → 'off'
instead of collapsing to '' and triggering the toggle path.
- ui-tui/packages/hermes-ink/src/ink/components/AlternateScreen.tsx
(comment #3284837620): always prepend DISABLE_MOUSE_TRACKING before
enableMouseTrackingFor(...) on mount. Otherwise selecting
'wheel'/'buttons' from a state where DEC 1003 was already asserted
(crash, another app, debugger) would silently leave hover on. Also
unconditionally DISABLE on unmount so a crash mid-mount can't leak
DEC modes back to the host shell.
* chore(release): map nat@nthrow.io to @nthrow for #26681 salvage
* fix(tui): drop redundant setAltScreenMouseTracking in AlternateScreen
Copilot review #4341356637 (comment #3284880417). The explicit
setAltScreenMouseTracking(mouseTracking) after setAltScreenActive(true,
mouseTracking) was defensive paranoia added in the previous fix commit
that's not actually reachable in practice:
- React's cleanup always runs before the next setup, so on any prop
change (mouseTracking or writeRaw) the cleanup sets active=false
first. Setup then sees active was false and applies the new mode
via setAltScreenActive without early-returning.
- On the impossible 'active stayed true' path, the writeRaw above has
already sent DISABLE_MOUSE_TRACKING + enableMouseTrackingFor(newMode)
to the terminal, so the in-memory mode would lag but the visible
state is already correct.
Removing the redundant call means a single DEC sequence per mount.
If the 'active stayed true' path ever manifests in practice, the
right fix is in setAltScreenActive (track mode regardless of the
active early-return), not here.
* fix(tui): always DISABLE before enableMouseTrackingFor in ink.tsx
Copilot review #4341379994 (comments #3284900825, #3284900840,
#3284900852). Three remaining call sites in ink.tsx still re-enabled
mouse tracking without first sending DISABLE_MOUSE_TRACKING:
- handleResize alt-screen recovery (line ~577)
- reassertTerminalModes stdin-gap re-assertion (line ~1351)
- reenterAltScreen SIGCONT/resize/stdin-gap self-heal (line ~1408)
For 'wheel'/'buttons' presets, omitting DISABLE leaves any externally-
asserted DEC 1003 (other apps, prior crash, tmux state) still active
and the hover-free preset silently has hover on. DISABLE_MOUSE_TRACKING
is idempotent and safe to send unconditionally — it resets all four
modes. Matches the pattern already in setAltScreenMouseTracking and
the AlternateScreen mount path.
* fix(tui): always DISABLE before enableMouseTrackingFor in exitAlternateScreen
Copilot review #4341452823 (comment #3284959762). exitAlternateScreen()
was the last call site in ink.tsx still re-enabling mouse tracking
without DISABLE first. Editors (vim/nvim/less) and tmux can leave
DEC 1003 hover asserted across the handoff back; without DISABLE,
'wheel'/'buttons' presets silently kept hover on after the editor
quit. Now all five enableMouseTrackingFor() call sites in ink.tsx
prepend DISABLE_MOUSE_TRACKING — handleResize, reassertTerminalModes,
reenterAltScreen, setAltScreenMouseTracking, exitAlternateScreen.
* fix(tui): add defensive default to enableMouseTrackingFor switch
Copilot review #4341485231 (comment #3284979323). TS exhaustive switch
returns string per the type system, but a JS caller / corrupted config
/ hot-reload-in-dev could reach the function with an unknown value at
runtime. Without a default, that path returns undefined which then
concatenates as the literal string 'undefined' into the terminal byte
stream — visibly garbling output. Treat unknown as 'off' (no DEC
sequences) so the worst case is silent input loss rather than a
wrecked screen.
---------
Co-authored-by: Nat Thrower <nat@nthrow.io>
Add tests/tools/test_computer_use_capture_routing.py — 13 integration
tests that drive _capture_response end-to-end with deterministic stubs
for the routing helper, _run_async, vision_analyze_tool, and
get_hermes_dir, so the full code path is exercised without a live
cua-driver, real auxiliary client, or network access.
Coverage:
* TestCaptureResponseDefaultPath (3 cases)
- SOM PNG capture returns the legacy multimodal envelope when the
routing helper says 'native' (image/png MIME).
- Same path returns image/jpeg MIME for JPEG payloads (cua-driver
can return either).
- AX-only mode never even consults the routing helper because no
PNG is present.
* TestCaptureResponseRoutedToAuxVision (5 cases)
- SOM capture with routing on returns a JSON string with the
vision_analysis embedded, the AX/SOM index preserved, and NO
image_url parts. Verifies the aux call receives a path under
the configured cache and a prompt that grounds itself against
the AX summary.
- Temp screenshot file is unlinked after _capture_response returns,
including when the aux call raises (the finally block runs).
- Empty / malformed aux analysis falls back to the multimodal
envelope so the user always gets *something* useful.
* TestRoutingDecisionWiring (4 cases)
- Explicit auxiliary.vision in config flips routing on regardless of
main-model vision capability.
- Vision-capable main + native tool-result support keeps multimodal.
- Config load failure fails open (returns False, multimodal path
continues to work).
- Helper exception is swallowed and routes to legacy behaviour.
* TestBugReproductionAnchor (1 case) - directly pins the #24015
contract: when routing is on, the response must NEVER contain a
'data:image' or 'image_url' substring. That is exactly what tripped
the reporter's HTTP 404 ('No endpoints found that support image
input') on tencent/hy3-preview before the fix.
Bug-reproduction proof:
$ git checkout upstream/main -- tools/computer_use/tool.py
$ scripts/run_tests.sh tests/tools/test_computer_use_capture_routing.py
============================== 13 failed in 1.29s ==============================
$ # restore tool.py to this branch's HEAD
$ scripts/run_tests.sh tests/tools/test_computer_use_capture_routing.py
============================== 13 passed in 1.04s ==============================
Total branch coverage:
85 passed across test_computer_use.py, test_computer_use_vision_routing.py,
test_computer_use_capture_routing.py
Add tests/tools/test_computer_use_vision_routing.py — 28 unit tests
that pin the contract of the new vision-routing helper introduced in
the previous commit:
* TestExplicitAuxVisionOverride (12 cases): mirror the
auxiliary.vision detection rules used by agent.image_routing so
the capture path and the user-attached-image path agree on what
counts as an explicit override (provider/model/base_url with
non-blank, non-'auto' values).
* TestRouteDecision (7 cases): pin the policy itself — explicit
override always wins, vision-capable + native-tool-result keeps
multimodal, everything else fails closed and routes to aux.
* TestLookupHelpers (5 cases): defensive paths for the models.dev /
tool-result-support lookups (blank inputs, exceptions, missing
caps).
* TestModuleSurface (4 cases): pin the public/__all__ surface and
keep internal helpers addressable so the integration test in the
next commit can monkeypatch them deterministically.
Run with:
scripts/run_tests.sh tests/tools/test_computer_use_vision_routing.py
The bundled-skill sync stamp added in the cherry-picked salvage commit
parsed .git/HEAD and looked for a loose ref file in the worktree gitdir
only, so two real cases hit the unresolved branch:
- repos after `git gc` where active refs live in packed-refs
- linked worktrees, whose branch ref lives in <commondir>/refs/heads/
(verified on the worktree this salvage was built in)
Both fell back to a constant-string fingerprint, so post-commit launches
would never re-run the real skill sync. Now we resolve packed-refs and
check both the worktree gitdir and the common dir for loose refs.
Adds three tests covering: packed-refs resolution, worktree common-dir
packed lookup, worktree common-dir loose lookup, and the explicit
'unresolved' marker (still stable + version-fallback-safe).
`CuaDriverBackend.capture(app=X)` and `focus_app(app=X)` silently fell back
to the frontmost on-screen window when X matched no app — typically a
menu-bar utility (e.g. "Fuwari" in the bug reporter's case) rather than
the requested app. The agent then received UI elements for the wrong app
and clicked / typed into it.
The root cause is a localized macOS app name mismatch: `list_windows`
returns the localized `app_name` (e.g. "計算機" on a Japanese/Chinese
system) but callers naturally pass the English name ("Calculator"). The
substring filter doesn't match, and the code falls through to picking the
frontmost window with no signal that the filter was effectively dropped.
Fix:
- `capture(app=…)`: when the filter matches nothing, return a
`CaptureResult` with empty `app`/`elements` and a diagnostic
`window_title` pointing the caller at `list_apps` and noting the
localized-name convention. `_active_pid` / `_active_window_id` are left
untouched so a subsequent action doesn't inadvertently hit the wrong
process.
- `focus_app(app=…)`: when the filter matches nothing, set `target = None`
and let the existing `return ActionResult(ok=False, …, "No on-screen
window found for app …")` path fire instead of falsely reporting success
on the frontmost window.
This addresses bug 1 only from #24170. Bugs 2 & 5 are addressed in #30046;
bugs 3 & 4 in #30032.
Bug 2 (capture_after=True loses app context):
_maybe_follow_capture called backend.capture(mode='som') with no app=,
causing cua-driver to capture the frontmost window instead of the app
targeted by the preceding capture/focus_app. Fix: track _last_app on
CuaDriverBackend and thread it through the follow-up capture call so
the same app is re-captured regardless of which window has OS focus.
Bug 5 (element labels stripped in capture results):
_ELEMENT_LINE_RE matched the classic ' - [N] AXRole "label"' format
but not the '[N] AXRole (order) id=Label' format introduced in
cua-driver v0.1.6. All element labels were silently dropped as empty
strings, making element identification impossible.
Fix: extend regex to capture both group(3) (quoted label) and group(4)
(id= label), and update _parse_elements_from_tree to use group(4) as
fallback. Both old and new cua-driver output now produce populated
UIElement.label values.
focus_app() now also sets _last_app so that capture_after= on any
subsequent action re-targets the focused app.
5 new regression tests added.
Part of #24170 (bugs 1 and 3/4 addressed separately).
* fix(skills): skip dependency dirs in skill scan
* fix(skills): widen sibling rglob scanners to use shared exclusion set
Follow-up to PR #29968. The contributor's PR widened EXCLUDED_SKILL_DIRS
in the canonical walker (iter_skill_index_files), which fixes the
user-visible discovery path. This commit sweeps the ~12 other
rglob('SKILL.md') sites that did their own ad-hoc filtering — most only
checked .git/.hub, some had no filter at all — so dependency dirs
(.venv, node_modules, site-packages, etc.) cannot leak ghost skills
through the secondary paths.
Adds agent.skill_utils.is_excluded_skill_path(path) helper. Migrates
all 13 sites to use it. Removes 3 hardcoded duplicate filter sets.
Sites touched:
agent/curator_backup.py - skill backup file count
gateway/run.py - disabled-skill response (2 sites)
hermes_cli/dump.py - skill count in env dump
hermes_cli/profile_describer.py- profile description (2 sites)
hermes_cli/profile_distribution.py - profile install count
hermes_cli/profiles.py - profile skill count
hermes_cli/skills_hub.py - category detection
tools/skill_manager_tool.py - skill name lookup (already used set, now uses helper)
tools/skill_usage.py - usage tracking + skill dir lookup (2 sites)
tools/skills_hub.py - optional skills find + scan (2 sites)
tools/skills_sync.py - bundled skills sync
E2E verified with the exact reported shape
(bring/scripts/.venv/.../typer/.agents/skills/typer/SKILL.md): no
sibling site picks up the ghost skill, all five legit-skill counts
still return 1.
* chore(infographic): retro-pop-grid bento for PR #30042 skill-scanner sweep
---------
Co-authored-by: helix4u <4317663+helix4u@users.noreply.github.com>
* feat(secrets): Bitwarden Secrets Manager integration with lazy bws install
Pull API keys from Bitwarden Secrets Manager at process startup
instead of storing them all in plaintext in ~/.hermes/.env. One
bootstrap token (BWS_ACCESS_TOKEN) replaces N per-provider keys, and
rotating a credential becomes a single change in the Bitwarden web
app.
Bitwarden defaults to source of truth: secrets pulled from BSM
overwrite any matching env vars on startup so rotations actually
take effect. Set secrets.bitwarden.override_existing: false in
config.yaml to invert.
The bws binary is auto-downloaded into ~/.hermes/bin/bws on first
use (pinned to v2.0.0, SHA-256 verified against the GitHub release
checksum file). No apt, brew, or sudo required.
New surfaces:
hermes secrets bitwarden setup — interactive wizard
hermes secrets bitwarden status — config + binary + token state
hermes secrets bitwarden sync — dry-run fetch / --apply exports
hermes secrets bitwarden disable — flip enabled: false
hermes secrets bitwarden install — just download the binary
Failures (missing binary, bad token, no network) never block Hermes
startup — they emit a one-line warning to stderr and continue with
whatever credentials .env already had.
Docs: website/docs/user-guide/secrets/{index,bitwarden}.md
Tests: tests/test_bitwarden_secrets.py (26 tests, hermetic — bws
subprocess and HTTP downloads fully mocked)
* chore(infographic): add bitwarden-secrets-manager bento-grid retro-pop-grid
Generated for PR #30035 — Bitwarden Secrets Manager integration.
Style picked via pick_pr_infographic_style.py rotation:
layout: bento-grid
style: retro-pop-grid
aspect: 1:1 square
Saved at infographic/bitwarden-secrets-manager/infographic.png
Bug 3: The cua_backend type_text() method called MCP tool 'type_text_chars'
which does not exist in current cua-driver. Changed to 'type_text' which is
the correct MCP tool name.
Bug 4: The drag() method returned a hardcoded 'not supported' error even
though cua-driver exposes a 'drag' MCP tool. Implemented proper drag
dispatching with coordinate-based and element-based targeting.
Added dispatch-level validation for drag to ensure from/to coordinates
or elements are provided before calling any backend.
Fixes#24170 (bugs 3 and 4)
Allow custom OpenAI-compatible providers declared under `custom_providers:`
to set provider-specific `extra_body` fields and have Hermes merge them into
chat-completions requests when the matching custom endpoint is active.
This is a manual per-provider override rather than a model-name heuristic.
OpenAI-compatible Gemma thinking support is real, but the on-wire payload
shape is backend-specific: some servers want top-level `enable_thinking`,
while vLLM Gemma and NIM-style endpoints expect `chat_template_kwargs`.
A per-provider override is safer than picking one assumed payload.
Example config:
```yaml
custom_providers:
- name: gemma-local
base_url: http://localhost:8080/v1
model: google/gemma-4-31b-it
extra_body:
enable_thinking: true
reasoning_effort: high
```
For vLLM Gemma or NIM-style endpoints, use the nested shape those servers
expect:
```yaml
extra_body:
chat_template_kwargs:
enable_thinking: true
```
Changes:
- `hermes_cli/config.py`: preserve `extra_body` in normalized
`custom_providers:` entries and allow it in the validated field set.
- `hermes_cli/runtime_provider.py`: propagate custom-provider `extra_body`
as `request_overrides.extra_body` for named custom runtime resolution,
including credential-pool paths.
- `agent/agent_init.py`: at agent init, locate the matching custom-provider
entry by `base_url` (+ optional model) and merge its `extra_body` into
`AIAgent.request_overrides`, with caller-provided overrides winning on
conflicting top-level keys.
- `plugins/model-providers/custom/__init__.py`: keep existing CustomProfile
behavior (Ollama `num_ctx`, `think=False` when reasoning disabled);
user-configured `extra_body` flows through `request_overrides`.
- `website/docs/integrations/providers.md`: document the explicit
`extra_body` override and the vLLM/Gemma `chat_template_kwargs` variant.
- Tests cover config normalization, runtime propagation, model matching,
trailing-slash equivalence, fallback when no `model` field is set, and
caller-override merging precedence.
Verified end-to-end against `CustomProfile` via `ChatCompletionsTransport`:
configured `extra_body` reaches `kwargs.extra_body` on the wire request,
and coexists with profile-generated entries (Ollama `num_ctx`, `think=False`)
without clobber.
Salvaged from #29022 onto current `main`. Cosmetic typing edit in
`plugins/model-providers/custom/__init__.py` and a stale-base docs revert
in `providers.md` were dropped during cherry-pick.
Closes#29022
* ci(tests): install ripgrep from prebuilt tarball instead of apt
apt-get update + install of ripgrep takes ~4 min on the GHA Ubuntu
runners (the apt-get update against archive.ubuntu.com is the slow
part; ripgrep itself is small). Switching to the upstream musl
binary tarball cuts the step to a few seconds.
- Pinned to ripgrep 15.1.0 with sha256 verification (same hash as
published in the releases sha256 sidecar file).
- Drops the `rg` binary into /usr/local/bin so it is on PATH for
every subsequent step without GITHUB_PATH manipulation.
- Applied to both the test and e2e jobs in tests.yml.
* fix(cli): compile syntax check to tempdir, not source __pycache__
`_validate_critical_files_syntax` runs `py_compile.compile()` on each
critical bootstrap file after a successful `git pull`. The default
`py_compile` writes the resulting `.pyc` next to the source under
`__pycache__/`, which causes two real problems:
1. Parallel test workers walking the same source tree (e.g. running
the suite under per-file process isolation) can race against each
other on the `__pycache__` write — manifests as flaky 'directory
not empty' errors during teardown.
2. In production, the post-pull syntax check leaves a `.pyc` behind
that the next interpreter run might pick up — fine when the
interpreter version matches, sketchy if it doesn't.
Fix: write the compiled output to a `tempfile.TemporaryDirectory()`
that's discarded on function exit. We only care about the compile-or-not
signal, not the artifact.
* test(runner): per-file process isolation, drop manual state reset + xdist
Replace fragile manual _reset_module_state test fixtures with robust
per-file subprocess isolation. Each test file runs in a fresh
`python -m pytest <file>` subprocess via ThreadPoolExecutor. No xdist,
no custom pytest plugin, no shared worker state.
Key changes:
* scripts/run_tests_parallel.py — new runner: discovers test files,
runs N in parallel via ThreadPoolExecutor, captures stdout per file,
treats exit code 5 (no tests collected) as pass, kills all children
on exit. Change from cpu_count to cpu_count*2. The runner is
I/O-bound (waiting on subprocess.communicate() from pytest children)
The parent process does almost no CPU work, so 2x oversubscription
keeps more pipes full. When a file fails, immediately show the last
30 lines of pytest output (stack traces + FAILED summary) plus a
ready-to-copy repro command:
python -m pytest tests/agent/test_auxiliary_client.py
* scripts/run_tests.sh — delegates to run_tests_parallel.py
* .github/workflows/tests.yml — test step: python
scripts/run_tests_parallel.py
* pyproject.toml — drop pytest-xdist, pytest-split; simplify addopts
* tests/conftest.py — remove ~200 lines of manual state-reset fixtures
* AGENTS.md — update Testing section for per-file design
* test(runner): speed gateway test antipattern scan up
* fix(test): web search provider plugin test missing xai
* fix(tests): make 14 test files pass under per-file subprocess isolation
Tests that relied on cross-file state pollution from xdist workers
fail when run in isolation (per-file subprocess model). Root causes
and fixes:
Tool registry not populated:
- test_video_generation_tool_surface_matrix: add discover_builtin_tools()
- test_web_providers_brave_free/ddgs/searxng/general: autouse fixtures
registering all 8 bundled web providers, reset after each test
- test_website_policy: same provider registration pattern
- test_web_tools_tavily: same pattern across 3 dispatch test classes
- Also add is_safe_url/check_website_access mocks where SSRF check
blocks example.com (DNS resolution fails in isolated envs)
Stale check_fn cache:
- test_kanban_tools: invalidate_check_fn_cache() + _clear_tool_defs_cache()
in both kanban guidance tests (prior test cached False for kanban_show)
- test_discord_tool: cache invalidation in setup/teardown
- test_homeassistant_tool: invalidate_check_fn_cache() before registry queries
Module-level state pollution:
- test_auxiliary_client: autouse fixture clearing _aux_unhealthy_until cache
- test_skill_commands: set_session_vars() instead of patch.dict(os.environ)
(ContextVar takes precedence over os.environ)
- test_dm_topics: overwrite sys.modules + separate telegram.constants mock
+ force-reimport of gateway.platforms.telegram
- test_terminal_tool_requirements: removed duplicate class declaration,
autouse _clear_caches fixture
* change(tests): run_tests.sh explicitly includes env vars
instead of manually dropping some vars, now we just only include some
* fix(tests): 5 more isolation/NixOS fixes
- test_approval_plugin_hooks: isolate HERMES_HOME so real user's
command_allowlist doesn't short-circuit the approval path
- test_google_chat: skipif when Platform.GOOGLE_CHAT not in enum
(feature not merged on this branch)
- test_write_deny: test systemd prefix against tmp_path instead of
/etc/systemd which resolves to /nix/store on NixOS
- test_pty_bridge: use shutil.which('cat') instead of /bin/cat
(doesn't exist on NixOS)
- profiles.py: rmtree onexc handler chmod's parent dirs too, fixing
profile deletion when copytree preserved read-only modes from
nix store
* fix(tests): clear unhealthy cache in autouse fixture for auxiliary_client
* fix(tests): skip send_message when telegram not installed; handle missing worker_id in browser_supervisor
* fix: py3.11 rmtree onexc compat + belt-and-suspenders unhealthy cache clear for expired codex test
* fix: address PR #29016 review feedback
- Remove tracked .pytest-cache/ artifact and add to .gitignore
- Fix stale 'xdist worker' comment in conftest.py
- Deduplicate web provider registration into tests/tools/conftest.py
shared helper (register_all_web_providers), replacing 8 copy-pasted
blocks across 6 test files
- Update PR description: remove stale recovered-test-files claim,
fix worker count to match code (cpu_count*2)
* fix: eliminate race in stale-cache achievements test
The background scan thread could complete and overwrite _SNAPSHOT_CACHE
before evaluate_all() returned the stale data — only 10 fake sessions
made the scan finish instantly. Added scan_delay param to _FakeSessionDB
and set it to 2s in the stale-cache test so the background thread can't
win the race.
- Replace 18-line comment block with 3-line invariant statement
- Trim test docstrings from multi-paragraph to single-line summaries
- Trim assertion messages from 4-line to 2-line mismatch reports
- Replace 5-line WHAT comments in stubs with 1-line WHY comments
- Add ziliangdotme@gmail.com -> ziliangpeng to AUTHOR_MAP
## Summary
The background skill/memory-review fork constructed a child `AIAgent`
without propagating `enabled_toolsets` / `disabled_toolsets` from the
parent. When the parent narrowed its toolset (via `hermes tools
disable` or `config.yaml`), the fork's default `enabled_toolsets=None`
expanded to "all registered tools" — and the fork's outbound request
body sent a wider `tools[]` array than the parent's main-turn request.
Anthropic's prompt-cache key includes the `tools[]` array byte-for-byte,
so this divergence forked the cache lineage on every nudge and forced a
full prefix rewrite. On a captured ~4 hour Claude-via-Hermes session
this cost roughly 4.3 M cache-write tokens — about half of those
attributable to the per-nudge alternation between the main turn's
narrowed `tools[]` and the review fork's wider `tools[]`.
## Goal
Extend the byte-stability invariant established by PR #17276 (which
fixed `system`) to the `tools[]` slot of the request body, so the
review fork's outbound request hits the parent's warmed Anthropic
prefix cache regardless of how the parent's toolset is configured.
## Implementation
Two-line change in `agent/background_review.py`: pass
`enabled_toolsets=getattr(agent, "enabled_toolsets", None)` and the
matching `disabled_toolsets` kwarg into the `AIAgent(...)` call inside
`_spawn_background_review`. Adds an explanatory block comment that
calls out the cache-key dependency and the relationship to PR #17276.
The post-construction runtime whitelist
(`set_thread_tool_whitelist({memory, skills})`) is untouched — it
still gates which tools the model is allowed to *dispatch*. This
change aligns only what the request body *transmits*, not what the
review is allowed to do, so the safety contract from issue #15204
remains intact.
## Testing
- `tests/run_agent/test_background_review_cache_parity.py`: new
`test_review_fork_inherits_parent_toolset_config` asserts the
parent's `enabled_toolsets` and `disabled_toolsets` reach the
review-fork constructor as kwargs.
- `tests/run_agent/test_background_review_toolset_restriction.py`:
the existing `test_background_review_does_not_narrow_toolset_schema`
was inverted (its old "must NOT pass enabled_toolsets" rule was
built on the assumption that the parent always ran with the
registry default — wrong in practice when the parent is narrowed).
Renamed to `test_background_review_matches_parent_toolset_config`
and updated to assert the parent's value propagates verbatim.
- Verified the new positive test fails without the fix and passes
with it.
- Full suite for `test_background_review*`:
```
$ python -m pytest tests/run_agent/test_background_review.py \
tests/run_agent/test_background_review_summary.py \
tests/run_agent/test_background_review_toolset_restriction.py \
tests/run_agent/test_background_review_cache_parity.py -q
18 passed in 1.85s
```
## Scope
- `agent/background_review.py`: 2 added kwargs + explanatory comment.
- Two test files: one new positive test, one inverted existing test.
- No production code paths outside the review fork; no schema changes;
no public-API changes.
Refs: ziliangpeng/hermes-agent#1 (root-cause analysis with wire-level
cache-write measurements). Extends PR #17276's `system`-bytes
invariant to the `tools[]` slot.
_handle_location_message and _handle_media_message were skipped when the
observe-unmentioned-group-messages feature landed (a9db0e2c7). Both handlers
now:
1. Check _should_observe_unmentioned_group_message on the skipped path and
call _observe_unmentioned_group_message so group chatter is stored as
shared session context even when the bot is not addressed.
2. Call _apply_telegram_group_observe_attribution on the triggered path so
the dispatched event uses the shared (user_id=None) group session instead
of the per-user session, letting the model see previously observed context.
For stickers the attribution is applied after _handle_sticker completes
(which overwrites event.text with the vision description); for all other
media types it is applied once after caption cleaning.
Four new tests cover the observe and attribution paths for both handlers.
build_write_denied_paths() resolved the protected ``.env`` via
get_hermes_home(), which is profile-aware. When a profile is active
HERMES_HOME points at ``<root>/profiles/<name>`` and ``hermes_home / ".env"``
expands to the *profile* env file only — the global ``<root>/.env`` is left
off the deny list and a write_file call against it succeeds. Since the
top-level .env supplies credentials inherited by every profile, this is a
P0 credential-exfiltration / overwrite path.
Add a parallel ``_hermes_root_path()`` helper that returns the Hermes root
(via the existing ``get_default_hermes_root()`` constant) and include
``<root>/.env`` in the deny list alongside ``<active_profile>/.env``. Both
paths now refuse write_file/patch regardless of profile state. The active
HERMES_HOME .env entry is preserved so the protection in non-profile mode
is unchanged.
A regression test exercises the profile-active scenario by pointing
HERMES_HOME at ``<tmp>/profiles/coder`` and asserting that ``<tmp>/.env``
is denied.
Fixes#15981
The contributor PR (#17936) only patched the strip path in
`_model_supports_vision()`. The auto-mode router in
`agent/image_routing._lookup_supports_vision` still only read models.dev,
so a custom-provider model declared as vision-capable would still get its
images routed through vision_analyze in the default `agent.image_input_mode:
auto` setting. Users had to set both `supports_vision: true` AND
`image_input_mode: native` to bypass the text pipeline.
Single-knob behavior now: `supports_vision: true` alone is enough in auto
mode. The strip path and the routing path consult the same resolver.
- Extract override resolution into `_supports_vision_override()` in
agent/image_routing.py and wire it into `_lookup_supports_vision()`.
- Refactor `run_agent._model_supports_vision` to call the same helper
(DRY, single source of truth for the resolution order).
- Strict YAML boolean coercion: `supports_vision: "false"` (quoted —
a common YAML mistake) no longer coerces to True via bool() truthiness.
Recognised tokens: true/false/yes/no/on/off/1/0 plus real bools and 0/1.
Unrecognised values return None and fall through to models.dev.
- Add @CNSeniorious000 to AUTHOR_MAP for release attribution.
Tests: 26 new (TestCoerceCapabilityBool, TestSupportsVisionOverride,
TestLookupSupportsVisionOverride, TestAutoModeRespectsOverride). Existing
contributor tests + image_routing + vision_native_fast_path +
native_image_buffer_isolation all green (92/92).
Named custom providers are rewritten to provider="custom" at runtime
(hermes_cli/runtime_provider.py:_resolve_named_custom_runtime), so a
config under providers.my-vllm.models.my-llava.supports_vision was
unreachable via self.provider alone. Also try cfg.model.provider as a
candidate provider key, covering both runtime and config naming.
Adds a regression test for the named-provider path.
Custom/local provider models absent from models.dev get classified as
non-vision and have their image content stripped before reaching the
upstream API. Surface a user-facing override:
model:
supports_vision: true
providers:
my-vllm:
models:
my-llava:
supports_vision: true
The override short-circuits the models.dev lookup in
_model_supports_vision(), which is the single gate guarding image-strip
preprocessing on every transport path.
Refs #8731.
xAI partner integration requires Hermes to thread `encrypted_content`
reasoning items back to the Responses API on every turn so Grok can
maintain cross-turn reasoning coherence. PR #26644 (May 15) gated this
off for `is_xai_responses` on the theory that the OAuth/SuperGrok
surface rejected replayed encrypted blobs and produced the multi-turn
"Expected to have received \`response.created\` before \`error\`"
failure. That diagnosis was wrong — the prelude-SSE fallback added in
the same PR is what actually fixed that failure mode. Suppressing the
replay was an unnecessary side-effect that broke the whole point of
xAI's partnership integration.
Changes:
- agent/codex_responses_adapter.py — drop the `is_xai_responses` gate
in `_chat_messages_to_responses_input`. Keep the kwarg in the
signature for transport compatibility; update the docstring to
document the May 2026 reversal.
- agent/transports/codex.py — restore
`kwargs["include"] = ["reasoning.encrypted_content"]` on the xAI
Responses path so xAI echoes encrypted reasoning back to us.
- tests/run_agent/test_codex_xai_oauth_recovery.py — flip the three
xAI assertions (now: xAI MUST receive replayed reasoning AND we MUST
include encrypted_content in the request).
- tests/agent/transports/test_codex_transport.py — flip the
`include` assertions on `test_xai_reasoning_effort_passed` and
`test_xai_grok_4_omits_reasoning_effort`; update the allowlist
block comment.
The prelude-SSE fallback and the entitlement-403 surfacing fixes from
#26644 are untouched — they were independent fixes that happened to
ride along with the reasoning-replay gate.
Validation:
- Targeted: tests/run_agent/test_codex_xai_oauth_recovery.py +
tests/agent/transports/test_codex_transport.py → 65/65 pass
- Broader: tests/agent/transports/ + tests/run_agent/ →
1674 passed, 3 skipped, 0 failures
- E2E (real imports, isolated HERMES_HOME, ResponsesApiTransport
build_kwargs): turn-1 request carries
`include: ["reasoning.encrypted_content"]`; turn-2 input replays
the encrypted_content blob from turn-1's
`codex_reasoning_items`; native Codex unchanged.
Five call sites do os.chmod(path.parent, 0o700) without checking that
the parent resolves to a safe directory. If HERMES_HOME or another
path env var resolves to /, the chmod strips traversal permission from
the root inode and bricks the entire host.
Add secure_parent_dir() to hermes_constants.py that refuses to chmod
/ or any top-level directory (depth < 2). Replace all 5 call sites
with this helper.
Fixes#25821
After #28660's host-gating fix, users with provider=custom and base_url
pointed at a commercial endpoint (DeepSeek, Groq, Mistral, …) hit
no-key-required even when they had the vendor-named env var set
(DEEPSEEK_API_KEY, GROQ_API_KEY, …). The issue author flagged this as
'what users intuitively expect'.
Adds _host_derived_api_key() to derive an env var name from the base URL
host using the *registrable* label (second-to-last). Appended to all three
api_key_candidates chains (_resolve_named_custom_runtime direct-alias path,
named-custom path, _resolve_openrouter_runtime non-openrouter branch).
Lookalike resistance: api.deepseek.com.attacker.test resolves to vendor
label 'attacker', NOT 'deepseek' — DEEPSEEK_API_KEY stays put. IPs and
loopback yield no vendor label. Already-handled vendors (OPENAI/OPENROUTER/
OLLAMA) are filtered to prevent bypass of the explicit host-gated paths.
Adds 6 tests covering positive paths (DeepSeek, Groq), the lookalike attack,
loopback rejection, the already-handled-vendor filter, and direct helper
unit tests.
Also adds erhnysr to AUTHOR_MAP.
- Preserve OPENROUTER_API_KEY for explicit mirror/proxy configs when
requested provider is openrouter and OPENROUTER_BASE_URL is set
- Gate OPENAI_API_KEY and OPENROUTER_API_KEY in named custom provider
path (_resolve_named_custom_runtime) on authoritative hosts
- Gate same keys in direct-alias path
- Update tests to reflect secure-by-default behavior for local endpoints
Custom endpoint provider was forwarding OPENAI_API_KEY and OLLAMA_API_KEY
to arbitrary hosts. Keys should only be sent to their authoritative domains
(openai.com, ollama.com) or when explicitly configured via pool/env.
- Gate OPENAI_API_KEY to openai.com hosts only
- Gate OLLAMA_API_KEY to ollama.com hosts only
- Return 'no-key-required' for unrecognized custom endpoints
- Update tests to reflect secure-by-default behavior
Closes#28660
Sibling fix on top of @EloquentBrush0x's PR #29441.
- tools/skills_hub.py GitHubSource.search() had the same r.name dedup bug.
Two configured GitHub taps publishing same-named skills would collapse to one.
- tests/hermes_cli/test_skills_hub.py:test_browse_skills_dedup_uses_identifier_not_name
patched hermes_cli.skills_hub.create_source_router, but browse_skills() imports
it locally from tools.skills_hub. Fixed patch path.
browse_skills() is the TUI gateway's API for the web UI skills browser
(tui_gateway/server.py:6574). It had the same dedup-by-name bug as
do_browse() and unified_search() fixed in the parent commit: r.name is
not unique for browse-sh skills (Airbnb, Booking.com, Zillow all publish
"search-listings"), so the dedup loop silently dropped all but the first
skill with each task name.
Switch to r.identifier, which is always globally unique.
Add a regression test asserting that two browse-sh skills with the same
name but different hostnames both appear in the browse_skills() result.
Browse.sh exposes skills by task name (e.g. "search-listings"), which is
shared across hundreds of sites. Deduplicating by name silently dropped
every browse-sh skill after the first one with a given task name — e.g.
only Airbnb's "search-listings" would survive, collapsing Booking.com,
Zillow, and every other site's variant into nothing.
Switch unified_search() and do_browse() to use r.identifier as the dedup
key. identifier is always globally unique (e.g.
"browse-sh/airbnb.com/search-listings-ddgioa"), so same-named skills from
different browse-sh hostnames are preserved as distinct results.
Update existing TestUnifiedSearchDedup tests to model the real scenario
(same identifier appearing from two sources) and add a regression test
that asserts browse-sh skills with the same name but different hostnames
are never collapsed.
Ben