fix(tests): accept unavailable_models kwarg in _prompt_model_selection mock

- test_update_check: replace patch("hermes_cli.banner.os.getenv") with
  monkeypatch.setenv("HERMES_HOME") — banner.py no longer imports os
  directly, it uses get_hermes_home() from hermes_constants.

- test_telegram_conflict/approval_buttons: provide real exception classes
  for telegram.error mock (NetworkError, TimedOut, BadRequest) so the
  except clause in connect() doesn't fail with "catching classes that do
  not inherit from BaseException" when xdist pollutes sys.modules.

.github/workflows/tests.yml

@@ -19,6 +19,9 @@ jobs:
       - name: Checkout code
         uses: actions/checkout@v4
 
+      - name: Install system dependencies
+        run: sudo apt-get update && sudo apt-get install -y ripgrep
+
       - name: Install uv
         uses: astral-sh/setup-uv@v5
 

acp_adapter/entry.py

@@ -15,7 +15,6 @@ Usage::
 
 import asyncio
 import logging
-import os
 import sys
 from pathlib import Path
 from hermes_constants import get_hermes_home

acp_adapter/session.py

@@ -262,8 +262,6 @@ class SessionManager:
         if self._db_instance is not None:
             return self._db_instance
         try:
-            import os
-            from pathlib import Path
             from hermes_state import SessionDB
             hermes_home = get_hermes_home()
             self._db_instance = SessionDB(db_path=hermes_home / "state.db")

acp_adapter/tools.py

@@ -39,7 +39,6 @@ TOOL_KIND_MAP: Dict[str, ToolKind] = {
     "browser_scroll": "execute",
     "browser_press": "execute",
     "browser_back": "execute",
-    "browser_close": "execute",
     "browser_get_images": "read",
     # Agent internals
     "delegate_task": "execute",

agent/anthropic_adapter.py

@@ -188,9 +188,7 @@ def _requires_bearer_auth(base_url: str | None) -> bool:
     if not base_url:
         return False
     normalized = base_url.rstrip("/").lower()
-    return normalized.startswith("https://api.minimax.io/anthropic") or normalized.startswith(
-        "https://api.minimaxi.com/anthropic"
-    )
+    return normalized.startswith(("https://api.minimax.io/anthropic", "https://api.minimaxi.com/anthropic"))
 
 
 def build_anthropic_client(api_key: str, base_url: str = None):
@@ -708,29 +706,6 @@ def run_hermes_oauth_login_pure() -> Optional[Dict[str, Any]]:
     }
 
 
-def run_hermes_oauth_login() -> Optional[str]:
-    """Run Hermes-native OAuth PKCE flow for Claude Pro/Max subscription.
-
-    Opens a browser to claude.ai for authorization, prompts for the code,
-    exchanges it for tokens, and stores them in ~/.hermes/.anthropic_oauth.json.
-
-    Returns the access token on success, None on failure.
-    """
-    result = run_hermes_oauth_login_pure()
-    if not result:
-        return None
-
-    access_token = result["access_token"]
-    refresh_token = result["refresh_token"]
-    expires_at_ms = result["expires_at_ms"]
-
-    _save_hermes_oauth_credentials(access_token, refresh_token, expires_at_ms)
-    _write_claude_code_credentials(access_token, refresh_token, expires_at_ms)
-
-    print("Authentication successful!")
-    return access_token
-
-
 def _save_hermes_oauth_credentials(access_token: str, refresh_token: str, expires_at_ms: int) -> None:
     """Save OAuth credentials to ~/.hermes/.anthropic_oauth.json."""
     data = {
@@ -758,38 +733,6 @@ def read_hermes_oauth_credentials() -> Optional[Dict[str, Any]]:
     return None
 
 
-def refresh_hermes_oauth_token() -> Optional[str]:
-    """Refresh the Hermes-managed OAuth token using the stored refresh token.
-
-    Returns the new access token, or None if refresh fails.
-    """
-    creds = read_hermes_oauth_credentials()
-    if not creds or not creds.get("refreshToken"):
-        return None
-
-    try:
-        refreshed = refresh_anthropic_oauth_pure(
-            creds["refreshToken"],
-            use_json=True,
-        )
-        _save_hermes_oauth_credentials(
-            refreshed["access_token"],
-            refreshed["refresh_token"],
-            refreshed["expires_at_ms"],
-        )
-        _write_claude_code_credentials(
-            refreshed["access_token"],
-            refreshed["refresh_token"],
-            refreshed["expires_at_ms"],
-        )
-        logger.debug("Successfully refreshed Hermes OAuth token")
-        return refreshed["access_token"]
-    except Exception as e:
-        logger.debug("Failed to refresh Hermes OAuth token: %s", e)
-
-    return None
-
-
 # ---------------------------------------------------------------------------
 # Message / tool / response format conversion
 # ---------------------------------------------------------------------------
@@ -847,7 +790,7 @@ def _convert_openai_image_part_to_anthropic(part: Dict[str, Any]) -> Optional[Di
                 },
             }
 
-    if url.startswith("http://") or url.startswith("https://"):
+    if url.startswith(("http://", "https://")):
         return {
             "type": "image",
             "source": {
@@ -859,35 +802,6 @@ def _convert_openai_image_part_to_anthropic(part: Dict[str, Any]) -> Optional[Di
     return None
 
 
-def _convert_user_content_part_to_anthropic(part: Any) -> Optional[Dict[str, Any]]:
-    if isinstance(part, dict):
-        ptype = part.get("type")
-        if ptype == "text":
-            block = {"type": "text", "text": part.get("text", "")}
-            if isinstance(part.get("cache_control"), dict):
-                block["cache_control"] = dict(part["cache_control"])
-            return block
-        if ptype == "image_url":
-            return _convert_openai_image_part_to_anthropic(part)
-        if ptype == "image" and part.get("source"):
-            return dict(part)
-        if ptype == "image" and part.get("data"):
-            media_type = part.get("mimeType") or part.get("media_type") or "image/png"
-            return {
-                "type": "image",
-                "source": {
-                    "type": "base64",
-                    "media_type": media_type,
-                    "data": part.get("data", ""),
-                },
-            }
-        if ptype == "tool_result":
-            return dict(part)
-    elif part is not None:
-        return {"type": "text", "text": str(part)}
-    return None
-
-
 def convert_tools_to_anthropic(tools: List[Dict]) -> List[Dict]:
     """Convert OpenAI tool definitions to Anthropic format."""
     if not tools:

agent/auxiliary_client.py

@@ -91,6 +91,7 @@ auxiliary_is_nous: bool = False
 # Default auxiliary models per provider
 _OPENROUTER_MODEL = "google/gemini-3-flash-preview"
 _NOUS_MODEL = "google/gemini-3-flash-preview"
+_NOUS_FREE_TIER_VISION_MODEL = "xiaomi/mimo-v2-omni"
 _NOUS_DEFAULT_BASE_URL = "https://inference-api.nousresearch.com/v1"
 _ANTHROPIC_DEFAULT_BASE_URL = "https://api.anthropic.com"
 _AUTH_JSON_PATH = get_hermes_home() / "auth.json"
@@ -208,7 +209,6 @@ class _CodexCompletionsAdapter:
     def create(self, **kwargs) -> Any:
         messages = kwargs.get("messages", [])
         model = kwargs.get("model", self._model)
-        temperature = kwargs.get("temperature")
 
         # Separate system/instructions from conversation messages.
         # Convert chat.completions multimodal content blocks to Responses
@@ -260,26 +260,73 @@ class _CodexCompletionsAdapter:
         usage = None
 
         try:
+            # Collect output items and text deltas during streaming —
+            # the Codex backend can return empty response.output from
+            # get_final_response() even when items were streamed.
+            collected_output_items: List[Any] = []
+            collected_text_deltas: List[str] = []
+            has_function_calls = False
             with self._client.responses.stream(**resp_kwargs) as stream:
                 for _event in stream:
-                    pass
+                    _etype = getattr(_event, "type", "")
+                    if _etype == "response.output_item.done":
+                        _done = getattr(_event, "item", None)
+                        if _done is not None:
+                            collected_output_items.append(_done)
+                    elif "output_text.delta" in _etype:
+                        _delta = getattr(_event, "delta", "")
+                        if _delta:
+                            collected_text_deltas.append(_delta)
+                    elif "function_call" in _etype:
+                        has_function_calls = True
                 final = stream.get_final_response()
 
-            # Extract text and tool calls from the Responses output
+            # Backfill empty output from collected stream events
+            _output = getattr(final, "output", None)
+            if isinstance(_output, list) and not _output:
+                if collected_output_items:
+                    final.output = list(collected_output_items)
+                    logger.debug(
+                        "Codex auxiliary: backfilled %d output items from stream events",
+                        len(collected_output_items),
+                    )
+                elif collected_text_deltas and not has_function_calls:
+                    # Only synthesize text when no tool calls were streamed —
+                    # a function_call response with incidental text should not
+                    # be collapsed into a plain-text message.
+                    assembled = "".join(collected_text_deltas)
+                    final.output = [SimpleNamespace(
+                        type="message", role="assistant", status="completed",
+                        content=[SimpleNamespace(type="output_text", text=assembled)],
+                    )]
+                    logger.debug(
+                        "Codex auxiliary: synthesized from %d deltas (%d chars)",
+                        len(collected_text_deltas), len(assembled),
+                    )
+
+            # Extract text and tool calls from the Responses output.
+            # Items may be SDK objects (attrs) or dicts (raw/fallback paths),
+            # so use a helper that handles both shapes.
+            def _item_get(obj: Any, key: str, default: Any = None) -> Any:
+                val = getattr(obj, key, None)
+                if val is None and isinstance(obj, dict):
+                    val = obj.get(key, default)
+                return val if val is not None else default
+
             for item in getattr(final, "output", []):
-                item_type = getattr(item, "type", None)
+                item_type = _item_get(item, "type")
                 if item_type == "message":
-                    for part in getattr(item, "content", []):
-                        ptype = getattr(part, "type", None)
+                    for part in (_item_get(item, "content") or []):
+                        ptype = _item_get(part, "type")
                         if ptype in ("output_text", "text"):
-                            text_parts.append(getattr(part, "text", ""))
+                            text_parts.append(_item_get(part, "text", ""))
                 elif item_type == "function_call":
                     tool_calls_raw.append(SimpleNamespace(
-                        id=getattr(item, "call_id", ""),
+                        id=_item_get(item, "call_id", ""),
                         type="function",
                         function=SimpleNamespace(
-                            name=getattr(item, "name", ""),
-                            arguments=getattr(item, "arguments", "{}"),
+                            name=_item_get(item, "name", ""),
+                            arguments=_item_get(item, "arguments", "{}"),
                         ),
                     ))
 
@@ -673,7 +720,19 @@ def _try_nous() -> Tuple[Optional[OpenAI], Optional[str]]:
     global auxiliary_is_nous
     auxiliary_is_nous = True
     logger.debug("Auxiliary client: Nous Portal")
-    model = "gemini-3-flash" if nous.get("source") == "pool" else _NOUS_MODEL
+    if nous.get("source") == "pool":
+        model = "gemini-3-flash"
+    else:
+        model = _NOUS_MODEL
+    # Free-tier users can't use paid auxiliary models — use the free
+    # multimodal model instead so vision/browser-vision still works.
+    try:
+        from hermes_cli.models import check_nous_free_tier
+        if check_nous_free_tier():
+            model = _NOUS_FREE_TIER_VISION_MODEL
+            logger.debug("Free-tier Nous account — using %s for auxiliary/vision", model)
+    except Exception:
+        pass
     return (
         OpenAI(
             api_key=_nous_api_key(nous),
@@ -849,7 +908,7 @@ def _resolve_forced_provider(forced: str) -> Tuple[Optional[OpenAI], Optional[st
     if forced == "nous":
         client, model = _try_nous()
         if client is None:
-            logger.warning("auxiliary.provider=nous but Nous Portal not configured (run: hermes login)")
+            logger.warning("auxiliary.provider=nous but Nous Portal not configured (run: hermes auth)")
         return client, model
 
     if forced == "codex":
@@ -1119,7 +1178,7 @@ def resolve_provider_client(
         client, default = _try_nous()
         if client is None:
             logger.warning("resolve_provider_client: nous requested "
-                           "but Nous Portal not configured (run: hermes login)")
+                           "but Nous Portal not configured (run: hermes auth)")
             return None, None
         final_model = model or default
         return (_to_async_client(client, final_model) if async_mode

agent/builtin_memory_provider.py

@@ -13,9 +13,10 @@ from __future__ import annotations
 
 import json
 import logging
-from typing import Any, Dict, List, Optional
+from typing import Any, Dict, List
 
 from agent.memory_provider import MemoryProvider
+from tools.registry import tool_error
 
 logger = logging.getLogger(__name__)
 
@@ -92,7 +93,7 @@ class BuiltinMemoryProvider(MemoryProvider):
 
     def handle_tool_call(self, tool_name: str, args: Dict[str, Any], **kwargs) -> str:
         """Not used — the memory tool is intercepted in run_agent.py."""
-        return json.dumps({"error": "Built-in memory tool is handled by the agent loop"})
+        return tool_error("Built-in memory tool is handled by the agent loop")
 
     def shutdown(self) -> None:
         """No cleanup needed — files are saved on every write."""

agent/context_references.py

@@ -343,10 +343,9 @@ def _resolve_path(cwd: Path, target: str, *, allowed_root: Path | None = None) -
 
 
 def _ensure_reference_path_allowed(path: Path) -> None:
+    from hermes_constants import get_hermes_home
     home = Path(os.path.expanduser("~")).resolve()
-    hermes_home = Path(
-        os.getenv("HERMES_HOME", str(home / ".hermes"))
-    ).expanduser().resolve()
+    hermes_home = get_hermes_home().resolve()
 
     blocked_exact = {home / rel for rel in _SENSITIVE_HOME_FILES}
     blocked_exact.add(hermes_home / ".env")

agent/credential_pool.py

@@ -10,22 +10,21 @@ import uuid
 import os
 import re
 from dataclasses import dataclass, fields, replace
-from datetime import datetime, timezone
+from datetime import datetime
 from typing import Any, Dict, List, Optional, Set, Tuple
 
 from hermes_constants import OPENROUTER_BASE_URL
 import hermes_cli.auth as auth_mod
 from hermes_cli.auth import (
-    ACCESS_TOKEN_REFRESH_SKEW_SECONDS,
     CODEX_ACCESS_TOKEN_REFRESH_SKEW_SECONDS,
     DEFAULT_AGENT_KEY_MIN_TTL_SECONDS,
     PROVIDER_REGISTRY,
-    _agent_key_is_usable,
     _codex_access_token_is_expiring,
     _decode_jwt_claims,
-    _is_expiring,
+    _import_codex_cli_tokens,
     _load_auth_store,
     _load_provider_state,
+    _resolve_zai_base_url,
     read_credential_pool,
     write_credential_pool,
 )
@@ -347,6 +346,9 @@ def get_pool_strategy(provider: str) -> str:
     return STRATEGY_FILL_FIRST
 
 
+DEFAULT_MAX_CONCURRENT_PER_CREDENTIAL = 1
+
+
 class CredentialPool:
     def __init__(self, provider: str, entries: List[PooledCredential]):
         self.provider = provider
@@ -354,6 +356,8 @@ class CredentialPool:
         self._current_id: Optional[str] = None
         self._strategy = get_pool_strategy(provider)
         self._lock = threading.Lock()
+        self._active_leases: Dict[str, int] = {}
+        self._max_concurrent = DEFAULT_MAX_CONCURRENT_PER_CREDENTIAL
 
     def has_credentials(self) -> bool:
         return bool(self._entries)
@@ -440,6 +444,39 @@ class CredentialPool:
             logger.debug("Failed to sync from credentials file: %s", exc)
         return entry
 
+    def _sync_codex_entry_from_cli(self, entry: PooledCredential) -> PooledCredential:
+        """Sync an openai-codex pool entry from ~/.codex/auth.json if tokens differ.
+
+        OpenAI OAuth refresh tokens are single-use and rotate on every refresh.
+        When the Codex CLI (or another Hermes profile) refreshes its token,
+        the pool entry's refresh_token becomes stale.  This method detects that
+        by comparing against ~/.codex/auth.json and syncing the fresh pair.
+        """
+        if self.provider != "openai-codex":
+            return entry
+        try:
+            cli_tokens = _import_codex_cli_tokens()
+            if not cli_tokens:
+                return entry
+            cli_refresh = cli_tokens.get("refresh_token", "")
+            cli_access = cli_tokens.get("access_token", "")
+            if cli_refresh and cli_refresh != entry.refresh_token:
+                logger.debug("Pool entry %s: syncing tokens from ~/.codex/auth.json (refresh token changed)", entry.id)
+                updated = replace(
+                    entry,
+                    access_token=cli_access,
+                    refresh_token=cli_refresh,
+                    last_status=None,
+                    last_status_at=None,
+                    last_error_code=None,
+                )
+                self._replace_entry(entry, updated)
+                self._persist()
+                return updated
+        except Exception as exc:
+            logger.debug("Failed to sync from ~/.codex/auth.json: %s", exc)
+        return entry
+
     def _refresh_entry(self, entry: PooledCredential, *, force: bool) -> Optional[PooledCredential]:
         if entry.auth_type != AUTH_TYPE_OAUTH or not entry.refresh_token:
             if force:
@@ -629,6 +666,16 @@ class CredentialPool:
                 if synced is not entry:
                     entry = synced
                     cleared_any = True
+            # For openai-codex entries, sync from ~/.codex/auth.json before
+            # any status/refresh checks.  This picks up tokens refreshed by
+            # the Codex CLI or another Hermes profile.
+            if (self.provider == "openai-codex"
+                    and entry.last_status == STATUS_EXHAUSTED
+                    and entry.refresh_token):
+                synced = self._sync_codex_entry_from_cli(entry)
+                if synced is not entry:
+                    entry = synced
+                    cleared_any = True
             if entry.last_status == STATUS_EXHAUSTED:
                 exhausted_until = _exhausted_until(entry)
                 if exhausted_until is not None and now < exhausted_until:
@@ -716,6 +763,51 @@ class CredentialPool:
                 logger.info("credential pool: rotated to %s", _next_label)
             return next_entry
 
+    def acquire_lease(self, credential_id: Optional[str] = None) -> Optional[str]:
+        """Acquire a soft lease on a credential.
+
+        If a specific credential_id is provided, lease that entry directly.
+        Otherwise prefer the least-leased available credential, using priority as
+        a stable tie-breaker. When every credential is already at the soft cap,
+        still return the least-leased one instead of blocking.
+        """
+        with self._lock:
+            if credential_id:
+                self._active_leases[credential_id] = self._active_leases.get(credential_id, 0) + 1
+                self._current_id = credential_id
+                return credential_id
+
+            available = self._available_entries(clear_expired=True, refresh=True)
+            if not available:
+                return None
+
+            below_cap = [
+                entry for entry in available
+                if self._active_leases.get(entry.id, 0) < self._max_concurrent
+            ]
+            candidates = below_cap if below_cap else available
+            chosen = min(
+                candidates,
+                key=lambda entry: (self._active_leases.get(entry.id, 0), entry.priority),
+            )
+            self._active_leases[chosen.id] = self._active_leases.get(chosen.id, 0) + 1
+            self._current_id = chosen.id
+            return chosen.id
+
+    def release_lease(self, credential_id: str) -> None:
+        """Release a previously acquired credential lease."""
+        with self._lock:
+            count = self._active_leases.get(credential_id, 0)
+            if count <= 1:
+                self._active_leases.pop(credential_id, None)
+            else:
+                self._active_leases[credential_id] = count - 1
+
+    def active_lease_count(self, credential_id: str) -> int:
+        """Return the number of active leases for a credential."""
+        with self._lock:
+            return self._active_leases.get(credential_id, 0)
+
     def try_refresh_current(self) -> Optional[PooledCredential]:
         with self._lock:
             return self._try_refresh_current_unlocked()
@@ -992,6 +1084,8 @@ def _seed_from_env(provider: str, entries: List[PooledCredential]) -> Tuple[bool
         active_sources.add(source)
         auth_type = AUTH_TYPE_OAUTH if provider == "anthropic" and not token.startswith("sk-ant-api") else AUTH_TYPE_API_KEY
         base_url = env_url or pconfig.inference_base_url
+        if provider == "zai":
+            base_url = _resolve_zai_base_url(token, pconfig.inference_base_url, env_url)
         changed |= _upsert_entry(
             entries,
             provider,

agent/display.py

@@ -890,8 +890,6 @@ def get_cute_tool_message(
         return _wrap(f"┊ ◀️  back      {dur}")
     if tool_name == "browser_press":
         return _wrap(f"┊ ⌨️  press     {args.get('key', '?')}  {dur}")
-    if tool_name == "browser_close":
-        return _wrap(f"┊ 🚪 close     browser  {dur}")
     if tool_name == "browser_get_images":
         return _wrap(f"┊ 🖼️  images    extracting  {dur}")
     if tool_name == "browser_vision":
@@ -988,24 +986,6 @@ def _osc8_link(url: str, text: str) -> str:
     return f"\033]8;;{url}\033\\{text}\033]8;;\033\\"
 
 
-def honcho_session_line(workspace: str, session_name: str) -> str:
-    """One-line session indicator: `Honcho session: <clickable name>`."""
-    url = honcho_session_url(workspace, session_name)
-    linked_name = _osc8_link(url, f"{_SKY_BLUE}{session_name}{_ANSI_RESET}")
-    return f"{_DIM}Honcho session:{_ANSI_RESET} {linked_name}"
-
-
-def write_tty(text: str) -> None:
-    """Write directly to /dev/tty, bypassing stdout capture."""
-    try:
-        fd = os.open("/dev/tty", os.O_WRONLY)
-        os.write(fd, text.encode("utf-8"))
-        os.close(fd)
-    except OSError:
-        sys.stdout.write(text)
-        sys.stdout.flush()
-
-
 # =========================================================================
 # Context pressure display (CLI user-facing warnings)
 # =========================================================================

agent/memory_manager.py

@@ -34,6 +34,7 @@ import re
 from typing import Any, Dict, List, Optional
 
 from agent.memory_provider import MemoryProvider
+from tools.registry import tool_error
 
 logger = logging.getLogger(__name__)
 
@@ -249,7 +250,7 @@ class MemoryManager:
         """
         provider = self._tool_to_provider.get(tool_name)
         if provider is None:
-            return json.dumps({"error": f"No memory provider handles tool '{tool_name}'"})
+            return tool_error(f"No memory provider handles tool '{tool_name}'")
         try:
             return provider.handle_tool_call(tool_name, args, **kwargs)
         except Exception as e:
@@ -257,7 +258,7 @@ class MemoryManager:
                 "Memory provider '%s' handle_tool_call(%s) failed: %s",
                 provider.name, tool_name, e,
             )
-            return json.dumps({"error": f"Memory tool '{tool_name}' failed: {e}"})
+            return tool_error(f"Memory tool '{tool_name}' failed: {e}")
 
     # -- Lifecycle hooks -----------------------------------------------------
 

agent/memory_provider.py

@@ -34,7 +34,7 @@ from __future__ import annotations
 
 import logging
 from abc import ABC, abstractmethod
-from typing import Any, Dict, List, Optional
+from typing import Any, Dict, List
 
 logger = logging.getLogger(__name__)
 

agent/model_metadata.py

@@ -510,8 +510,8 @@ def fetch_endpoint_model_metadata(
 
 def _get_context_cache_path() -> Path:
     """Return path to the persistent context length cache file."""
-    hermes_home = Path(os.environ.get("HERMES_HOME", Path.home() / ".hermes"))
-    return hermes_home / "context_length_cache.yaml"
+    from hermes_constants import get_hermes_home
+    return get_hermes_home() / "context_length_cache.yaml"
 
 
 def _load_context_cache() -> Dict[str, int]:

agent/models_dev.py

@@ -23,9 +23,9 @@ import json
 import logging
 import os
 import time
-from dataclasses import dataclass, field
+from dataclasses import dataclass
 from pathlib import Path
-from typing import Any, Dict, List, Optional, Tuple, Union
+from typing import Any, Dict, List, Optional, Tuple
 
 from utils import atomic_json_write
 
@@ -185,9 +185,8 @@ def _get_reverse_mapping() -> Dict[str, str]:
 
 def _get_cache_path() -> Path:
     """Return path to disk cache file."""
-    env_val = os.environ.get("HERMES_HOME", "")
-    hermes_home = Path(env_val) if env_val else Path.home() / ".hermes"
-    return hermes_home / "models_dev_cache.json"
+    from hermes_constants import get_hermes_home
+    return get_hermes_home() / "models_dev_cache.json"
 
 
 def _load_disk_cache() -> Dict[str, Any]:
@@ -231,7 +230,7 @@ def fetch_models_dev(force_refresh: bool = False) -> Dict[str, Any]:
         response = requests.get(MODELS_DEV_URL, timeout=15)
         response.raise_for_status()
         data = response.json()
-        if isinstance(data, dict) and len(data) > 0:
+        if isinstance(data, dict) and data:
             _models_dev_cache = data
             _models_dev_cache_time = time.time()
             _save_disk_cache(data)

agent/prompt_builder.py

@@ -744,7 +744,6 @@ def build_nous_subscription_prompt(valid_tool_names: "set[str] | None" = None) -
         "browser_type",
         "browser_scroll",
         "browser_console",
-        "browser_close",
         "browser_press",
         "browser_get_images",
         "browser_vision",
@@ -774,13 +773,13 @@ def build_nous_subscription_prompt(valid_tool_names: "set[str] | None" = None) -
 
     lines = [
         "# Nous Subscription",
-        "Nous subscription includes managed web tools (Firecrawl), image generation (FAL), OpenAI TTS, and browser automation (Browserbase) by default. Modal execution is optional.",
+        "Nous subscription includes managed web tools (Firecrawl), image generation (FAL), OpenAI TTS, and browser automation (Browser Use) by default. Modal execution is optional.",
         "Current capability status:",
     ]
     lines.extend(_status_line(feature) for feature in features.items())
     lines.extend(
         [
-            "When a Nous-managed feature is active, do not ask the user for Firecrawl, FAL, OpenAI TTS, or Browserbase API keys.",
+            "When a Nous-managed feature is active, do not ask the user for Firecrawl, FAL, OpenAI TTS, or Browser-Use API keys.",
             "If the user is not subscribed and asks for a capability that Nous subscription would unlock or simplify, suggest Nous subscription as one option alongside direct setup or local alternatives.",
             "Do not mention subscription unless the user asks about it or it directly solves the current missing capability.",
             "Useful commands: hermes setup, hermes setup tools, hermes setup terminal, hermes status.",

agent/skill_utils.py

@@ -10,7 +10,7 @@ import os
 import re
 import sys
 from pathlib import Path
-from typing import Any, Dict, List, Optional, Set, Tuple
+from typing import Any, Dict, List, Set, Tuple
 
 from hermes_constants import get_hermes_home
 

agent/subdirectory_hints.py

@@ -15,7 +15,6 @@ Inspired by Block/goose's SubdirectoryHintTracker.
 
 import logging
 import os
-import re
 import shlex
 from pathlib import Path
 from typing import Dict, Any, Optional, Set

batch_runner.py

@@ -31,6 +31,8 @@ from multiprocessing import Pool, Lock
 import traceback
 from rich.progress import Progress, SpinnerColumn, BarColumn, TextColumn, TimeRemainingColumn, MofNCompleteColumn
 from rich.console import Console
+
+logger = logging.getLogger(__name__)
 import fire
 
 from run_agent import AIAgent
@@ -1016,7 +1018,7 @@ class BatchRunner:
                             tool_stats = data.get('tool_stats', {})
                             
                             # Check for invalid tool names (model hallucinations)
-                            invalid_tools = [k for k in tool_stats.keys() if k not in VALID_TOOLS]
+                            invalid_tools = [k for k in tool_stats if k not in VALID_TOOLS]
                             
                             if invalid_tools:
                                 filtered_entries += 1

cli-config.yaml.example

@@ -539,7 +539,7 @@ platform_toolsets:
 #   terminal     - terminal, process
 #   file         - read_file, write_file, patch, search
 #   browser      - browser_navigate, browser_snapshot, browser_click, browser_type,
-#                  browser_scroll, browser_back, browser_press, browser_close,
+#                  browser_scroll, browser_back, browser_press,
 #                  browser_get_images, browser_vision  (requires BROWSERBASE_API_KEY)
 #   vision       - vision_analyze  (requires OPENROUTER_API_KEY)
 #   image_gen    - image_generate  (requires FAL_KEY)

cli.py

@@ -70,7 +70,7 @@ _COMMAND_SPINNER_FRAMES = ("⠋", "⠙", "⠹", "⠸", "⠼", "⠴", "⠦", "⠧
 
 # Load .env from ~/.hermes/.env first, then project root as dev fallback.
 # User-managed env files should override stale shell exports on restart.
-from hermes_constants import get_hermes_home, display_hermes_home, OPENROUTER_BASE_URL
+from hermes_constants import get_hermes_home, display_hermes_home
 from hermes_cli.env_loader import load_hermes_dotenv
 
 _hermes_home = get_hermes_home()
@@ -1920,6 +1920,12 @@ class HermesCLI:
             _cprint(f"{_DIM}└{'─' * (w - 2)}┘{_RST}")
             self._reasoning_box_opened = False
 
+            # Flush any content that was deferred while reasoning was rendering.
+            deferred = getattr(self, "_deferred_content", "")
+            if deferred:
+                self._deferred_content = ""
+                self._emit_stream_text(deferred)
+
     def _stream_delta(self, text) -> None:
         """Line-buffered streaming callback for real-time token rendering.
 
@@ -2022,6 +2028,13 @@ class HermesCLI:
         if not text:
             return
 
+        # When show_reasoning is on and reasoning is still rendering,
+        # defer content until the reasoning box closes.  This ensures the
+        # reasoning block always appears BEFORE the response in the terminal.
+        if self.show_reasoning and getattr(self, "_reasoning_box_opened", False):
+            self._deferred_content = getattr(self, "_deferred_content", "") + text
+            return
+
         # Close the live reasoning box before opening the response box
         self._close_reasoning_box()
 
@@ -2088,6 +2101,7 @@ class HermesCLI:
         self._reasoning_box_opened = False
         self._reasoning_buf = ""
         self._reasoning_preview_buf = ""
+        self._deferred_content = ""
 
     def _slow_command_status(self, command: str) -> str:
         """Return a user-facing status message for slower slash commands."""
@@ -3522,13 +3536,6 @@ class HermesCLI:
         _cprint(f"  Original session: {parent_session_id}")
         _cprint(f"  Branch session:   {new_session_id}")
 
-    def reset_conversation(self):
-        """Reset the conversation by starting a new session."""
-        # Shut down memory provider before resetting — actual session boundary
-        if hasattr(self, 'agent') and self.agent:
-            self.agent.shutdown_memory_provider(self.conversation_history)
-        self.new_session()
-    
     def save_conversation(self):
         """Save the current conversation to a file."""
         if not self.conversation_history:
@@ -4232,7 +4239,6 @@ class HermesCLI:
         
         try:
             config = load_gateway_config()
-            connected = config.get_connected_platforms()
             
             print("  Messaging Platform Configuration:")
             print("  " + "-" * 55)
@@ -5023,13 +5029,13 @@ class HermesCLI:
                     pass
                 print()
                 print("🌐 Browser disconnected from live Chrome")
-                print("   Browser tools reverted to default mode (local headless or Browserbase)")
+                print("   Browser tools reverted to default mode (local headless or cloud provider)")
                 print()
 
                 if hasattr(self, '_pending_input'):
                     self._pending_input.put(
                         "[System note: The user has disconnected the browser tools from their live Chrome. "
-                        "Browser tools are back to default mode (headless local browser or Browserbase cloud).]"
+                        "Browser tools are back to default mode (headless local browser or cloud provider).]"
                     )
             else:
                 print()
@@ -5056,10 +5062,17 @@ class HermesCLI:
                     print("   Status: ✓ reachable")
                 except (OSError, Exception):
                     print("   Status: ⚠ not reachable (Chrome may not be running)")
-            elif os.environ.get("BROWSERBASE_API_KEY"):
-                print("🌐 Browser: Browserbase (cloud)")
             else:
-                print("🌐 Browser: local headless Chromium (agent-browser)")
+                try:
+                    from tools.browser_tool import _get_cloud_provider
+                    provider = _get_cloud_provider()
+                except Exception:
+                    provider = None
+
+                if provider is not None:
+                    print(f"🌐 Browser: {provider.provider_name()} (cloud)")
+                else:
+                    print("🌐 Browser: local headless Chromium (agent-browser)")
             print()
             print("   /browser connect      — connect to your live Chrome")
             print("   /browser disconnect   — revert to default")
@@ -5987,7 +6000,7 @@ class HermesCLI:
 
         timeout = CLI_CONFIG.get("clarify", {}).get("timeout", 120)
         response_queue = queue.Queue()
-        is_open_ended = not choices or len(choices) == 0
+        is_open_ended = not choices
 
         self._clarify_state = {
             "question": question,
@@ -6270,14 +6283,6 @@ class HermesCLI:
             except Exception:
                 pass
 
-    def _clear_current_input(self) -> None:
-        if getattr(self, "_app", None):
-            try:
-                self._app.current_buffer.text = ""
-            except Exception:
-                pass
-
-
     def chat(self, message, images: list = None) -> Optional[str]:
         """
         Send a message to the agent and get a response.
@@ -7818,7 +7823,6 @@ class HermesCLI:
             title = '🔐 Sudo Password Required'
             body = 'Enter password below (hidden), or press Enter to skip'
             box_width = _panel_box_width(title, [body])
-            inner = max(0, box_width - 2)
             lines = []
             lines.append(('class:sudo-border', '╭─ '))
             lines.append(('class:sudo-title', title))
@@ -8120,6 +8124,25 @@ class HermesCLI:
                         # Periodic config watcher — auto-reload MCP on mcp_servers change
                         if not self._agent_running:
                             self._check_config_mcp_changes()
+                            # Check for background process completion notifications
+                            # while the agent is idle (user hasn't typed anything yet).
+                            try:
+                                from tools.process_registry import process_registry
+                                if not process_registry.completion_queue.empty():
+                                    completion = process_registry.completion_queue.get_nowait()
+                                    _exit = completion.get("exit_code", "?")
+                                    _cmd = completion.get("command", "unknown")
+                                    _sid = completion.get("session_id", "unknown")
+                                    _out = completion.get("output", "")
+                                    _synth = (
+                                        f"[SYSTEM: Background process {_sid} completed "
+                                        f"(exit code {_exit}).\n"
+                                        f"Command: {_cmd}\n"
+                                        f"Output:\n{_out}]"
+                                    )
+                                    self._pending_input.put(_synth)
+                            except Exception:
+                                pass
                         continue
                     
                     if not user_input:
@@ -8233,7 +8256,29 @@ class HermesCLI:
                                 except Exception as e:
                                     _cprint(f"{_DIM}Voice auto-restart failed: {e}{_RST}")
                             threading.Thread(target=_restart_recording, daemon=True).start()
-                    
+
+                        # Drain process completion notifications — any background
+                        # process that finished with notify_on_complete while the
+                        # agent was running (or before) gets auto-injected as a
+                        # new user message so the agent can react to it.
+                        try:
+                            from tools.process_registry import process_registry
+                            while not process_registry.completion_queue.empty():
+                                completion = process_registry.completion_queue.get_nowait()
+                                _exit = completion.get("exit_code", "?")
+                                _cmd = completion.get("command", "unknown")
+                                _sid = completion.get("session_id", "unknown")
+                                _out = completion.get("output", "")
+                                _synth = (
+                                    f"[SYSTEM: Background process {_sid} completed "
+                                    f"(exit code {_exit}).\n"
+                                    f"Command: {_cmd}\n"
+                                    f"Output:\n{_out}]"
+                                )
+                                self._pending_input.put(_synth)
+                        except Exception:
+                            pass  # Non-fatal — don't break the main loop
+
                 except Exception as e:
                     print(f"Error: {e}")
         

cron/scheduler.py

@@ -25,7 +25,6 @@ except ImportError:
         import msvcrt
     except ImportError:
         msvcrt = None
-import time
 from pathlib import Path
 from typing import Optional
 
@@ -159,6 +158,44 @@ def _resolve_delivery_target(job: dict) -> Optional[dict]:
     }
 
 
+# Media extension sets — keep in sync with gateway/platforms/base.py:_process_message_background
+_AUDIO_EXTS = frozenset({'.ogg', '.opus', '.mp3', '.wav', '.m4a'})
+_VIDEO_EXTS = frozenset({'.mp4', '.mov', '.avi', '.mkv', '.webm', '.3gp'})
+_IMAGE_EXTS = frozenset({'.jpg', '.jpeg', '.png', '.webp', '.gif'})
+
+
+def _send_media_via_adapter(adapter, chat_id: str, media_files: list, metadata: dict | None, loop, job: dict) -> None:
+    """Send extracted MEDIA files as native platform attachments via a live adapter.
+
+    Routes each file to the appropriate adapter method (send_voice, send_image_file,
+    send_video, send_document) based on file extension — mirroring the routing logic
+    in ``BasePlatformAdapter._process_message_background``.
+    """
+    from pathlib import Path
+
+    for media_path, _is_voice in media_files:
+        try:
+            ext = Path(media_path).suffix.lower()
+            if ext in _AUDIO_EXTS:
+                coro = adapter.send_voice(chat_id=chat_id, audio_path=media_path, metadata=metadata)
+            elif ext in _VIDEO_EXTS:
+                coro = adapter.send_video(chat_id=chat_id, video_path=media_path, metadata=metadata)
+            elif ext in _IMAGE_EXTS:
+                coro = adapter.send_image_file(chat_id=chat_id, image_path=media_path, metadata=metadata)
+            else:
+                coro = adapter.send_document(chat_id=chat_id, file_path=media_path, metadata=metadata)
+
+            future = asyncio.run_coroutine_threadsafe(coro, loop)
+            result = future.result(timeout=30)
+            if result and not getattr(result, "success", True):
+                logger.warning(
+                    "Job '%s': media send failed for %s: %s",
+                    job.get("id", "?"), media_path, getattr(result, "error", "unknown"),
+                )
+        except Exception as e:
+            logger.warning("Job '%s': failed to send media %s: %s", job.get("id", "?"), media_path, e)
+
+
 def _deliver_result(job: dict, content: str, adapters=None, loop=None) -> None:
     """
     Deliver job output to the configured target (origin chat, specific platform, etc.).
@@ -247,18 +284,28 @@ def _deliver_result(job: dict, content: str, adapters=None, loop=None) -> None:
     if runtime_adapter is not None and loop is not None and getattr(loop, "is_running", lambda: False)():
         send_metadata = {"thread_id": thread_id} if thread_id else None
         try:
-            future = asyncio.run_coroutine_threadsafe(
-                runtime_adapter.send(chat_id, delivery_content, metadata=send_metadata),
-                loop,
-            )
-            send_result = future.result(timeout=60)
-            if send_result and not getattr(send_result, "success", True):
-                err = getattr(send_result, "error", "unknown")
-                logger.warning(
-                    "Job '%s': live adapter send to %s:%s failed (%s), falling back to standalone",
-                    job["id"], platform_name, chat_id, err,
+            # Send cleaned text (MEDIA tags stripped) — not the raw content
+            text_to_send = cleaned_delivery_content.strip()
+            adapter_ok = True
+            if text_to_send:
+                future = asyncio.run_coroutine_threadsafe(
+                    runtime_adapter.send(chat_id, text_to_send, metadata=send_metadata),
+                    loop,
                 )
-            else:
+                send_result = future.result(timeout=60)
+                if send_result and not getattr(send_result, "success", True):
+                    err = getattr(send_result, "error", "unknown")
+                    logger.warning(
+                        "Job '%s': live adapter send to %s:%s failed (%s), falling back to standalone",
+                        job["id"], platform_name, chat_id, err,
+                    )
+                    adapter_ok = False  # fall through to standalone path
+
+            # Send extracted media files as native attachments via the live adapter
+            if adapter_ok and media_files:
+                _send_media_via_adapter(runtime_adapter, chat_id, media_files, send_metadata, loop, job)
+
+            if adapter_ok:
                 logger.info("Job '%s': delivered to %s:%s via live adapter", job["id"], platform_name, chat_id)
                 return
         except Exception as e:

gateway/builtin_hooks/boot_md.py

@@ -24,7 +24,8 @@ from pathlib import Path
 
 logger = logging.getLogger("hooks.boot-md")
 
-HERMES_HOME = Path(os.environ.get("HERMES_HOME", Path.home() / ".hermes"))
+from hermes_constants import get_hermes_home
+HERMES_HOME = get_hermes_home()
 BOOT_FILE = HERMES_HOME / "BOOT.md"
 
 

gateway/channel_directory.py

@@ -124,7 +124,6 @@ def _build_discord(adapter) -> List[Dict[str, str]]:
 
 def _build_slack(adapter) -> List[Dict[str, str]]:
     """List Slack channels the bot has joined."""
-    channels = []
     # Slack adapter may expose a web client
     client = getattr(adapter, "_app", None) or getattr(adapter, "_client", None)
     if not client:

gateway/delivery.py

@@ -314,38 +314,4 @@ def parse_deliver_spec(
     return deliver
 
 
-def build_delivery_context_for_tool(
-    config: GatewayConfig,
-    origin: Optional[SessionSource] = None
-) -> Dict[str, Any]:
-    """
-    Build context for the unified cronjob tool to understand delivery options.
-    
-    This is passed to the tool so it can validate and explain delivery targets.
-    """
-    connected = config.get_connected_platforms()
-    
-    options = {
-        "origin": {
-            "description": "Back to where this job was created",
-            "available": origin is not None,
-        },
-        "local": {
-            "description": "Save to local files only",
-            "available": True,
-        }
-    }
-    
-    for platform in connected:
-        home = config.get_home_channel(platform)
-        options[platform.value] = {
-            "description": f"{platform.value.title()} home channel",
-            "available": True,
-            "home_channel": home.to_dict() if home else None,
-        }
-    
-    return {
-        "origin": origin.to_dict() if origin else None,
-        "options": options,
-        "always_log_local": config.always_log_local,
-    }
+

gateway/platforms/base.py

@@ -27,7 +27,6 @@ sys.path.insert(0, str(_Path(__file__).resolve().parents[2]))
 
 from gateway.config import Platform, PlatformConfig
 from gateway.session import SessionSource, build_session_key
-from hermes_cli.config import get_hermes_home
 from hermes_constants import get_hermes_dir
 
 
@@ -485,6 +484,9 @@ class BasePlatformAdapter(ABC):
         self._background_tasks: set[asyncio.Task] = set()
         # Chats where auto-TTS on voice input is disabled (set by /voice off)
         self._auto_tts_disabled_chats: set = set()
+        # Chats where typing indicator is paused (e.g. during approval waits).
+        # _keep_typing skips send_typing when the chat_id is in this set.
+        self._typing_paused: set = set()
 
     @property
     def has_fatal_error(self) -> bool:
@@ -569,6 +571,16 @@ class BasePlatformAdapter(ABC):
         """
         self._message_handler = handler
     
+    def set_session_store(self, session_store: Any) -> None:
+        """
+        Set the session store for checking active sessions.
+        
+        Used by adapters that need to check if a thread/conversation
+        has an active session before processing messages (e.g., Slack
+        thread replies without explicit mentions).
+        """
+        self._session_store = session_store
+    
     @abstractmethod
     async def connect(self) -> bool:
         """
@@ -934,10 +946,16 @@ class BasePlatformAdapter(ABC):
         
         Telegram/Discord typing status expires after ~5 seconds, so we refresh every 2
         to recover quickly after progress messages interrupt it.
+        
+        Skips send_typing when the chat is in ``_typing_paused`` (e.g. while
+        the agent is waiting for dangerous-command approval).  This is critical
+        for Slack's Assistant API where ``assistant_threads_setStatus`` disables
+        the compose box — pausing lets the user type ``/approve`` or ``/deny``.
         """
         try:
             while True:
-                await self.send_typing(chat_id, metadata=metadata)
+                if chat_id not in self._typing_paused:
+                    await self.send_typing(chat_id, metadata=metadata)
                 await asyncio.sleep(interval)
         except asyncio.CancelledError:
             pass  # Normal cancellation when handler completes
@@ -951,7 +969,20 @@ class BasePlatformAdapter(ABC):
                     await self.stop_typing(chat_id)
                 except Exception:
                     pass
-    
+            self._typing_paused.discard(chat_id)
+
+    def pause_typing_for_chat(self, chat_id: str) -> None:
+        """Pause typing indicator for a chat (e.g. during approval waits).
+
+        Thread-safe (CPython GIL) — can be called from the sync agent thread
+        while ``_keep_typing`` runs on the async event loop.
+        """
+        self._typing_paused.add(chat_id)
+
+    def resume_typing_for_chat(self, chat_id: str) -> None:
+        """Resume typing indicator for a chat after approval resolves."""
+        self._typing_paused.discard(chat_id)
+
     # ── Processing lifecycle hooks ──────────────────────────────────────────
     # Subclasses override these to react to message processing events
     # (e.g. Discord adds 👀/✅/❌ reactions).
@@ -1093,16 +1124,20 @@ class BasePlatformAdapter(ABC):
         
         # Check if there's already an active handler for this session
         if session_key in self._active_sessions:
-            # /approve and /deny must bypass the active-session guard.
-            # The agent thread is blocked on threading.Event.wait() inside
-            # tools/approval.py — queuing these commands creates a deadlock:
-            # the agent waits for approval, approval waits for agent to finish.
-            # Dispatch directly to the message handler without touching session
-            # lifecycle (no competing background task, no session guard removal).
+            # Certain commands must bypass the active-session guard and be
+            # dispatched directly to the gateway runner.  Without this, they
+            # are queued as pending messages and either:
+            #   - leak into the conversation as user text (/stop, /new), or
+            #   - deadlock (/approve, /deny — agent is blocked on Event.wait)
+            #
+            # Dispatch inline: call the message handler directly and send the
+            # response.  Do NOT use _process_message_background — it manages
+            # session lifecycle and its cleanup races with the running task
+            # (see PR #4926).
             cmd = event.get_command()
-            if cmd in ("approve", "deny"):
+            if cmd in ("approve", "deny", "status", "stop", "new", "reset"):
                 logger.debug(
-                    "[%s] Approval command '/%s' bypassing active-session guard for %s",
+                    "[%s] Command '/%s' bypassing active-session guard for %s",
                     self.name, cmd, session_key,
                 )
                 try:
@@ -1116,29 +1151,7 @@ class BasePlatformAdapter(ABC):
                             metadata=_thread_meta,
                         )
                 except Exception as e:
-                    logger.error("[%s] Approval dispatch failed: %s", self.name, e, exc_info=True)
-                return
-
-            # /status must also bypass the active-session guard so it always
-            # returns a system-generated response instead of being queued as
-            # user text and passed to the agent (#5046).
-            if cmd == "status":
-                logger.debug(
-                    "[%s] Status command bypassing active-session guard for %s",
-                    self.name, session_key,
-                )
-                try:
-                    _thread_meta = {"thread_id": event.source.thread_id} if event.source.thread_id else None
-                    response = await self._message_handler(event)
-                    if response:
-                        await self._send_with_retry(
-                            chat_id=event.source.chat_id,
-                            content=response,
-                            reply_to=event.message_id,
-                            metadata=_thread_meta,
-                        )
-                except Exception as e:
-                    logger.error("[%s] Status dispatch failed: %s", self.name, e, exc_info=True)
+                    logger.error("[%s] Command '/%s' dispatch failed: %s", self.name, cmd, e, exc_info=True)
                 return
 
             # Special case: photo bursts/albums frequently arrive as multiple near-

gateway/platforms/discord.py

@@ -2039,6 +2039,66 @@ class DiscordAdapter(BasePlatformAdapter):
         except Exception as e:
             return SendResult(success=False, error=str(e))
 
+    async def send_model_picker(
+        self,
+        chat_id: str,
+        providers: list,
+        current_model: str,
+        current_provider: str,
+        session_key: str,
+        on_model_selected,
+        metadata: Optional[Dict[str, Any]] = None,
+    ) -> SendResult:
+        """Send an interactive select-menu model picker.
+
+        Two-step drill-down: provider dropdown → model dropdown.
+        Uses Discord embeds + Select menus via ``ModelPickerView``.
+        """
+        if not self._client or not DISCORD_AVAILABLE:
+            return SendResult(success=False, error="Not connected")
+
+        try:
+            # Resolve target channel (use thread_id if present)
+            target_id = chat_id
+            if metadata and metadata.get("thread_id"):
+                target_id = metadata["thread_id"]
+
+            channel = self._client.get_channel(int(target_id))
+            if not channel:
+                channel = await self._client.fetch_channel(int(target_id))
+
+            try:
+                from hermes_cli.providers import get_label
+                provider_label = get_label(current_provider)
+            except Exception:
+                provider_label = current_provider
+
+            embed = discord.Embed(
+                title="⚙ Model Configuration",
+                description=(
+                    f"Current model: `{current_model or 'unknown'}`\n"
+                    f"Provider: {provider_label}\n\n"
+                    f"Select a provider:"
+                ),
+                color=discord.Color.blue(),
+            )
+
+            view = ModelPickerView(
+                providers=providers,
+                current_model=current_model,
+                current_provider=current_provider,
+                session_key=session_key,
+                on_model_selected=on_model_selected,
+                allowed_user_ids=self._allowed_user_ids,
+            )
+
+            msg = await channel.send(embed=embed, view=view)
+            return SendResult(success=True, message_id=str(msg.id))
+
+        except Exception as e:
+            logger.warning("[%s] send_model_picker failed: %s", self.name, e)
+            return SendResult(success=False, error=str(e))
+
     def _get_parent_channel_id(self, channel: Any) -> Optional[str]:
         """Return the parent channel ID for a Discord thread-like channel, if present."""
         parent = getattr(channel, "parent", None)
@@ -2530,3 +2590,218 @@ if DISCORD_AVAILABLE:
             self.resolved = True
             for child in self.children:
                 child.disabled = True
+
+    class ModelPickerView(discord.ui.View):
+        """Interactive select-menu view for model switching.
+
+        Two-step drill-down: provider dropdown → model dropdown.
+        Edits the original message in-place as the user navigates.
+        Times out after 2 minutes.
+        """
+
+        def __init__(
+            self,
+            providers: list,
+            current_model: str,
+            current_provider: str,
+            session_key: str,
+            on_model_selected,
+            allowed_user_ids: set,
+        ):
+            super().__init__(timeout=120)
+            self.providers = providers
+            self.current_model = current_model
+            self.current_provider = current_provider
+            self.session_key = session_key
+            self.on_model_selected = on_model_selected
+            self.allowed_user_ids = allowed_user_ids
+            self.resolved = False
+            self._selected_provider: str = ""
+
+            self._build_provider_select()
+
+        def _check_auth(self, interaction: discord.Interaction) -> bool:
+            if not self.allowed_user_ids:
+                return True
+            return str(interaction.user.id) in self.allowed_user_ids
+
+        def _build_provider_select(self):
+            """Build the provider dropdown menu."""
+            self.clear_items()
+            options = []
+            for p in self.providers:
+                count = p.get("total_models", len(p.get("models", [])))
+                label = f"{p['name']} ({count} models)"
+                desc = "current" if p.get("is_current") else None
+                options.append(
+                    discord.SelectOption(
+                        label=label[:100],
+                        value=p["slug"],
+                        description=desc,
+                    )
+                )
+            if not options:
+                return
+
+            select = discord.ui.Select(
+                placeholder="Choose a provider...",
+                options=options[:25],
+                custom_id="model_provider_select",
+            )
+            select.callback = self._on_provider_selected
+            self.add_item(select)
+
+            cancel_btn = discord.ui.Button(
+                label="Cancel", style=discord.ButtonStyle.red, custom_id="model_cancel"
+            )
+            cancel_btn.callback = self._on_cancel
+            self.add_item(cancel_btn)
+
+        def _build_model_select(self, provider_slug: str):
+            """Build the model dropdown for a specific provider."""
+            self.clear_items()
+            provider = next(
+                (p for p in self.providers if p["slug"] == provider_slug), None
+            )
+            if not provider:
+                return
+
+            models = provider.get("models", [])
+            options = []
+            for model_id in models[:25]:
+                short = model_id.split("/")[-1] if "/" in model_id else model_id
+                options.append(
+                    discord.SelectOption(
+                        label=short[:100],
+                        value=model_id[:100],
+                    )
+                )
+            if not options:
+                return
+
+            select = discord.ui.Select(
+                placeholder=f"Choose a model from {provider.get('name', provider_slug)}...",
+                options=options,
+                custom_id="model_model_select",
+            )
+            select.callback = self._on_model_selected
+            self.add_item(select)
+
+            back_btn = discord.ui.Button(
+                label="◀ Back", style=discord.ButtonStyle.grey, custom_id="model_back"
+            )
+            back_btn.callback = self._on_back
+            self.add_item(back_btn)
+
+            cancel_btn = discord.ui.Button(
+                label="Cancel", style=discord.ButtonStyle.red, custom_id="model_cancel2"
+            )
+            cancel_btn.callback = self._on_cancel
+            self.add_item(cancel_btn)
+
+        async def _on_provider_selected(self, interaction: discord.Interaction):
+            if not self._check_auth(interaction):
+                await interaction.response.send_message(
+                    "You're not authorized~", ephemeral=True
+                )
+                return
+
+            provider_slug = interaction.data["values"][0]
+            self._selected_provider = provider_slug
+            provider = next(
+                (p for p in self.providers if p["slug"] == provider_slug), None
+            )
+            pname = provider.get("name", provider_slug) if provider else provider_slug
+
+            self._build_model_select(provider_slug)
+
+            total = provider.get("total_models", 0) if provider else 0
+            shown = min(len(provider.get("models", [])), 25) if provider else 0
+            extra = f"\n*{total - shown} more available — type `/model <name>` directly*" if total > shown else ""
+
+            await interaction.response.edit_message(
+                embed=discord.Embed(
+                    title="⚙ Model Configuration",
+                    description=f"Provider: **{pname}**\nSelect a model:{extra}",
+                    color=discord.Color.blue(),
+                ),
+                view=self,
+            )
+
+        async def _on_model_selected(self, interaction: discord.Interaction):
+            if self.resolved:
+                await interaction.response.send_message(
+                    "Already resolved~", ephemeral=True
+                )
+                return
+            if not self._check_auth(interaction):
+                await interaction.response.send_message(
+                    "You're not authorized~", ephemeral=True
+                )
+                return
+
+            self.resolved = True
+            model_id = interaction.data["values"][0]
+
+            try:
+                result_text = await self.on_model_selected(
+                    str(interaction.channel_id),
+                    model_id,
+                    self._selected_provider,
+                )
+            except Exception as exc:
+                result_text = f"Error switching model: {exc}"
+
+            self.clear_items()
+            await interaction.response.edit_message(
+                embed=discord.Embed(
+                    title="⚙ Model Switched",
+                    description=result_text,
+                    color=discord.Color.green(),
+                ),
+                view=self,
+            )
+
+        async def _on_back(self, interaction: discord.Interaction):
+            if not self._check_auth(interaction):
+                await interaction.response.send_message(
+                    "You're not authorized~", ephemeral=True
+                )
+                return
+
+            self._build_provider_select()
+
+            try:
+                from hermes_cli.providers import get_label
+                provider_label = get_label(self.current_provider)
+            except Exception:
+                provider_label = self.current_provider
+
+            await interaction.response.edit_message(
+                embed=discord.Embed(
+                    title="⚙ Model Configuration",
+                    description=(
+                        f"Current model: `{self.current_model or 'unknown'}`\n"
+                        f"Provider: {provider_label}\n\n"
+                        f"Select a provider:"
+                    ),
+                    color=discord.Color.blue(),
+                ),
+                view=self,
+            )
+
+        async def _on_cancel(self, interaction: discord.Interaction):
+            self.resolved = True
+            self.clear_items()
+            await interaction.response.edit_message(
+                embed=discord.Embed(
+                    title="⚙ Model Configuration",
+                    description="Model selection cancelled.",
+                    color=discord.Color.greyple(),
+                ),
+                view=self,
+            )
+
+        async def on_timeout(self):
+            self.resolved = True
+            self.clear_items()

gateway/platforms/feishu.py

@@ -60,7 +60,6 @@ try:
         CreateMessageRequestBody,
         GetChatRequest,
         GetMessageRequest,
-        GetImageRequest,
         GetMessageResourceRequest,
         P2ImMessageMessageReadV1,
         ReplyMessageRequest,
@@ -388,10 +387,6 @@ def _coerce_required_int(value: Any, default: int, min_value: int = 0) -> int:
     return default if parsed is None else parsed
 
 
-def _is_loop_ready(loop: Optional[asyncio.AbstractEventLoop]) -> bool:
-    return loop is not None and not bool(getattr(loop, "is_closed", lambda: False)())
-
-
 # ---------------------------------------------------------------------------
 # Post payload builders and parsers
 # ---------------------------------------------------------------------------

gateway/platforms/matrix.py

@@ -1057,7 +1057,7 @@ class MatrixAdapter(BasePlatformAdapter):
 
         # Message type.
         msg_type = MessageType.TEXT
-        if body.startswith("!") or body.startswith("/"):
+        if body.startswith(("!", "/")):
             msg_type = MessageType.COMMAND
 
         source = self.build_source(

gateway/platforms/mattermost.py

@@ -430,7 +430,6 @@ class MattermostAdapter(BasePlatformAdapter):
                     ct = resp.content_type or "application/octet-stream"
                     break
             except (aiohttp.ClientError, asyncio.TimeoutError) as exc:
-                last_exc = exc
                 if attempt < 2:
                     await asyncio.sleep(1.5 * (attempt + 1))
                     continue

gateway/platforms/slack.py

@@ -84,6 +84,17 @@ class SlackAdapter(BasePlatformAdapter):
         self._seen_messages: Dict[str, float] = {}
         self._SEEN_TTL = 300   # 5 minutes
         self._SEEN_MAX = 2000  # prune threshold
+        # Track pending approval message_ts → resolved flag to prevent
+        # double-clicks on approval buttons.
+        self._approval_resolved: Dict[str, bool] = {}
+        # Track timestamps of messages sent by the bot so we can respond
+        # to thread replies even without an explicit @mention.
+        self._bot_message_ts: set = set()
+        self._BOT_TS_MAX = 5000  # cap to avoid unbounded growth
+        # Track threads where the bot has been @mentioned — once mentioned,
+        # respond to ALL subsequent messages in that thread automatically.
+        self._mentioned_threads: set = set()
+        self._MENTIONED_THREADS_MAX = 5000
 
     async def connect(self) -> bool:
         """Connect to Slack via Socket Mode."""
@@ -176,6 +187,15 @@ class SlackAdapter(BasePlatformAdapter):
                 await ack()
                 await self._handle_slash_command(command)
 
+            # Register Block Kit action handlers for approval buttons
+            for _action_id in (
+                "hermes_approve_once",
+                "hermes_approve_session",
+                "hermes_approve_always",
+                "hermes_deny",
+            ):
+                self._app.action(_action_id)(self._handle_approval_action)
+
             # Start Socket Mode handler in background
             self._handler = AsyncSocketModeHandler(self._app, app_token)
             self._socket_mode_task = asyncio.create_task(self._handler.start_async())
@@ -256,9 +276,22 @@ class SlackAdapter(BasePlatformAdapter):
 
                 last_result = await self._get_client(chat_id).chat_postMessage(**kwargs)
 
+            # Track the sent message ts so we can auto-respond to thread
+            # replies without requiring @mention.
+            sent_ts = last_result.get("ts") if last_result else None
+            if sent_ts:
+                self._bot_message_ts.add(sent_ts)
+                # Also register the thread root so replies-to-my-replies work
+                if thread_ts:
+                    self._bot_message_ts.add(thread_ts)
+                if len(self._bot_message_ts) > self._BOT_TS_MAX:
+                    excess = len(self._bot_message_ts) - self._BOT_TS_MAX // 2
+                    for old_ts in list(self._bot_message_ts)[:excess]:
+                        self._bot_message_ts.discard(old_ts)
+
             return SendResult(
                 success=True,
-                message_id=last_result.get("ts") if last_result else None,
+                message_id=sent_ts,
                 raw_response=last_result,
             )
 
@@ -276,10 +309,13 @@ class SlackAdapter(BasePlatformAdapter):
         if not self._app:
             return SendResult(success=False, error="Not connected")
         try:
+            # Convert standard markdown → Slack mrkdwn
+            formatted = self.format_message(content)
+
             await self._get_client(chat_id).chat_update(
                 channel=chat_id,
                 ts=message_id,
-                text=content,
+                text=formatted,
             )
             return SendResult(success=True, message_id=message_id)
         except Exception as e:  # pragma: no cover - defensive logging
@@ -763,13 +799,61 @@ class SlackAdapter(BasePlatformAdapter):
         else:
             thread_ts = event.get("thread_ts") or ts  # ts fallback for channels
 
-        # In channels, only respond if bot is mentioned
+        # In channels, respond if:
+        #   1. The bot is @mentioned in this message, OR
+        #   2. The message is a reply in a thread the bot started/participated in, OR
+        #   3. The message is in a thread where the bot was previously @mentioned, OR
+        #   4. There's an existing session for this thread (survives restarts)
         bot_uid = self._team_bot_user_ids.get(team_id, self._bot_user_id)
-        if not is_dm and bot_uid:
-            if f"<@{bot_uid}>" not in text:
+        is_mentioned = bot_uid and f"<@{bot_uid}>" in text
+        event_thread_ts = event.get("thread_ts")
+        is_thread_reply = bool(event_thread_ts and event_thread_ts != ts)
+
+        if not is_dm and bot_uid and not is_mentioned:
+            reply_to_bot_thread = (
+                is_thread_reply and event_thread_ts in self._bot_message_ts
+            )
+            in_mentioned_thread = (
+                event_thread_ts is not None
+                and event_thread_ts in self._mentioned_threads
+            )
+            has_session = (
+                is_thread_reply
+                and self._has_active_session_for_thread(
+                    channel_id=channel_id,
+                    thread_ts=event_thread_ts,
+                    user_id=user_id,
+                )
+            )
+            if not reply_to_bot_thread and not in_mentioned_thread and not has_session:
                 return
+
+        if is_mentioned:
             # Strip the bot mention from the text
             text = text.replace(f"<@{bot_uid}>", "").strip()
+            # Register this thread so all future messages auto-trigger the bot
+            if event_thread_ts:
+                self._mentioned_threads.add(event_thread_ts)
+                if len(self._mentioned_threads) > self._MENTIONED_THREADS_MAX:
+                    to_remove = list(self._mentioned_threads)[:self._MENTIONED_THREADS_MAX // 2]
+                    for t in to_remove:
+                        self._mentioned_threads.discard(t)
+
+        # When entering a thread for the first time (no existing session),
+        # fetch thread context so the agent understands the conversation.
+        if is_thread_reply and not self._has_active_session_for_thread(
+            channel_id=channel_id,
+            thread_ts=event_thread_ts,
+            user_id=user_id,
+        ):
+            thread_context = await self._fetch_thread_context(
+                channel_id=channel_id,
+                thread_ts=event_thread_ts,
+                current_ts=ts,
+                team_id=team_id,
+            )
+            if thread_context:
+                text = thread_context + text
 
         # Determine message type
         msg_type = MessageType.TEXT
@@ -892,6 +976,233 @@ class SlackAdapter(BasePlatformAdapter):
         await self._remove_reaction(channel_id, ts, "eyes")
         await self._add_reaction(channel_id, ts, "white_check_mark")
 
+    # ----- Approval button support (Block Kit) -----
+
+    async def send_exec_approval(
+        self, chat_id: str, command: str, session_key: str,
+        description: str = "dangerous command",
+        metadata: Optional[Dict[str, Any]] = None,
+    ) -> SendResult:
+        """Send a Block Kit approval prompt with interactive buttons.
+
+        The buttons call ``resolve_gateway_approval()`` to unblock the waiting
+        agent thread — same mechanism as the text ``/approve`` flow.
+        """
+        if not self._app:
+            return SendResult(success=False, error="Not connected")
+
+        try:
+            cmd_preview = command[:2900] + "..." if len(command) > 2900 else command
+            thread_ts = self._resolve_thread_ts(None, metadata)
+
+            blocks = [
+                {
+                    "type": "section",
+                    "text": {
+                        "type": "mrkdwn",
+                        "text": (
+                            f":warning: *Command Approval Required*\n"
+                            f"```{cmd_preview}```\n"
+                            f"Reason: {description}"
+                        ),
+                    },
+                },
+                {
+                    "type": "actions",
+                    "elements": [
+                        {
+                            "type": "button",
+                            "text": {"type": "plain_text", "text": "Allow Once"},
+                            "style": "primary",
+                            "action_id": "hermes_approve_once",
+                            "value": session_key,
+                        },
+                        {
+                            "type": "button",
+                            "text": {"type": "plain_text", "text": "Allow Session"},
+                            "action_id": "hermes_approve_session",
+                            "value": session_key,
+                        },
+                        {
+                            "type": "button",
+                            "text": {"type": "plain_text", "text": "Always Allow"},
+                            "action_id": "hermes_approve_always",
+                            "value": session_key,
+                        },
+                        {
+                            "type": "button",
+                            "text": {"type": "plain_text", "text": "Deny"},
+                            "style": "danger",
+                            "action_id": "hermes_deny",
+                            "value": session_key,
+                        },
+                    ],
+                },
+            ]
+
+            kwargs: Dict[str, Any] = {
+                "channel": chat_id,
+                "text": f"⚠️ Command approval required: {cmd_preview[:100]}",
+                "blocks": blocks,
+            }
+            if thread_ts:
+                kwargs["thread_ts"] = thread_ts
+
+            result = await self._get_client(chat_id).chat_postMessage(**kwargs)
+            msg_ts = result.get("ts", "")
+            if msg_ts:
+                self._approval_resolved[msg_ts] = False
+
+            return SendResult(success=True, message_id=msg_ts, raw_response=result)
+        except Exception as e:
+            logger.error("[Slack] send_exec_approval failed: %s", e, exc_info=True)
+            return SendResult(success=False, error=str(e))
+
+    async def _handle_approval_action(self, ack, body, action) -> None:
+        """Handle an approval button click from Block Kit."""
+        await ack()
+
+        action_id = action.get("action_id", "")
+        session_key = action.get("value", "")
+        message = body.get("message", {})
+        msg_ts = message.get("ts", "")
+        channel_id = body.get("channel", {}).get("id", "")
+        user_name = body.get("user", {}).get("name", "unknown")
+
+        # Map action_id to approval choice
+        choice_map = {
+            "hermes_approve_once": "once",
+            "hermes_approve_session": "session",
+            "hermes_approve_always": "always",
+            "hermes_deny": "deny",
+        }
+        choice = choice_map.get(action_id, "deny")
+
+        # Prevent double-clicks
+        if self._approval_resolved.get(msg_ts, False):
+            return
+        self._approval_resolved[msg_ts] = True
+
+        # Update the message to show the decision and remove buttons
+        label_map = {
+            "once": f"✅ Approved once by {user_name}",
+            "session": f"✅ Approved for session by {user_name}",
+            "always": f"✅ Approved permanently by {user_name}",
+            "deny": f"❌ Denied by {user_name}",
+        }
+        decision_text = label_map.get(choice, f"Resolved by {user_name}")
+
+        # Get original text from the section block
+        original_text = ""
+        for block in message.get("blocks", []):
+            if block.get("type") == "section":
+                original_text = block.get("text", {}).get("text", "")
+                break
+
+        updated_blocks = [
+            {
+                "type": "section",
+                "text": {
+                    "type": "mrkdwn",
+                    "text": original_text or "Command approval request",
+                },
+            },
+            {
+                "type": "context",
+                "elements": [
+                    {"type": "mrkdwn", "text": decision_text},
+                ],
+            },
+        ]
+
+        try:
+            await self._get_client(channel_id).chat_update(
+                channel=channel_id,
+                ts=msg_ts,
+                text=decision_text,
+                blocks=updated_blocks,
+            )
+        except Exception as e:
+            logger.warning("[Slack] Failed to update approval message: %s", e)
+
+        # Resolve the approval — this unblocks the agent thread
+        try:
+            from tools.approval import resolve_gateway_approval
+            count = resolve_gateway_approval(session_key, choice)
+            logger.info(
+                "Slack button resolved %d approval(s) for session %s (choice=%s, user=%s)",
+                count, session_key, choice, user_name,
+            )
+        except Exception as exc:
+            logger.error("Failed to resolve gateway approval from Slack button: %s", exc)
+
+        # Clean up stale approval state
+        self._approval_resolved.pop(msg_ts, None)
+
+    # ----- Thread context fetching -----
+
+    async def _fetch_thread_context(
+        self, channel_id: str, thread_ts: str, current_ts: str,
+        team_id: str = "", limit: int = 30,
+    ) -> str:
+        """Fetch recent thread messages to provide context when the bot is
+        mentioned mid-thread for the first time.
+
+        Returns a formatted string with thread history, or empty string on
+        failure or if the thread is empty (just the parent message).
+        """
+        try:
+            client = self._get_client(channel_id)
+            result = await client.conversations_replies(
+                channel=channel_id,
+                ts=thread_ts,
+                limit=limit + 1,  # +1 because it includes the current message
+                inclusive=True,
+            )
+            messages = result.get("messages", [])
+            if not messages:
+                return ""
+
+            context_parts = []
+            for msg in messages:
+                msg_ts = msg.get("ts", "")
+                # Skip the current message (the one that triggered this fetch)
+                if msg_ts == current_ts:
+                    continue
+                # Skip bot messages from ourselves
+                if msg.get("bot_id") or msg.get("subtype") == "bot_message":
+                    continue
+
+                msg_user = msg.get("user", "unknown")
+                msg_text = msg.get("text", "").strip()
+                if not msg_text:
+                    continue
+
+                # Strip bot mentions from context messages
+                bot_uid = self._team_bot_user_ids.get(team_id, self._bot_user_id)
+                if bot_uid:
+                    msg_text = msg_text.replace(f"<@{bot_uid}>", "").strip()
+
+                # Mark the thread parent
+                is_parent = msg_ts == thread_ts
+                prefix = "[thread parent] " if is_parent else ""
+
+                # Resolve user name (cached)
+                name = await self._resolve_user_name(msg_user, chat_id=channel_id)
+                context_parts.append(f"{prefix}{name}: {msg_text}")
+
+            if not context_parts:
+                return ""
+
+            return (
+                "[Thread context — previous messages in this thread:]\n"
+                + "\n".join(context_parts)
+                + "\n[End of thread context]\n\n"
+            )
+        except Exception as e:
+            logger.warning("[Slack] Failed to fetch thread context: %s", e)
+            return ""
+
     async def _handle_slash_command(self, command: dict) -> None:
         """Handle /hermes slash command."""
         text = command.get("text", "").strip()
@@ -933,6 +1244,53 @@ class SlackAdapter(BasePlatformAdapter):
 
         await self.handle_message(event)
 
+    def _has_active_session_for_thread(
+        self,
+        channel_id: str,
+        thread_ts: str,
+        user_id: str,
+    ) -> bool:
+        """Check if there's an active session for a thread.
+
+        Used to determine if thread replies without @mentions should be
+        processed (they should if there's an active session).
+
+        Uses ``build_session_key()`` as the single source of truth for key
+        construction — avoids the bug where manual key building didn't
+        respect ``thread_sessions_per_user`` and ``group_sessions_per_user``
+        settings correctly.
+        """
+        session_store = getattr(self, "_session_store", None)
+        if not session_store:
+            return False
+
+        try:
+            from gateway.session import SessionSource, build_session_key
+
+            source = SessionSource(
+                platform=Platform.SLACK,
+                chat_id=channel_id,
+                chat_type="group",
+                user_id=user_id,
+                thread_id=thread_ts,
+            )
+
+            # Read session isolation settings from the store's config
+            store_cfg = getattr(session_store, "config", None)
+            gspu = getattr(store_cfg, "group_sessions_per_user", True) if store_cfg else True
+            tspu = getattr(store_cfg, "thread_sessions_per_user", False) if store_cfg else False
+
+            session_key = build_session_key(
+                source,
+                group_sessions_per_user=gspu,
+                thread_sessions_per_user=tspu,
+            )
+
+            session_store._ensure_loaded()
+            return session_key in session_store._entries
+        except Exception:
+            return False
+
     async def _download_slack_file(self, url: str, ext: str, audio: bool = False, team_id: str = "") -> str:
         """Download a Slack file using the bot token for auth, with retry."""
         import asyncio

gateway/platforms/telegram.py

@@ -151,6 +151,10 @@ class TelegramAdapter(BasePlatformAdapter):
         self._dm_topics: Dict[str, int] = {}
         # DM Topics config from extra.dm_topics
         self._dm_topics_config: List[Dict[str, Any]] = self.config.extra.get("dm_topics", [])
+        # Interactive model picker state per chat
+        self._model_picker_state: Dict[str, dict] = {}
+        # Approval button state: message_id → session_key
+        self._approval_state: Dict[int, str] = {}
 
     def _fallback_ips(self) -> list[str]:
         """Return validated fallback IPs from config (populated by _apply_env_overrides)."""
@@ -1008,14 +1012,432 @@ class TelegramAdapter(BasePlatformAdapter):
             logger.warning("[%s] send_update_prompt failed: %s", self.name, e)
             return SendResult(success=False, error=str(e))
 
+    async def send_exec_approval(
+        self, chat_id: str, command: str, session_key: str,
+        description: str = "dangerous command",
+        metadata: Optional[Dict[str, Any]] = None,
+    ) -> SendResult:
+        """Send an inline-keyboard approval prompt with interactive buttons.
+
+        The buttons call ``resolve_gateway_approval()`` to unblock the waiting
+        agent thread — same mechanism as the text ``/approve`` flow.
+        """
+        if not self._bot:
+            return SendResult(success=False, error="Not connected")
+
+        try:
+            cmd_preview = command[:3800] + "..." if len(command) > 3800 else command
+            text = (
+                f"⚠️ *Command Approval Required*\n\n"
+                f"`{cmd_preview}`\n\n"
+                f"Reason: {description}"
+            )
+
+            # Resolve thread context for thread replies
+            thread_id = None
+            if metadata:
+                thread_id = metadata.get("thread_id") or metadata.get("message_thread_id")
+
+            # We'll use the message_id as part of callback_data to look up session_key
+            # Send a placeholder first, then update — or use a counter.
+            # Simpler: use a monotonic counter to generate short IDs.
+            import itertools
+            if not hasattr(self, "_approval_counter"):
+                self._approval_counter = itertools.count(1)
+            approval_id = next(self._approval_counter)
+
+            keyboard = InlineKeyboardMarkup([
+                [
+                    InlineKeyboardButton("✅ Allow Once", callback_data=f"ea:once:{approval_id}"),
+                    InlineKeyboardButton("✅ Session", callback_data=f"ea:session:{approval_id}"),
+                ],
+                [
+                    InlineKeyboardButton("✅ Always", callback_data=f"ea:always:{approval_id}"),
+                    InlineKeyboardButton("❌ Deny", callback_data=f"ea:deny:{approval_id}"),
+                ],
+            ])
+
+            kwargs: Dict[str, Any] = {
+                "chat_id": int(chat_id),
+                "text": text,
+                "parse_mode": ParseMode.MARKDOWN,
+                "reply_markup": keyboard,
+            }
+            if thread_id:
+                kwargs["message_thread_id"] = int(thread_id)
+
+            msg = await self._bot.send_message(**kwargs)
+
+            # Store session_key keyed by approval_id for the callback handler
+            self._approval_state[approval_id] = session_key
+
+            return SendResult(success=True, message_id=str(msg.message_id))
+        except Exception as e:
+            logger.warning("[%s] send_exec_approval failed: %s", self.name, e)
+            return SendResult(success=False, error=str(e))
+
+    async def send_model_picker(
+        self,
+        chat_id: str,
+        providers: list,
+        current_model: str,
+        current_provider: str,
+        session_key: str,
+        on_model_selected,
+        metadata: Optional[Dict[str, Any]] = None,
+    ) -> SendResult:
+        """Send an interactive inline-keyboard model picker.
+
+        Two-step drill-down: provider selection → model selection.
+        Edits the same message in-place as the user navigates.
+        """
+        if not self._bot:
+            return SendResult(success=False, error="Not connected")
+
+        try:
+            from hermes_cli.providers import get_label
+        except ImportError:
+            def get_label(slug):
+                return slug
+
+        try:
+            # Build provider buttons — 2 per row
+            buttons: list = []
+            for p in providers:
+                count = p.get("total_models", len(p.get("models", [])))
+                label = f"{p['name']} ({count})"
+                if p.get("is_current"):
+                    label = f"✓ {label}"
+                # Compact callback data: mp:<slug>  (max 64 bytes)
+                buttons.append(
+                    InlineKeyboardButton(label, callback_data=f"mp:{p['slug']}")
+                )
+
+            rows = [buttons[i : i + 2] for i in range(0, len(buttons), 2)]
+            rows.append([InlineKeyboardButton("✗ Cancel", callback_data="mx")])
+            keyboard = InlineKeyboardMarkup(rows)
+
+            provider_label = get_label(current_provider)
+            text = (
+                f"⚙ *Model Configuration*\n\n"
+                f"Current model: `{current_model or 'unknown'}`\n"
+                f"Provider: {provider_label}\n\n"
+                f"Select a provider:"
+            )
+
+            thread_id = metadata.get("thread_id") if metadata else None
+            msg = await self._bot.send_message(
+                chat_id=int(chat_id),
+                text=text,
+                parse_mode=ParseMode.MARKDOWN,
+                reply_markup=keyboard,
+                message_thread_id=int(thread_id) if thread_id else None,
+            )
+
+            # Store picker state keyed by chat_id
+            self._model_picker_state[str(chat_id)] = {
+                "msg_id": msg.message_id,
+                "providers": providers,
+                "session_key": session_key,
+                "on_model_selected": on_model_selected,
+                "current_model": current_model,
+                "current_provider": current_provider,
+            }
+
+            return SendResult(success=True, message_id=str(msg.message_id))
+        except Exception as e:
+            logger.warning("[%s] send_model_picker failed: %s", self.name, e)
+            return SendResult(success=False, error=str(e))
+
+    _MODEL_PAGE_SIZE = 8
+
+    def _build_model_keyboard(self, models: list, page: int) -> tuple:
+        """Build paginated model buttons. Returns (keyboard, page_info_text)."""
+        page_size = self._MODEL_PAGE_SIZE
+        total = len(models)
+        total_pages = max(1, (total + page_size - 1) // page_size)
+        page = max(0, min(page, total_pages - 1))
+
+        start = page * page_size
+        end = min(start + page_size, total)
+        page_models = models[start:end]
+
+        buttons: list = []
+        for i, model_id in enumerate(page_models):
+            abs_idx = start + i
+            short = model_id.split("/")[-1] if "/" in model_id else model_id
+            if len(short) > 38:
+                short = short[:35] + "..."
+            buttons.append(
+                InlineKeyboardButton(short, callback_data=f"mm:{abs_idx}")
+            )
+
+        rows = [buttons[i : i + 2] for i in range(0, len(buttons), 2)]
+
+        # Pagination row (if needed)
+        if total_pages > 1:
+            nav: list = []
+            if page > 0:
+                nav.append(InlineKeyboardButton("◀ Prev", callback_data=f"mg:{page - 1}"))
+            nav.append(InlineKeyboardButton(f"{page + 1}/{total_pages}", callback_data="mx:noop"))
+            if page < total_pages - 1:
+                nav.append(InlineKeyboardButton("Next ▶", callback_data=f"mg:{page + 1}"))
+            rows.append(nav)
+
+        rows.append([
+            InlineKeyboardButton("◀ Back", callback_data="mb"),
+            InlineKeyboardButton("✗ Cancel", callback_data="mx"),
+        ])
+
+        page_info = f" ({start + 1}–{end} of {total})" if total_pages > 1 else ""
+        return InlineKeyboardMarkup(rows), page_info
+
+    async def _handle_model_picker_callback(
+        self, query, data: str, chat_id: str
+    ) -> None:
+        """Handle model picker inline keyboard callbacks (mp:/mm:/mb:/mx:/mg:)."""
+        state = self._model_picker_state.get(chat_id)
+        if not state:
+            await query.answer(text="Picker expired — use /model again.")
+            return
+
+        try:
+            from hermes_cli.providers import get_label
+        except ImportError:
+            def get_label(slug):
+                return slug
+
+        if data.startswith("mp:"):
+            # --- Provider selected: show model buttons (page 0) ---
+            provider_slug = data[3:]
+            provider = next(
+                (p for p in state["providers"] if p["slug"] == provider_slug),
+                None,
+            )
+            if not provider:
+                await query.answer(text="Provider not found.")
+                return
+
+            models = provider.get("models", [])
+            state["selected_provider"] = provider_slug
+            state["selected_provider_name"] = provider.get("name", provider_slug)
+            state["model_list"] = models
+            state["model_page"] = 0
+
+            keyboard, page_info = self._build_model_keyboard(models, 0)
+
+            pname = provider.get("name", provider_slug)
+            total = provider.get("total_models", len(models))
+            shown = len(models)
+            extra = f"\n_{total - shown} more available — type `/model <name>` directly_" if total > shown else ""
+
+            await query.edit_message_text(
+                text=(
+                    f"⚙ *Model Configuration*\n\n"
+                    f"Provider: *{pname}*{page_info}\n"
+                    f"Select a model:{extra}"
+                ),
+                parse_mode=ParseMode.MARKDOWN,
+                reply_markup=keyboard,
+            )
+            await query.answer()
+
+        elif data.startswith("mg:"):
+            # --- Page navigation ---
+            try:
+                page = int(data[3:])
+            except ValueError:
+                await query.answer(text="Invalid page.")
+                return
+
+            models = state.get("model_list", [])
+            state["model_page"] = page
+
+            keyboard, page_info = self._build_model_keyboard(models, page)
+
+            pname = state.get("selected_provider_name", "")
+            provider_slug = state.get("selected_provider", "")
+            provider = next(
+                (p for p in state["providers"] if p["slug"] == provider_slug),
+                None,
+            )
+            total = provider.get("total_models", len(models)) if provider else len(models)
+            shown = len(models)
+            extra = f"\n_{total - shown} more available — type `/model <name>` directly_" if total > shown else ""
+
+            await query.edit_message_text(
+                text=(
+                    f"⚙ *Model Configuration*\n\n"
+                    f"Provider: *{pname}*{page_info}\n"
+                    f"Select a model:{extra}"
+                ),
+                parse_mode=ParseMode.MARKDOWN,
+                reply_markup=keyboard,
+            )
+            await query.answer()
+
+        elif data.startswith("mm:"):
+            # --- Model selected: perform the switch ---
+            try:
+                idx = int(data[3:])
+            except ValueError:
+                await query.answer(text="Invalid selection.")
+                return
+
+            model_list = state.get("model_list", [])
+            if idx < 0 or idx >= len(model_list):
+                await query.answer(text="Invalid model index.")
+                return
+
+            model_id = model_list[idx]
+            provider_slug = state.get("selected_provider", "")
+            callback = state.get("on_model_selected")
+
+            if not callback:
+                await query.answer(text="Picker expired.")
+                return
+
+            try:
+                result_text = await callback(chat_id, model_id, provider_slug)
+            except Exception as exc:
+                logger.error("Model picker switch failed: %s", exc)
+                result_text = f"Error switching model: {exc}"
+
+            # Edit message to show confirmation, remove buttons
+            try:
+                await query.edit_message_text(
+                    text=result_text,
+                    parse_mode=ParseMode.MARKDOWN,
+                    reply_markup=None,
+                )
+            except Exception:
+                # Markdown parse failure — retry as plain text
+                try:
+                    await query.edit_message_text(
+                        text=result_text,
+                        parse_mode=None,
+                        reply_markup=None,
+                    )
+                except Exception:
+                    pass
+            await query.answer(text="Model switched!")
+
+            # Clean up state
+            self._model_picker_state.pop(chat_id, None)
+
+        elif data == "mb":
+            # --- Back to provider list ---
+            buttons = []
+            for p in state["providers"]:
+                count = p.get("total_models", len(p.get("models", [])))
+                label = f"{p['name']} ({count})"
+                if p.get("is_current"):
+                    label = f"✓ {label}"
+                buttons.append(
+                    InlineKeyboardButton(label, callback_data=f"mp:{p['slug']}")
+                )
+
+            rows = [buttons[i : i + 2] for i in range(0, len(buttons), 2)]
+            rows.append([InlineKeyboardButton("✗ Cancel", callback_data="mx")])
+            keyboard = InlineKeyboardMarkup(rows)
+
+            try:
+                provider_label = get_label(state["current_provider"])
+            except Exception:
+                provider_label = state["current_provider"]
+
+            await query.edit_message_text(
+                text=(
+                    f"⚙ *Model Configuration*\n\n"
+                    f"Current model: `{state['current_model'] or 'unknown'}`\n"
+                    f"Provider: {provider_label}\n\n"
+                    f"Select a provider:"
+                ),
+                parse_mode=ParseMode.MARKDOWN,
+                reply_markup=keyboard,
+            )
+            await query.answer()
+
+        elif data == "mx":
+            # --- Cancel ---
+            self._model_picker_state.pop(chat_id, None)
+            await query.edit_message_text(
+                text="Model selection cancelled.",
+                reply_markup=None,
+            )
+            await query.answer()
+
+        else:
+            # Catch-all (e.g. page counter button "mx:noop")
+            await query.answer()
+
     async def _handle_callback_query(
         self, update: "Update", context: "ContextTypes.DEFAULT_TYPE"
     ) -> None:
-        """Handle inline keyboard button clicks (update prompts)."""
+        """Handle inline keyboard button clicks."""
         query = update.callback_query
         if not query or not query.data:
             return
         data = query.data
+
+        # --- Model picker callbacks ---
+        if data.startswith(("mp:", "mm:", "mb", "mx", "mg:")):
+            chat_id = str(query.message.chat_id) if query.message else None
+            if chat_id:
+                await self._handle_model_picker_callback(query, data, chat_id)
+            return
+
+        # --- Exec approval callbacks (ea:choice:id) ---
+        if data.startswith("ea:"):
+            parts = data.split(":", 2)
+            if len(parts) == 3:
+                choice = parts[1]  # once, session, always, deny
+                try:
+                    approval_id = int(parts[2])
+                except (ValueError, IndexError):
+                    await query.answer(text="Invalid approval data.")
+                    return
+
+                session_key = self._approval_state.pop(approval_id, None)
+                if not session_key:
+                    await query.answer(text="This approval has already been resolved.")
+                    return
+
+                # Map choice to human-readable label
+                label_map = {
+                    "once": "✅ Approved once",
+                    "session": "✅ Approved for session",
+                    "always": "✅ Approved permanently",
+                    "deny": "❌ Denied",
+                }
+                user_display = getattr(query.from_user, "first_name", "User")
+                label = label_map.get(choice, "Resolved")
+
+                await query.answer(text=label)
+
+                # Edit message to show decision, remove buttons
+                try:
+                    await query.edit_message_text(
+                        text=f"{label} by {user_display}",
+                        parse_mode=ParseMode.MARKDOWN,
+                        reply_markup=None,
+                    )
+                except Exception:
+                    pass  # non-fatal if edit fails
+
+                # Resolve the approval — unblocks the agent thread
+                try:
+                    from tools.approval import resolve_gateway_approval
+                    count = resolve_gateway_approval(session_key, choice)
+                    logger.info(
+                        "Telegram button resolved %d approval(s) for session %s (choice=%s, user=%s)",
+                        count, session_key, choice, user_display,
+                    )
+                except Exception as exc:
+                    logger.error("Failed to resolve gateway approval from Telegram button: %s", exc)
+            return
+
+        # --- Update prompt callbacks ---
         if not data.startswith("update_prompt:"):
             return
         answer = data.split(":", 1)[1]  # "y" or "n"
@@ -1063,7 +1485,7 @@ class TelegramAdapter(BasePlatformAdapter):
             
             with open(audio_path, "rb") as audio_file:
                 # .ogg files -> send as voice (round playable bubble)
-                if audio_path.endswith(".ogg") or audio_path.endswith(".opus"):
+                if audio_path.endswith((".ogg", ".opus")):
                     _voice_thread = metadata.get("thread_id") if metadata else None
                     msg = await self._bot.send_voice(
                         chat_id=int(chat_id),

gateway/platforms/webhook.py

@@ -203,10 +203,8 @@ class WebhookAdapter(BasePlatformAdapter):
 
     def _reload_dynamic_routes(self) -> None:
         """Reload agent-created subscriptions from disk if the file changed."""
-        from pathlib import Path as _Path
-        hermes_home = _Path(
-            os.getenv("HERMES_HOME", str(_Path.home() / ".hermes"))
-        ).expanduser()
+        from hermes_constants import get_hermes_home
+        hermes_home = get_hermes_home()
         subs_path = hermes_home / _DYNAMIC_ROUTES_FILENAME
         if not subs_path.exists():
             if self._dynamic_routes:

gateway/platforms/wecom.py

@@ -653,7 +653,7 @@ class WeComAdapter(BasePlatformAdapter):
             return ".png"
         if data.startswith(b"\xff\xd8\xff"):
             return ".jpg"
-        if data.startswith(b"GIF87a") or data.startswith(b"GIF89a"):
+        if data.startswith((b"GIF87a", b"GIF89a")):
             return ".gif"
         if data.startswith(b"RIFF") and data[8:12] == b"WEBP":
             return ".webp"
@@ -689,7 +689,7 @@ class WeComAdapter(BasePlatformAdapter):
     @staticmethod
     def _derive_message_type(body: Dict[str, Any], text: str, media_types: List[str]) -> MessageType:
         """Choose the normalized inbound message type."""
-        if any(mtype.startswith("application/") or mtype.startswith("text/") for mtype in media_types):
+        if any(mtype.startswith(("application/", "text/")) for mtype in media_types):
             return MessageType.DOCUMENT
         if any(mtype.startswith("image/") for mtype in media_types):
             return MessageType.TEXT if text else MessageType.PHOTO

gateway/platforms/whatsapp.py

@@ -27,7 +27,6 @@ _IS_WINDOWS = platform.system() == "Windows"
 from pathlib import Path
 from typing import Dict, Optional, Any
 
-from hermes_cli.config import get_hermes_home
 from hermes_constants import get_hermes_dir
 
 logger = logging.getLogger(__name__)

gateway/run.py

@@ -24,7 +24,6 @@ import signal
 import tempfile
 import threading
 import time
-import uuid
 from pathlib import Path
 from datetime import datetime
 from typing import Dict, Optional, Any, List
@@ -378,7 +377,7 @@ def _check_unavailable_skill(command_name: str) -> str | None:
                     )
 
         # Check optional skills (shipped with repo but not installed)
-        from hermes_constants import get_hermes_home, get_optional_skills_dir
+        from hermes_constants import get_optional_skills_dir
         repo_root = Path(__file__).resolve().parent.parent
         optional_dir = get_optional_skills_dir(repo_root / "optional-skills")
         if optional_dir.exists():
@@ -1127,6 +1126,7 @@ class GatewayRunner:
             # Set up message + fatal error handlers
             adapter.set_message_handler(self._handle_message)
             adapter.set_fatal_error_handler(self._handle_adapter_fatal_error)
+            adapter.set_session_store(self.session_store)
             
             # Try to connect
             logger.info("Connecting to %s...", platform.value)
@@ -1424,6 +1424,7 @@ class GatewayRunner:
 
                     adapter.set_message_handler(self._handle_message)
                     adapter.set_fatal_error_handler(self._handle_adapter_fatal_error)
+                    adapter.set_session_store(self.session_store)
 
                     success = await adapter.connect()
                     if success:
@@ -1856,6 +1857,11 @@ class GatewayRunner:
         if _quick_key in self._running_agents and _stale_ts:
             _stale_age = time.time() - _stale_ts
             _stale_agent = self._running_agents.get(_quick_key)
+            # Never evict the pending sentinel — it was just placed moments
+            # ago during the async setup phase before the real agent is
+            # created.  Sentinels have no get_activity_summary(), so the
+            # idle check below would always evaluate to inf >= timeout and
+            # immediately evict them, racing with the setup path.
             _stale_idle = float("inf")  # assume idle if we can't check
             _stale_detail = ""
             if _stale_agent and hasattr(_stale_agent, "get_activity_summary"):
@@ -1874,8 +1880,11 @@ class GatewayRunner:
             # cases where the agent object was garbage-collected).
             _wall_ttl = max(_raw_stale_timeout * 10, 7200) if _raw_stale_timeout > 0 else float("inf")
             _should_evict = (
-                (_raw_stale_timeout > 0 and _stale_idle >= _raw_stale_timeout)
-                or _stale_age > _wall_ttl
+                _stale_agent is not _AGENT_PENDING_SENTINEL
+                and (
+                    (_raw_stale_timeout > 0 and _stale_idle >= _raw_stale_timeout)
+                    or _stale_age > _wall_ttl
+                )
             )
             if _should_evict:
                 logger.warning(
@@ -2812,7 +2821,7 @@ class GatewayRunner:
                         guessed, _ = _mimetypes.guess_type(path)
                         if guessed:
                             mtype = guessed
-                if not (mtype.startswith("application/") or mtype.startswith("text/")):
+                if not mtype.startswith(("application/", "text/")):
                     continue
                 # Extract display filename by stripping the doc_{uuid12}_ prefix
                 import os as _os
@@ -3462,11 +3471,11 @@ class GatewayRunner:
             lines.append(f"_(Requested page {requested_page} was out of range, showing page {page}.)_")
         return "\n".join(lines)
     
-    async def _handle_model_command(self, event: MessageEvent) -> str:
+    async def _handle_model_command(self, event: MessageEvent) -> Optional[str]:
         """Handle /model command — switch model for this session.
 
         Supports:
-          /model                              — show current model info
+          /model                              — interactive picker (Telegram/Discord) or text list
           /model <name>                       — switch for this session only
           /model <name> --global              — switch and persist to config.yaml
           /model <name> --provider <provider> — switch provider + model
@@ -3514,8 +3523,118 @@ class GatewayRunner:
             current_base_url = override.get("base_url", current_base_url)
             current_api_key = override.get("api_key", current_api_key)
 
-        # No args: show authenticated providers with models
+        # No args: show interactive picker (Telegram/Discord) or text list
         if not model_input and not explicit_provider:
+            # Try interactive picker if the platform supports it
+            adapter = self.adapters.get(source.platform)
+            has_picker = (
+                adapter is not None
+                and getattr(type(adapter), "send_model_picker", None) is not None
+            )
+
+            if has_picker:
+                try:
+                    providers = list_authenticated_providers(
+                        current_provider=current_provider,
+                        user_providers=user_provs,
+                        max_models=50,
+                    )
+                except Exception:
+                    providers = []
+
+                if providers:
+                    # Build a callback closure for when the user picks a model.
+                    # Captures self + locals needed for the switch logic.
+                    _self = self
+                    _session_key = session_key
+                    _cur_model = current_model
+                    _cur_provider = current_provider
+                    _cur_base_url = current_base_url
+                    _cur_api_key = current_api_key
+
+                    async def _on_model_selected(
+                        _chat_id: str, model_id: str, provider_slug: str
+                    ) -> str:
+                        """Perform the model switch and return confirmation text."""
+                        result = _switch_model(
+                            raw_input=model_id,
+                            current_provider=_cur_provider,
+                            current_model=_cur_model,
+                            current_base_url=_cur_base_url,
+                            current_api_key=_cur_api_key,
+                            is_global=False,
+                            explicit_provider=provider_slug,
+                        )
+                        if not result.success:
+                            return f"Error: {result.error_message}"
+
+                        # Update cached agent in-place
+                        cached_entry = None
+                        _cache_lock = getattr(_self, "_agent_cache_lock", None)
+                        _cache = getattr(_self, "_agent_cache", None)
+                        if _cache_lock and _cache is not None:
+                            with _cache_lock:
+                                cached_entry = _cache.get(_session_key)
+                        if cached_entry and cached_entry[0] is not None:
+                            try:
+                                cached_entry[0].switch_model(
+                                    new_model=result.new_model,
+                                    new_provider=result.target_provider,
+                                    api_key=result.api_key,
+                                    base_url=result.base_url,
+                                    api_mode=result.api_mode,
+                                )
+                            except Exception as exc:
+                                logger.warning("Picker model switch failed for cached agent: %s", exc)
+
+                        # Store model note + session override
+                        if not hasattr(_self, "_pending_model_notes"):
+                            _self._pending_model_notes = {}
+                        _self._pending_model_notes[_session_key] = (
+                            f"[Note: model was just switched from {_cur_model} to {result.new_model} "
+                            f"via {result.provider_label or result.target_provider}. "
+                            f"Adjust your self-identification accordingly.]"
+                        )
+                        if not hasattr(_self, "_session_model_overrides"):
+                            _self._session_model_overrides = {}
+                        _self._session_model_overrides[_session_key] = {
+                            "model": result.new_model,
+                            "provider": result.target_provider,
+                            "api_key": result.api_key,
+                            "base_url": result.base_url,
+                            "api_mode": result.api_mode,
+                        }
+
+                        # Build confirmation text
+                        plabel = result.provider_label or result.target_provider
+                        lines = [f"Model switched to `{result.new_model}`"]
+                        lines.append(f"Provider: {plabel}")
+                        mi = result.model_info
+                        if mi:
+                            if mi.context_window:
+                                lines.append(f"Context: {mi.context_window:,} tokens")
+                            if mi.max_output:
+                                lines.append(f"Max output: {mi.max_output:,} tokens")
+                            if mi.has_cost_data():
+                                lines.append(f"Cost: {mi.format_cost()}")
+                            lines.append(f"Capabilities: {mi.format_capabilities()}")
+                        lines.append("_(session only — use `/model <name> --global` to persist)_")
+                        return "\n".join(lines)
+
+                    metadata = {"thread_id": source.thread_id} if source.thread_id else None
+                    result = await adapter.send_model_picker(
+                        chat_id=source.chat_id,
+                        providers=providers,
+                        current_model=current_model,
+                        current_provider=current_provider,
+                        session_key=session_key,
+                        on_model_selected=_on_model_selected,
+                        metadata=metadata,
+                    )
+                    if result.success:
+                        return None  # Picker sent — adapter handles the response
+
+            # Fallback: text list (for platforms without picker or if picker failed)
             provider_label = get_label(current_provider)
             lines = [f"Current: `{current_model or 'unknown'}` on {provider_label}", ""]
 
@@ -3789,7 +3908,7 @@ class GatewayRunner:
 
             return f"🎭 Personality set to **{args}**\n_(takes effect on next message)_"
 
-        available = "`none`, " + ", ".join(f"`{n}`" for n in personalities.keys())
+        available = "`none`, " + ", ".join(f"`{n}`" for n in personalities)
         return f"Unknown personality: `{args}`\n\nAvailable: {available}"
     
     async def _handle_retry_command(self, event: MessageEvent) -> str:
@@ -4432,6 +4551,7 @@ class GatewayRunner:
                     provider_data_collection=pr.get("data_collection"),
                     session_id=task_id,
                     platform=platform_key,
+                    user_id=source.user_id,
                     session_db=self._session_db,
                     fallback_model=self._fallback_model,
                 )
@@ -5201,9 +5321,6 @@ class GatewayRunner:
                 old_servers = set(_servers.keys())
 
             # Read new config before shutting down, so we know what will be added/removed
-            new_config = _load_mcp_config()
-            new_server_names = set(new_config.keys())
-
             # Shutdown existing connections
             await loop.run_in_executor(None, shutdown_mcp_servers)
 
@@ -5291,7 +5408,6 @@ class GatewayRunner:
 
         from tools.approval import (
             resolve_gateway_approval, has_blocking_approval,
-            pending_approval_count,
         )
 
         if not has_blocking_approval(session_key):
@@ -5319,6 +5435,11 @@ class GatewayRunner:
         if not count:
             return "No pending command to approve."
 
+        # Resume typing indicator — agent is about to continue processing.
+        _adapter = self.adapters.get(source.platform)
+        if _adapter:
+            _adapter.resume_typing_for_chat(source.chat_id)
+
         count_msg = f" ({count} commands)" if count > 1 else ""
         logger.info("User approved %d dangerous command(s) via /approve%s", count, scope_msg)
         return f"✅ Command{'s' if count > 1 else ''} approved{scope_msg}{count_msg}. The agent is resuming..."
@@ -5351,6 +5472,11 @@ class GatewayRunner:
         if not count:
             return "No pending command to deny."
 
+        # Resume typing indicator — agent continues (with BLOCKED result).
+        _adapter = self.adapters.get(source.platform)
+        if _adapter:
+            _adapter.resume_typing_for_chat(source.chat_id)
+
         count_msg = f" ({count} commands)" if count > 1 else ""
         logger.info("User denied %d dangerous command(s) via /deny", count)
         return f"❌ Command{'s' if count > 1 else ''} denied{count_msg}."
@@ -5936,12 +6062,13 @@ class GatewayRunner:
         platform_name = watcher.get("platform", "")
         chat_id = watcher.get("chat_id", "")
         thread_id = watcher.get("thread_id", "")
+        agent_notify = watcher.get("notify_on_complete", False)
         notify_mode = self._load_background_notifications_mode()
 
-        logger.debug("Process watcher started: %s (every %ss, notify=%s)",
-                      session_id, interval, notify_mode)
+        logger.debug("Process watcher started: %s (every %ss, notify=%s, agent_notify=%s)",
+                      session_id, interval, notify_mode, agent_notify)
 
-        if notify_mode == "off":
+        if notify_mode == "off" and not agent_notify:
             # Still wait for the process to exit so we can log it, but don't
             # push any messages to the user.
             while True:
@@ -5965,6 +6092,47 @@ class GatewayRunner:
             last_output_len = current_output_len
 
             if session.exited:
+                # --- Agent-triggered completion: inject synthetic message ---
+                if agent_notify:
+                    from tools.ansi_strip import strip_ansi
+                    _out = strip_ansi(session.output_buffer[-2000:]) if session.output_buffer else ""
+                    synth_text = (
+                        f"[SYSTEM: Background process {session_id} completed "
+                        f"(exit code {session.exit_code}).\n"
+                        f"Command: {session.command}\n"
+                        f"Output:\n{_out}]"
+                    )
+                    adapter = None
+                    for p, a in self.adapters.items():
+                        if p.value == platform_name:
+                            adapter = a
+                            break
+                    if adapter and chat_id:
+                        try:
+                            from gateway.platforms.base import MessageEvent, MessageType
+                            from gateway.session import SessionSource
+                            from gateway.config import Platform
+                            _platform_enum = Platform(platform_name)
+                            _source = SessionSource(
+                                platform=_platform_enum,
+                                chat_id=chat_id,
+                                thread_id=thread_id or None,
+                            )
+                            synth_event = MessageEvent(
+                                text=synth_text,
+                                message_type=MessageType.TEXT,
+                                source=_source,
+                            )
+                            logger.info(
+                                "Process %s finished — injecting agent notification for session %s",
+                                session_id, session_key,
+                            )
+                            await adapter.handle_message(synth_event)
+                        except Exception as e:
+                            logger.error("Agent notify injection error: %s", e)
+                    break
+
+                # --- Normal text-only notification ---
                 # Decide whether to notify based on mode
                 should_notify = (
                     notify_mode in ("all", "result")
@@ -5989,8 +6157,9 @@ class GatewayRunner:
                             logger.error("Watcher delivery error: %s", e)
                 break
 
-            elif has_new_output and notify_mode == "all":
+            elif has_new_output and notify_mode == "all" and not agent_notify:
                 # New output available -- deliver status update (only in "all" mode)
+                # Skip periodic updates for agent_notify watchers (they only care about completion)
                 new_output = session.output_buffer[-500:] if session.output_buffer else ""
                 message_text = (
                     f"[Background process {session_id} is still running~ "
@@ -6477,6 +6646,7 @@ class GatewayRunner:
                     provider_data_collection=pr.get("data_collection"),
                     session_id=session_id,
                     platform=platform_key,
+                    user_id=source.user_id,
                     session_db=self._session_db,
                     fallback_model=self._fallback_model,
                 )
@@ -6601,6 +6771,15 @@ class GatewayRunner:
                 UX.  Otherwise fall back to a plain text message with
                 ``/approve`` instructions.
                 """
+                # Pause the typing indicator while the agent waits for
+                # user approval.  Critical for Slack's Assistant API where
+                # assistant_threads_setStatus disables the compose box — the
+                # user literally cannot type /approve while "is thinking..."
+                # is active.  The approval message send auto-clears the Slack
+                # status; pausing prevents _keep_typing from re-setting it.
+                # Typing resumes in _handle_approve_command/_handle_deny_command.
+                _status_adapter.pause_typing_for_chat(_status_chat_id)
+
                 cmd = approval_data.get("command", "")
                 desc = approval_data.get("description", "dangerous command")
 
@@ -7019,6 +7198,27 @@ class GatewayRunner:
                     if pending:
                         logger.debug("Processing queued message after agent completion: '%s...'", pending[:40])
             
+            # Safety net: if the pending text is a slash command (e.g. "/stop",
+            # "/new"), discard it — commands should never be passed to the agent
+            # as user input.  The primary fix is in base.py (commands bypass the
+            # active-session guard), but this catches edge cases where command
+            # text leaks through the interrupt_message fallback.
+            if pending and pending.strip().startswith("/"):
+                _pending_parts = pending.strip().split(None, 1)
+                _pending_cmd_word = _pending_parts[0][1:].lower() if _pending_parts else ""
+                if _pending_cmd_word:
+                    try:
+                        from hermes_cli.commands import resolve_command as _rc_pending
+                        if _rc_pending(_pending_cmd_word):
+                            logger.info(
+                                "Discarding command '/%s' from pending queue — "
+                                "commands must not be passed as agent input",
+                                _pending_cmd_word,
+                            )
+                            pending = None
+                    except Exception:
+                        pass
+
             if pending:
                 logger.debug("Processing pending message: '%s...'", pending[:40])
                 

gateway/stream_consumer.py

@@ -28,6 +28,10 @@ logger = logging.getLogger("gateway.stream_consumer")
 # Sentinel to signal the stream is complete
 _DONE = object()
 
+# Sentinel to signal a tool boundary — finalize current message and start a
+# new one so that subsequent text appears below tool progress messages.
+_NEW_SEGMENT = object()
+
 
 @dataclass
 class StreamConsumerConfig:
@@ -78,9 +82,16 @@ class GatewayStreamConsumer:
         return self._already_sent
 
     def on_delta(self, text: str) -> None:
-        """Thread-safe callback — called from the agent's worker thread."""
+        """Thread-safe callback — called from the agent's worker thread.
+
+        When *text* is ``None``, signals a tool boundary: the current message
+        is finalized and subsequent text will be sent as a new message so it
+        appears below any tool-progress messages the gateway sent in between.
+        """
         if text:
             self._queue.put(text)
+        elif text is None:
+            self._queue.put(_NEW_SEGMENT)
 
     def finish(self) -> None:
         """Signal that the stream is complete."""
@@ -96,12 +107,16 @@ class GatewayStreamConsumer:
             while True:
                 # Drain all available items from the queue
                 got_done = False
+                got_segment_break = False
                 while True:
                     try:
                         item = self._queue.get_nowait()
                         if item is _DONE:
                             got_done = True
                             break
+                        if item is _NEW_SEGMENT:
+                            got_segment_break = True
+                            break
                         self._accumulated += item
                     except queue.Empty:
                         break
@@ -111,8 +126,9 @@ class GatewayStreamConsumer:
                 elapsed = now - self._last_edit_time
                 should_edit = (
                     got_done
+                    or got_segment_break
                     or (elapsed >= self.cfg.edit_interval
-                        and len(self._accumulated) > 0)
+                        and self._accumulated)
                     or len(self._accumulated) >= self.cfg.buffer_threshold
                 )
 
@@ -133,7 +149,7 @@ class GatewayStreamConsumer:
                         self._last_sent_text = ""
 
                     display_text = self._accumulated
-                    if not got_done:
+                    if not got_done and not got_segment_break:
                         display_text += self.cfg.cursor
 
                     await self._send_or_edit(display_text)
@@ -145,6 +161,15 @@ class GatewayStreamConsumer:
                         await self._send_or_edit(self._accumulated)
                     return
 
+                # Tool boundary: the should_edit block above already flushed
+                # accumulated text without a cursor.  Reset state so the next
+                # text chunk creates a fresh message below any tool-progress
+                # messages the gateway sent in between.
+                if got_segment_break:
+                    self._message_id = None
+                    self._accumulated = ""
+                    self._last_sent_text = ""
+
                 await asyncio.sleep(0.05)  # Small yield to not busy-loop
 
         except asyncio.CancelledError:

hermes_cli/auth.py

@@ -404,6 +404,47 @@ def detect_zai_endpoint(api_key: str, timeout: float = 8.0) -> Optional[Dict[str
     return None
 
 
+def _resolve_zai_base_url(api_key: str, default_url: str, env_override: str) -> str:
+    """Return the correct Z.AI base URL by probing endpoints.
+
+    If the user has explicitly set GLM_BASE_URL, that always wins.
+    Otherwise, probe the candidate endpoints to find one that accepts the
+    key.  The detected endpoint is cached in provider state (auth.json) keyed
+    on a hash of the API key so subsequent starts skip the probe.
+    """
+    if env_override:
+        return env_override
+
+    # Check provider-state cache for a previously-detected endpoint.
+    auth_store = _load_auth_store()
+    state = _load_provider_state(auth_store, "zai") or {}
+    cached = state.get("detected_endpoint")
+    if isinstance(cached, dict) and cached.get("base_url"):
+        key_hash = cached.get("key_hash", "")
+        if key_hash == hashlib.sha256(api_key.encode()).hexdigest()[:16]:
+            logger.debug("Z.AI: using cached endpoint %s", cached["base_url"])
+            return cached["base_url"]
+
+    # Probe — may take up to ~8s per endpoint.
+    detected = detect_zai_endpoint(api_key)
+    if detected and detected.get("base_url"):
+        # Persist the detection result keyed on the API key hash.
+        key_hash = hashlib.sha256(api_key.encode()).hexdigest()[:16]
+        state["detected_endpoint"] = {
+            "base_url": detected["base_url"],
+            "endpoint_id": detected.get("id", ""),
+            "model": detected.get("model", ""),
+            "label": detected.get("label", ""),
+            "key_hash": key_hash,
+        }
+        _save_provider_state(auth_store, "zai", state)
+        logger.info("Z.AI: auto-detected endpoint %s (%s)", detected["label"], detected["base_url"])
+        return detected["base_url"]
+
+    logger.debug("Z.AI: probe failed, falling back to default %s", default_url)
+    return default_url
+
+
 # =============================================================================
 # Error Types
 # =============================================================================
@@ -936,7 +977,7 @@ def _read_codex_tokens(*, _lock: bool = True) -> Dict[str, Any]:
     state = _load_provider_state(auth_store, "openai-codex")
     if not state:
         raise AuthError(
-            "No Codex credentials stored. Run `hermes login` to authenticate.",
+            "No Codex credentials stored. Run `hermes auth` to authenticate.",
             provider="openai-codex",
             code="codex_auth_missing",
             relogin_required=True,
@@ -944,7 +985,7 @@ def _read_codex_tokens(*, _lock: bool = True) -> Dict[str, Any]:
     tokens = state.get("tokens")
     if not isinstance(tokens, dict):
         raise AuthError(
-            "Codex auth state is missing tokens. Run `hermes login` to re-authenticate.",
+            "Codex auth state is missing tokens. Run `hermes auth` to re-authenticate.",
             provider="openai-codex",
             code="codex_auth_invalid_shape",
             relogin_required=True,
@@ -953,14 +994,14 @@ def _read_codex_tokens(*, _lock: bool = True) -> Dict[str, Any]:
     refresh_token = tokens.get("refresh_token")
     if not isinstance(access_token, str) or not access_token.strip():
         raise AuthError(
-            "Codex auth is missing access_token. Run `hermes login` to re-authenticate.",
+            "Codex auth is missing access_token. Run `hermes auth` to re-authenticate.",
             provider="openai-codex",
             code="codex_auth_missing_access_token",
             relogin_required=True,
         )
     if not isinstance(refresh_token, str) or not refresh_token.strip():
         raise AuthError(
-            "Codex auth is missing refresh_token. Run `hermes login` to re-authenticate.",
+            "Codex auth is missing refresh_token. Run `hermes auth` to re-authenticate.",
             provider="openai-codex",
             code="codex_auth_missing_refresh_token",
             relogin_required=True,
@@ -995,7 +1036,7 @@ def refresh_codex_oauth_pure(
     del access_token  # Access token is only used by callers to decide whether to refresh.
     if not isinstance(refresh_token, str) or not refresh_token.strip():
         raise AuthError(
-            "Codex auth is missing refresh_token. Run `hermes login` to re-authenticate.",
+            "Codex auth is missing refresh_token. Run `hermes auth` to re-authenticate.",
             provider="openai-codex",
             code="codex_auth_missing_refresh_token",
             relogin_required=True,
@@ -1035,7 +1076,7 @@ def refresh_codex_oauth_pure(
                 "Codex refresh token was already consumed by another client "
                 "(e.g. Codex CLI or VS Code extension). "
                 "Run `codex` in your terminal to generate fresh tokens, "
-                "then run `hermes login --provider openai-codex` to re-authenticate."
+                "then run `hermes auth` to re-authenticate."
             )
             relogin_required = True
         raise AuthError(
@@ -1099,7 +1140,8 @@ def _refresh_codex_auth_tokens(
 def _import_codex_cli_tokens() -> Optional[Dict[str, str]]:
     """Try to read tokens from ~/.codex/auth.json (Codex CLI shared file).
     
-    Returns tokens dict if valid, None otherwise. Does NOT write to the shared file.
+    Returns tokens dict if valid and not expired, None otherwise.
+    Does NOT write to the shared file.
     """
     codex_home = os.getenv("CODEX_HOME", "").strip()
     if not codex_home:
@@ -1112,7 +1154,17 @@ def _import_codex_cli_tokens() -> Optional[Dict[str, str]]:
         tokens = payload.get("tokens")
         if not isinstance(tokens, dict):
             return None
-        if not tokens.get("access_token") or not tokens.get("refresh_token"):
+        access_token = tokens.get("access_token")
+        refresh_token = tokens.get("refresh_token")
+        if not access_token or not refresh_token:
+            return None
+        # Reject expired tokens — importing stale tokens from ~/.codex/
+        # that can't be refreshed leaves the user stuck with "Login successful!"
+        # but no working credentials.
+        if _codex_access_token_is_expiring(access_token, 0):
+            logger.debug(
+                "Codex CLI tokens at %s are expired — skipping import.", auth_path,
+            )
             return None
         return dict(tokens)
     except Exception:
@@ -1140,7 +1192,7 @@ def resolve_codex_runtime_credentials(
             logger.info("Migrating Codex credentials from ~/.codex/ to Hermes auth store")
             print("⚠️  Migrating Codex credentials to Hermes's own auth store.")
             print("   This avoids conflicts with Codex CLI and VS Code.")
-            print("   Run `hermes login` to create a fully independent session.\n")
+            print("   Run `hermes auth` to create a fully independent session.\n")
             _save_codex_tokens(cli_tokens)
             data = _read_codex_tokens()
         else:
@@ -1904,7 +1956,36 @@ def get_nous_auth_status() -> Dict[str, Any]:
 
 
 def get_codex_auth_status() -> Dict[str, Any]:
-    """Status snapshot for Codex auth."""
+    """Status snapshot for Codex auth.
+    
+    Checks the credential pool first (where `hermes auth` stores credentials),
+    then falls back to the legacy provider state.
+    """
+    # Check credential pool first — this is where `hermes auth` and
+    # `hermes model` store device_code tokens.
+    try:
+        from agent.credential_pool import load_pool
+        pool = load_pool("openai-codex")
+        if pool and pool.has_credentials():
+            entry = pool.select()
+            if entry is not None:
+                api_key = (
+                    getattr(entry, "runtime_api_key", None)
+                    or getattr(entry, "access_token", "")
+                )
+                if api_key and not _codex_access_token_is_expiring(api_key, 0):
+                    return {
+                        "logged_in": True,
+                        "auth_store": str(_auth_file_path()),
+                        "last_refresh": getattr(entry, "last_refresh", None),
+                        "auth_mode": "chatgpt",
+                        "source": f"pool:{getattr(entry, 'label', 'unknown')}",
+                        "api_key": api_key,
+                    }
+    except Exception:
+        pass
+
+    # Fall back to legacy provider state
     try:
         creds = resolve_codex_runtime_credentials()
         return {
@@ -1913,6 +1994,7 @@ def get_codex_auth_status() -> Dict[str, Any]:
             "last_refresh": creds.get("last_refresh"),
             "auth_mode": creds.get("auth_mode"),
             "source": creds.get("source"),
+            "api_key": creds.get("api_key"),
         }
     except AuthError as exc:
         return {
@@ -2022,6 +2104,8 @@ def resolve_api_key_provider_credentials(provider_id: str) -> Dict[str, Any]:
 
     if provider_id == "kimi-coding":
         base_url = _resolve_kimi_base_url(api_key, pconfig.inference_base_url, env_url)
+    elif provider_id == "zai":
+        base_url = _resolve_zai_base_url(api_key, pconfig.inference_base_url, env_url)
     elif env_url:
         base_url = env_url.rstrip("/")
     else:
@@ -2096,7 +2180,7 @@ def detect_external_credentials() -> List[Dict[str, Any]]:
         found.append({
             "provider": "openai-codex",
             "path": str(codex_path),
-            "label": f"Codex CLI credentials found ({codex_path}) — run `hermes login` to create a separate session",
+            "label": f"Codex CLI credentials found ({codex_path}) — run `hermes auth` to create a separate session",
         })
 
     return found
@@ -2195,14 +2279,21 @@ def _prompt_model_selection(
     model_ids: List[str],
     current_model: str = "",
     pricing: Optional[Dict[str, Dict[str, str]]] = None,
+    unavailable_models: Optional[List[str]] = None,
+    portal_url: str = "",
 ) -> Optional[str]:
     """Interactive model selection. Puts current_model first with a marker. Returns chosen model ID or None.
 
     If *pricing* is provided (``{model_id: {prompt, completion}}``), a compact
     price indicator is shown next to each model in aligned columns.
+
+    If *unavailable_models* is provided, those models are shown grayed out
+    and unselectable, with an upgrade link to *portal_url*.
     """
     from hermes_cli.models import _format_price_per_mtok
 
+    _unavailable = unavailable_models or []
+
     # Reorder: current model first, then the rest (deduplicated)
     ordered = []
     if current_model and current_model in model_ids:
@@ -2211,9 +2302,12 @@ def _prompt_model_selection(
         if mid not in ordered:
             ordered.append(mid)
 
+    # All models for column-width computation (selectable + unavailable)
+    all_models = list(ordered) + list(_unavailable)
+
     # Column-aligned labels when pricing is available
-    has_pricing = bool(pricing and any(pricing.get(m) for m in ordered))
-    name_col = max((len(m) for m in ordered), default=0) + 2 if has_pricing else 0
+    has_pricing = bool(pricing and any(pricing.get(m) for m in all_models))
+    name_col = max((len(m) for m in all_models), default=0) + 2 if has_pricing else 0
 
     # Pre-compute formatted prices and dynamic column widths
     _price_cache: dict[str, tuple[str, str, str]] = {}
@@ -2221,7 +2315,7 @@ def _prompt_model_selection(
     cache_col = 0  # only set if any model has cache pricing
     has_cache = False
     if has_pricing:
-        for mid in ordered:
+        for mid in all_models:
             p = pricing.get(mid)  # type: ignore[union-attr]
             if p:
                 inp = _format_price_per_mtok(p.get("prompt", ""))
@@ -2266,12 +2360,35 @@ def _prompt_model_selection(
             header += f"  {'Cache':>{cache_col}}"
         menu_title += header + "  /Mtok"
 
+    # ANSI escape for dim text
+    _DIM = "\033[2m"
+    _RESET = "\033[0m"
+
     # Try arrow-key menu first, fall back to number input
     try:
         from simple_term_menu import TerminalMenu
+
         choices = [f"  {_label(mid)}" for mid in ordered]
         choices.append("  Enter custom model name")
         choices.append("  Skip (keep current)")
+
+        # Print the unavailable block BEFORE the menu via regular print().
+        # simple_term_menu pads title lines to terminal width (causes wrapping),
+        # so we keep the title minimal and use stdout for the static block.
+        # clear_screen=False means our printed output stays visible above.
+        _upgrade_url = (portal_url or DEFAULT_NOUS_PORTAL_URL).rstrip("/")
+        if _unavailable:
+            print(menu_title)
+            print()
+            for mid in _unavailable:
+                print(f"{_DIM}     {_label(mid)}{_RESET}")
+            print()
+            print(f"{_DIM}  ── Upgrade at {_upgrade_url} for paid models ──{_RESET}")
+            print()
+            effective_title = "Available free models:"
+        else:
+            effective_title = menu_title
+
         menu = TerminalMenu(
             choices,
             cursor_index=default_idx,
@@ -2280,7 +2397,7 @@ def _prompt_model_selection(
             menu_highlight_style=("fg_green",),
             cycle_cursor=True,
             clear_screen=False,
-            title=menu_title,
+            title=effective_title,
         )
         idx = menu.show()
         if idx is None:
@@ -2303,6 +2420,13 @@ def _prompt_model_selection(
     n = len(ordered)
     print(f"  {n + 1:>{num_width}}. Enter custom model name")
     print(f"  {n + 2:>{num_width}}. Skip (keep current)")
+
+    if _unavailable:
+        _upgrade_url = (portal_url or DEFAULT_NOUS_PORTAL_URL).rstrip("/")
+        print()
+        print(f"  {_DIM}── Unavailable models (requires paid tier — upgrade at {_upgrade_url}) ──{_RESET}")
+        for mid in _unavailable:
+            print(f"  {'':>{num_width}}  {_DIM}{_label(mid)}{_RESET}")
     print()
 
     while True:
@@ -2345,8 +2469,8 @@ def _save_model_choice(model_id: str) -> None:
 def login_command(args) -> None:
     """Deprecated: use 'hermes model' or 'hermes setup' instead."""
     print("The 'hermes login' command has been removed.")
-    print("Use 'hermes model' to select a provider and model,")
-    print("or 'hermes setup' for full interactive setup.")
+    print("Use 'hermes auth' to manage credentials,")
+    print("'hermes model' to select a provider, or 'hermes setup' for full setup.")
     raise SystemExit(0)
 
 
@@ -2356,17 +2480,25 @@ def _login_openai_codex(args, pconfig: ProviderConfig) -> None:
     # Check for existing Hermes-owned credentials
     try:
         existing = resolve_codex_runtime_credentials()
-        print("Existing Codex credentials found in Hermes auth store.")
-        try:
-            reuse = input("Use existing credentials? [Y/n]: ").strip().lower()
-        except (EOFError, KeyboardInterrupt):
-            reuse = "y"
-        if reuse in ("", "y", "yes"):
-            config_path = _update_config_for_provider("openai-codex", existing.get("base_url", DEFAULT_CODEX_BASE_URL))
-            print()
-            print("Login successful!")
-            print(f"  Config updated: {config_path} (model.provider=openai-codex)")
-            return
+        # Verify the resolved token is actually usable (not expired).
+        # resolve_codex_runtime_credentials attempts refresh, so if we get
+        # here the token should be valid — but double-check before telling
+        # the user "Login successful!".
+        _resolved_key = existing.get("api_key", "")
+        if isinstance(_resolved_key, str) and _resolved_key and not _codex_access_token_is_expiring(_resolved_key, 60):
+            print("Existing Codex credentials found in Hermes auth store.")
+            try:
+                reuse = input("Use existing credentials? [Y/n]: ").strip().lower()
+            except (EOFError, KeyboardInterrupt):
+                reuse = "y"
+            if reuse in ("", "y", "yes"):
+                config_path = _update_config_for_provider("openai-codex", existing.get("base_url", DEFAULT_CODEX_BASE_URL))
+                print()
+                print("Login successful!")
+                print(f"  Config updated: {config_path} (model.provider=openai-codex)")
+                return
+        else:
+            print("Existing Codex credentials are expired. Starting fresh login...")
     except AuthError:
         pass
 
@@ -2707,7 +2839,6 @@ def _login_nous(args, pconfig: ProviderConfig) -> None:
         )
 
         inference_base_url = auth_state["inference_base_url"]
-        verify: bool | str = False if insecure else (ca_bundle if ca_bundle else True)
 
         with _auth_store_lock():
             auth_store = _load_auth_store()
@@ -2729,16 +2860,37 @@ def _login_nous(args, pconfig: ProviderConfig) -> None:
                     code="invalid_token",
                 )
 
-            from hermes_cli.models import _PROVIDER_MODELS
+            from hermes_cli.models import (
+                _PROVIDER_MODELS, get_pricing_for_provider, filter_nous_free_models,
+                check_nous_free_tier, partition_nous_models_by_tier,
+            )
             model_ids = _PROVIDER_MODELS.get("nous", [])
 
             print()
+            unavailable_models: list = []
+            if model_ids:
+                pricing = get_pricing_for_provider("nous")
+                model_ids = filter_nous_free_models(model_ids, pricing)
+                free_tier = check_nous_free_tier()
+                if free_tier:
+                    model_ids, unavailable_models = partition_nous_models_by_tier(
+                        model_ids, pricing, free_tier=True,
+                    )
+            _portal = auth_state.get("portal_base_url", "")
             if model_ids:
                 print(f"Showing {len(model_ids)} curated models — use \"Enter custom model name\" for others.")
-                selected_model = _prompt_model_selection(model_ids)
+                selected_model = _prompt_model_selection(
+                    model_ids, pricing=pricing,
+                    unavailable_models=unavailable_models,
+                    portal_url=_portal,
+                )
                 if selected_model:
                     _save_model_choice(selected_model)
                     print(f"Default model set to: {selected_model}")
+            elif unavailable_models:
+                _url = (_portal or DEFAULT_NOUS_PORTAL_URL).rstrip("/")
+                print("No free models currently available.")
+                print(f"Upgrade at {_url} to access paid models.")
             else:
                 print("No curated models available for Nous Portal.")
         except Exception as exc:

hermes_cli/auth_commands.py

@@ -18,7 +18,6 @@ from agent.credential_pool import (
     STRATEGY_ROUND_ROBIN,
     STRATEGY_RANDOM,
     STRATEGY_LEAST_USED,
-    SUPPORTED_POOL_STRATEGIES,
     PooledCredential,
     _exhausted_until,
     _normalize_custom_pool_name,
@@ -305,6 +304,32 @@ def auth_remove_command(args) -> None:
             if cleared:
                 print(f"Cleared {env_var} from .env")
 
+    # If this was a singleton-seeded credential (OAuth device_code, hermes_pkce),
+    # clear the underlying auth store / credential file so it doesn't get
+    # re-seeded on the next load_pool() call.
+    elif removed.source == "device_code" and provider in ("openai-codex", "nous"):
+        from hermes_cli.auth import (
+            _load_auth_store, _save_auth_store, _auth_store_lock,
+        )
+        with _auth_store_lock():
+            auth_store = _load_auth_store()
+            providers_dict = auth_store.get("providers")
+            if isinstance(providers_dict, dict) and provider in providers_dict:
+                del providers_dict[provider]
+                _save_auth_store(auth_store)
+                print(f"Cleared {provider} OAuth tokens from auth store")
+
+    elif removed.source == "hermes_pkce" and provider == "anthropic":
+        from hermes_constants import get_hermes_home
+        oauth_file = get_hermes_home() / ".anthropic_oauth.json"
+        if oauth_file.exists():
+            oauth_file.unlink()
+            print("Cleared Hermes Anthropic OAuth credentials")
+
+    elif removed.source == "claude_code" and provider == "anthropic":
+        print("Note: Claude Code credentials live in ~/.claude/.credentials.json")
+        print("      Remove them manually if you want to deauthorize Claude Code.")
+
 
 def auth_reset_command(args) -> None:
     provider = _normalize_provider(getattr(args, "provider", ""))

hermes_cli/banner.py

@@ -5,7 +5,6 @@ Pure display functions with no HermesCLI state dependency.
 
 import json
 import logging
-import os
 import shutil
 import subprocess
 import threading

hermes_cli/callbacks.py

@@ -25,7 +25,7 @@ def clarify_callback(cli, question, choices):
 
     timeout = CLI_CONFIG.get("clarify", {}).get("timeout", 120)
     response_queue = queue.Queue()
-    is_open_ended = not choices or len(choices) == 0
+    is_open_ended = not choices
 
     cli._clarify_state = {
         "question": question,
@@ -63,47 +63,6 @@ def clarify_callback(cli, question, choices):
     )
 
 
-def sudo_password_callback(cli) -> str:
-    """Prompt for sudo password through the TUI.
-
-    Sets up a password input area and blocks until the user responds.
-    """
-    timeout = 45
-    response_queue = queue.Queue()
-
-    cli._sudo_state = {"response_queue": response_queue}
-    cli._sudo_deadline = _time.monotonic() + timeout
-
-    if hasattr(cli, "_app") and cli._app:
-        cli._app.invalidate()
-
-    while True:
-        try:
-            result = response_queue.get(timeout=1)
-            cli._sudo_state = None
-            cli._sudo_deadline = 0
-            if hasattr(cli, "_app") and cli._app:
-                cli._app.invalidate()
-            if result:
-                cprint(f"\n{_DIM}  ✓ Password received (cached for session){_RST}")
-            else:
-                cprint(f"\n{_DIM}  ⏭ Skipped{_RST}")
-            return result
-        except queue.Empty:
-            remaining = cli._sudo_deadline - _time.monotonic()
-            if remaining <= 0:
-                break
-            if hasattr(cli, "_app") and cli._app:
-                cli._app.invalidate()
-
-    cli._sudo_state = None
-    cli._sudo_deadline = 0
-    if hasattr(cli, "_app") and cli._app:
-        cli._app.invalidate()
-    cprint(f"\n{_DIM}  ⏱ Timeout — continuing without sudo{_RST}")
-    return ""
-
-
 def prompt_for_secret(cli, var_name: str, prompt: str, metadata=None) -> dict:
     """Prompt for a secret value through the TUI (e.g. API keys for skills).
 

hermes_cli/claw.py

@@ -10,7 +10,6 @@ Usage:
 
 import importlib.util
 import logging
-import shutil
 import sys
 from datetime import datetime
 from pathlib import Path
@@ -24,7 +23,6 @@ from hermes_cli.setup import (
     print_info,
     print_success,
     print_error,
-    print_warning,
     prompt_yes_no,
 )
 

hermes_cli/clipboard.py

@@ -1,4 +1,4 @@
-"""Clipboard image extraction for macOS, Linux, and WSL2.
+"""Clipboard image extraction for macOS, Windows, Linux, and WSL2.
 
 Provides a single function `save_clipboard_image(dest)` that checks the
 system clipboard for image data, saves it to *dest* as PNG, and returns
@@ -6,9 +6,10 @@ True on success.  No external Python dependencies — uses only OS-level
 CLI tools that ship with the platform (or are commonly installed).
 
 Platform support:
-  macOS  — osascript (always available), pngpaste (if installed)
-  WSL2   — powershell.exe via .NET System.Windows.Forms.Clipboard
-  Linux  — wl-paste (Wayland), xclip (X11)
+  macOS   — osascript (always available), pngpaste (if installed)
+  Windows — PowerShell via .NET System.Windows.Forms.Clipboard
+  WSL2    — powershell.exe via .NET System.Windows.Forms.Clipboard
+  Linux   — wl-paste (Wayland), xclip (X11)
 """
 
 import base64
@@ -32,6 +33,8 @@ def save_clipboard_image(dest: Path) -> bool:
     dest.parent.mkdir(parents=True, exist_ok=True)
     if sys.platform == "darwin":
         return _macos_save(dest)
+    if sys.platform == "win32":
+        return _windows_save(dest)
     return _linux_save(dest)
 
 
@@ -42,6 +45,8 @@ def has_clipboard_image() -> bool:
     """
     if sys.platform == "darwin":
         return _macos_has_image()
+    if sys.platform == "win32":
+        return _windows_has_image()
     if _is_wsl():
         return _wsl_has_image()
     if os.environ.get("WAYLAND_DISPLAY"):
@@ -112,6 +117,104 @@ def _macos_osascript(dest: Path) -> bool:
     return False
 
 
+# ── Shared PowerShell scripts (native Windows + WSL2) ─────────────────────
+
+# .NET System.Windows.Forms.Clipboard — used by both native Windows (powershell)
+# and WSL2 (powershell.exe) paths.
+_PS_CHECK_IMAGE = (
+    "Add-Type -AssemblyName System.Windows.Forms;"
+    "[System.Windows.Forms.Clipboard]::ContainsImage()"
+)
+
+_PS_EXTRACT_IMAGE = (
+    "Add-Type -AssemblyName System.Windows.Forms;"
+    "Add-Type -AssemblyName System.Drawing;"
+    "$img = [System.Windows.Forms.Clipboard]::GetImage();"
+    "if ($null -eq $img) { exit 1 }"
+    "$ms = New-Object System.IO.MemoryStream;"
+    "$img.Save($ms, [System.Drawing.Imaging.ImageFormat]::Png);"
+    "[System.Convert]::ToBase64String($ms.ToArray())"
+)
+
+
+# ── Native Windows ────────────────────────────────────────────────────────
+
+# Native Windows uses ``powershell`` (Windows PowerShell 5.1, always present)
+# or ``pwsh`` (PowerShell 7+, optional).  Discovery is cached per-process.
+
+
+def _find_powershell() -> str | None:
+    """Return the first available PowerShell executable, or None."""
+    for name in ("powershell", "pwsh"):
+        try:
+            r = subprocess.run(
+                [name, "-NoProfile", "-NonInteractive", "-Command", "echo ok"],
+                capture_output=True, text=True, timeout=5,
+            )
+            if r.returncode == 0 and "ok" in r.stdout:
+                return name
+        except FileNotFoundError:
+            continue
+        except Exception:
+            continue
+    return None
+
+
+# Cache the resolved PowerShell executable (checked once per process)
+_ps_exe: str | None | bool = False  # False = not yet checked
+
+
+def _get_ps_exe() -> str | None:
+    global _ps_exe
+    if _ps_exe is False:
+        _ps_exe = _find_powershell()
+    return _ps_exe
+
+
+def _windows_has_image() -> bool:
+    """Check if the Windows clipboard contains an image."""
+    ps = _get_ps_exe()
+    if ps is None:
+        return False
+    try:
+        r = subprocess.run(
+            [ps, "-NoProfile", "-NonInteractive", "-Command", _PS_CHECK_IMAGE],
+            capture_output=True, text=True, timeout=5,
+        )
+        return r.returncode == 0 and "True" in r.stdout
+    except Exception as e:
+        logger.debug("Windows clipboard image check failed: %s", e)
+    return False
+
+
+def _windows_save(dest: Path) -> bool:
+    """Extract clipboard image on native Windows via PowerShell → base64 PNG."""
+    ps = _get_ps_exe()
+    if ps is None:
+        logger.debug("No PowerShell found — Windows clipboard image paste unavailable")
+        return False
+    try:
+        r = subprocess.run(
+            [ps, "-NoProfile", "-NonInteractive", "-Command", _PS_EXTRACT_IMAGE],
+            capture_output=True, text=True, timeout=15,
+        )
+        if r.returncode != 0:
+            return False
+
+        b64_data = r.stdout.strip()
+        if not b64_data:
+            return False
+
+        png_bytes = base64.b64decode(b64_data)
+        dest.write_bytes(png_bytes)
+        return dest.exists() and dest.stat().st_size > 0
+
+    except Exception as e:
+        logger.debug("Windows clipboard image extraction failed: %s", e)
+        dest.unlink(missing_ok=True)
+    return False
+
+
 # ── Linux ────────────────────────────────────────────────────────────────
 
 def _is_wsl() -> bool:
@@ -142,24 +245,7 @@ def _linux_save(dest: Path) -> bool:
 
 
 # ── WSL2 (powershell.exe) ────────────────────────────────────────────────
-
-# PowerShell script: get clipboard image as base64-encoded PNG on stdout.
-# Using .NET System.Windows.Forms.Clipboard — always available on Windows.
-_PS_CHECK_IMAGE = (
-    "Add-Type -AssemblyName System.Windows.Forms;"
-    "[System.Windows.Forms.Clipboard]::ContainsImage()"
-)
-
-_PS_EXTRACT_IMAGE = (
-    "Add-Type -AssemblyName System.Windows.Forms;"
-    "Add-Type -AssemblyName System.Drawing;"
-    "$img = [System.Windows.Forms.Clipboard]::GetImage();"
-    "if ($null -eq $img) { exit 1 }"
-    "$ms = New-Object System.IO.MemoryStream;"
-    "$img.Save($ms, [System.Drawing.Imaging.ImageFormat]::Png);"
-    "[System.Convert]::ToBase64String($ms.ToArray())"
-)
-
+# Reuses _PS_CHECK_IMAGE / _PS_EXTRACT_IMAGE defined above.
 
 def _wsl_has_image() -> bool:
     """Check if Windows clipboard has an image (via powershell.exe)."""

hermes_cli/commands.py

@@ -294,10 +294,8 @@ def _resolve_config_gates() -> set[str]:
         return set()
     try:
         import yaml
-        config_path = os.path.join(
-            os.getenv("HERMES_HOME", os.path.expanduser("~/.hermes")),
-            "config.yaml",
-        )
+        from hermes_constants import get_hermes_home
+        config_path = str(get_hermes_home() / "config.yaml")
         if os.path.exists(config_path):
             with open(config_path, encoding="utf-8") as f:
                 cfg = yaml.safe_load(f) or {}

hermes_cli/config.py

@@ -1881,6 +1881,24 @@ def _normalize_max_turns_config(config: Dict[str, Any]) -> Dict[str, Any]:
 
 
 
+def read_raw_config() -> Dict[str, Any]:
+    """Read ~/.hermes/config.yaml as-is, without merging defaults or migrating.
+
+    Returns the raw YAML dict, or ``{}`` if the file doesn't exist or can't
+    be parsed.  Use this for lightweight config reads where you just need a
+    single value and don't want the overhead of ``load_config()``'s deep-merge
+    + migration pipeline.
+    """
+    try:
+        config_path = get_config_path()
+        if config_path.exists():
+            with open(config_path, encoding="utf-8") as f:
+                return yaml.safe_load(f) or {}
+    except Exception:
+        pass
+    return {}
+
+
 def load_config() -> Dict[str, Any]:
     """Load configuration from ~/.hermes/config.yaml."""
     import copy
@@ -1932,8 +1950,8 @@ _FALLBACK_COMMENT = """
 #
 # Supported providers:
 #   openrouter   (OPENROUTER_API_KEY)  — routes to any model
-#   openai-codex (OAuth — hermes login) — OpenAI Codex
-#   nous         (OAuth — hermes login) — Nous Portal
+#   openai-codex (OAuth — hermes auth) — OpenAI Codex
+#   nous         (OAuth — hermes auth) — Nous Portal
 #   zai          (ZAI_API_KEY)         — Z.AI / GLM
 #   kimi-coding  (KIMI_API_KEY)        — Kimi / Moonshot
 #   minimax      (MINIMAX_API_KEY)     — MiniMax
@@ -1975,8 +1993,8 @@ _COMMENTED_SECTIONS = """
 #
 # Supported providers:
 #   openrouter   (OPENROUTER_API_KEY)  — routes to any model
-#   openai-codex (OAuth — hermes login) — OpenAI Codex
-#   nous         (OAuth — hermes login) — Nous Portal
+#   openai-codex (OAuth — hermes auth) — OpenAI Codex
+#   nous         (OAuth — hermes auth) — Nous Portal
 #   zai          (ZAI_API_KEY)         — Z.AI / GLM
 #   kimi-coding  (KIMI_API_KEY)        — Kimi / Moonshot
 #   minimax      (MINIMAX_API_KEY)     — MiniMax
@@ -2520,7 +2538,7 @@ def set_config_value(key: str, value: str):
         'TINKER_API_KEY',
     ]
     
-    if key.upper() in api_keys or key.upper().endswith('_API_KEY') or key.upper().endswith('_TOKEN') or key.upper().startswith('TERMINAL_SSH'):
+    if key.upper() in api_keys or key.upper().endswith(('_API_KEY', '_TOKEN')) or key.upper().startswith('TERMINAL_SSH'):
         save_env_value(key.upper(), value)
         print(f"✓ Set {key} in {get_env_path()}")
         return

hermes_cli/doctor.py

@@ -920,8 +920,8 @@ def run_doctor(args):
                         pass
     except ImportError:
         pass
-    except Exception as _e:
-        logger.debug("Profile health check failed: %s", _e)
+    except Exception:
+        pass
 
     # =========================================================================
     # Summary

hermes_cli/gateway.py

@@ -1803,8 +1803,7 @@ def _setup_signal():
         print_warning("signal-cli not found on PATH.")
         print_info("  Signal requires signal-cli running as an HTTP daemon.")
         print_info("  Install options:")
-        print_info("    Linux:  sudo apt install signal-cli")
-        print_info("            or download from https://github.com/AsamK/signal-cli")
+        print_info("    Linux:  download from https://github.com/AsamK/signal-cli/releases")
         print_info("    macOS:  brew install signal-cli")
         print_info("    Docker: bbernhard/signal-cli-rest-api")
         print()

hermes_cli/logs.py

@@ -15,7 +15,6 @@ Usage examples::
     hermes logs --since 30m -f     # follow, starting 30 min ago
 """
 
-import os
 import re
 import sys
 import time

hermes_cli/main.py

@@ -1154,7 +1154,7 @@ def _model_flow_nous(config, current_model="", args=None):
     from hermes_cli.auth import (
         get_provider_auth_state, _prompt_model_selection, _save_model_choice,
         _update_config_for_provider, resolve_nous_runtime_credentials,
-        fetch_nous_models, AuthError, format_auth_error,
+        AuthError, format_auth_error,
         _login_nous, PROVIDER_REGISTRY,
     )
     from hermes_cli.config import get_env_value, save_config, save_env_value
@@ -1195,14 +1195,15 @@ def _model_flow_nous(config, current_model="", args=None):
     # Already logged in — use curated model list (same as OpenRouter defaults).
     # The live /models endpoint returns hundreds of models; the curated list
     # shows only agentic models users recognize from OpenRouter.
-    from hermes_cli.models import _PROVIDER_MODELS, get_pricing_for_provider
+    from hermes_cli.models import (
+        _PROVIDER_MODELS, get_pricing_for_provider, filter_nous_free_models,
+        check_nous_free_tier, partition_nous_models_by_tier,
+    )
     model_ids = _PROVIDER_MODELS.get("nous", [])
     if not model_ids:
         print("No curated models available for Nous Portal.")
         return
 
-    print(f"Showing {len(model_ids)} curated models — use \"Enter custom model name\" for others.")
-
     # Verify credentials are still valid (catches expired sessions early)
     try:
         creds = resolve_nous_runtime_credentials(min_key_ttl_seconds=5 * 60)
@@ -1228,7 +1229,44 @@ def _model_flow_nous(config, current_model="", args=None):
     # Fetch live pricing (non-blocking — returns empty dict on failure)
     pricing = get_pricing_for_provider("nous")
 
-    selected = _prompt_model_selection(model_ids, current_model=current_model, pricing=pricing)
+    # Check if user is on free tier
+    free_tier = check_nous_free_tier()
+
+    # For both tiers: apply the allowlist filter first (removes non-allowlisted
+    # free models and allowlist models that aren't actually free).
+    # Then for free users: partition remaining models into selectable/unavailable.
+    model_ids = filter_nous_free_models(model_ids, pricing)
+    unavailable_models: list[str] = []
+    if free_tier:
+        model_ids, unavailable_models = partition_nous_models_by_tier(model_ids, pricing, free_tier=True)
+
+    if not model_ids and not unavailable_models:
+        print("No models available for Nous Portal after filtering.")
+        return
+
+    # Resolve portal URL for upgrade links (may differ on staging)
+    _nous_portal_url = ""
+    try:
+        _nous_state = get_provider_auth_state("nous")
+        if _nous_state:
+            _nous_portal_url = _nous_state.get("portal_base_url", "")
+    except Exception:
+        pass
+
+    if free_tier and not model_ids:
+        print("No free models currently available.")
+        if unavailable_models:
+            from hermes_cli.auth import DEFAULT_NOUS_PORTAL_URL
+            _url = (_nous_portal_url or DEFAULT_NOUS_PORTAL_URL).rstrip("/")
+            print(f"Upgrade at {_url} to access paid models.")
+        return
+
+    print(f"Showing {len(model_ids)} curated models — use \"Enter custom model name\" for others.")
+
+    selected = _prompt_model_selection(
+        model_ids, current_model=current_model, pricing=pricing,
+        unavailable_models=unavailable_models, portal_url=_nous_portal_url,
+    )
     if selected:
         _save_model_choice(selected)
         # Reactivate Nous as the provider and update config
@@ -1276,7 +1314,6 @@ def _model_flow_openai_codex(config, current_model=""):
         PROVIDER_REGISTRY, DEFAULT_CODEX_BASE_URL,
     )
     from hermes_cli.codex_models import get_codex_model_ids
-    from hermes_cli.config import get_env_value, save_env_value
     import argparse
 
     status = get_codex_auth_status()
@@ -1294,12 +1331,21 @@ def _model_flow_openai_codex(config, current_model=""):
             return
 
     _codex_token = None
+    # Prefer credential pool (where `hermes auth` stores device_code tokens),
+    # fall back to legacy provider state.
     try:
-        from hermes_cli.auth import resolve_codex_runtime_credentials
-        _codex_creds = resolve_codex_runtime_credentials()
-        _codex_token = _codex_creds.get("api_key")
+        _codex_status = get_codex_auth_status()
+        if _codex_status.get("logged_in"):
+            _codex_token = _codex_status.get("api_key")
     except Exception:
         pass
+    if not _codex_token:
+        try:
+            from hermes_cli.auth import resolve_codex_runtime_credentials
+            _codex_creds = resolve_codex_runtime_credentials()
+            _codex_token = _codex_creds.get("api_key")
+        except Exception:
+            pass
 
     codex_models = get_codex_model_ids(access_token=_codex_token)
 
@@ -1320,7 +1366,7 @@ def _model_flow_custom(config):
     so it appears in the provider menu on subsequent runs.
     """
     from hermes_cli.auth import _save_model_choice, deactivate_provider
-    from hermes_cli.config import get_env_value, save_env_value, load_config, save_config
+    from hermes_cli.config import get_env_value, load_config, save_config
 
     current_url = get_env_value("OPENAI_BASE_URL") or ""
     current_key = get_env_value("OPENAI_API_KEY") or ""
@@ -1582,7 +1628,7 @@ def _model_flow_named_custom(config, provider_info):
     Otherwise probes the endpoint's /models API to let the user pick one.
     """
     from hermes_cli.auth import _save_model_choice, deactivate_provider
-    from hermes_cli.config import save_env_value, load_config, save_config
+    from hermes_cli.config import load_config, save_config
     from hermes_cli.models import fetch_api_models
 
     name = provider_info["name"]
@@ -1792,7 +1838,7 @@ def _model_flow_copilot(config, current_model=""):
         deactivate_provider,
         resolve_api_key_provider_credentials,
     )
-    from hermes_cli.config import get_env_value, save_env_value, load_config, save_config
+    from hermes_cli.config import save_env_value, load_config, save_config
     from hermes_cli.models import (
         fetch_api_models,
         fetch_github_model_catalog,
@@ -2383,8 +2429,6 @@ def _model_flow_anthropic(config, current_model=""):
     )
     from hermes_cli.models import _PROVIDER_MODELS
 
-    pconfig = PROVIDER_REGISTRY["anthropic"]
-
     # Check ALL credential sources
     existing_key = (
         get_env_value("ANTHROPIC_TOKEN")
@@ -3557,7 +3601,7 @@ def cmd_update(args):
         try:
             from hermes_cli.profiles import list_profiles, get_active_profile_name, seed_profile_skills
             active = get_active_profile_name()
-            other_profiles = [p for p in list_profiles() if not p.is_default and p.name != active]
+            other_profiles = [p for p in list_profiles() if p.name != active]
             if other_profiles:
                 print()
                 print("→ Syncing bundled skills to other profiles...")
@@ -3653,7 +3697,7 @@ def cmd_update(args):
         try:
             from hermes_cli.gateway import (
                 is_macos, is_linux, _ensure_user_systemd_env,
-                get_systemd_linger_status, find_gateway_pids,
+                find_gateway_pids,
                 _get_service_pids,
             )
             import signal as _signal
@@ -3809,7 +3853,7 @@ def cmd_profile(args):
     """Profile management — create, delete, list, switch, alias."""
     from hermes_cli.profiles import (
         list_profiles, create_profile, delete_profile, seed_profile_skills,
-        get_active_profile, set_active_profile, get_active_profile_name,
+        set_active_profile, get_active_profile_name,
         check_alias_collision, create_wrapper_script, remove_wrapper_script,
         _is_wrapper_dir_in_path, _get_wrapper_dir,
     )
@@ -3937,7 +3981,6 @@ def cmd_profile(args):
             print(f"  {name} chat               Start chatting")
             print(f"  {name} gateway start      Start the messaging gateway")
             if clone or clone_all:
-                from hermes_constants import get_hermes_home
                 profile_dir_display = f"~/.hermes/profiles/{name}"
                 print(f"\n  Edit {profile_dir_display}/.env for different API keys")
                 print(f"  Edit {profile_dir_display}/SOUL.md for different personality")
@@ -4360,7 +4403,7 @@ For more help on a command:
     gateway_uninstall.add_argument("--system", action="store_true", help="Target the Linux system-level gateway service")
 
     # gateway setup
-    gateway_setup = gateway_subparsers.add_parser("setup", help="Configure messaging platforms")
+    gateway_subparsers.add_parser("setup", help="Configure messaging platforms")
 
     gateway_parser.set_defaults(func=cmd_gateway)
     
@@ -4635,10 +4678,10 @@ For more help on a command:
     config_subparsers = config_parser.add_subparsers(dest="config_command")
     
     # config show (default)
-    config_show = config_subparsers.add_parser("show", help="Show current configuration")
+    config_subparsers.add_parser("show", help="Show current configuration")
     
     # config edit
-    config_edit = config_subparsers.add_parser("edit", help="Open config file in editor")
+    config_subparsers.add_parser("edit", help="Open config file in editor")
     
     # config set
     config_set = config_subparsers.add_parser("set", help="Set a configuration value")
@@ -4646,16 +4689,16 @@ For more help on a command:
     config_set.add_argument("value", nargs="?", help="Value to set")
     
     # config path
-    config_path = config_subparsers.add_parser("path", help="Print config file path")
+    config_subparsers.add_parser("path", help="Print config file path")
     
     # config env-path
-    config_env = config_subparsers.add_parser("env-path", help="Print .env file path")
+    config_subparsers.add_parser("env-path", help="Print .env file path")
     
     # config check
-    config_check = config_subparsers.add_parser("check", help="Check for missing/outdated config")
+    config_subparsers.add_parser("check", help="Check for missing/outdated config")
     
     # config migrate
-    config_migrate = config_subparsers.add_parser("migrate", help="Update config with new options")
+    config_subparsers.add_parser("migrate", help="Update config with new options")
     
     config_parser.set_defaults(func=cmd_config)
     
@@ -4669,7 +4712,7 @@ For more help on a command:
     )
     pairing_sub = pairing_parser.add_subparsers(dest="pairing_action")
 
-    pairing_list_parser = pairing_sub.add_parser("list", help="Show pending + approved users")
+    pairing_sub.add_parser("list", help="Show pending + approved users")
 
     pairing_approve_parser = pairing_sub.add_parser("approve", help="Approve a pairing code")
     pairing_approve_parser.add_argument("platform", help="Platform name (telegram, discord, slack, whatsapp)")
@@ -4679,7 +4722,7 @@ For more help on a command:
     pairing_revoke_parser.add_argument("platform", help="Platform name")
     pairing_revoke_parser.add_argument("user_id", help="User ID to revoke")
 
-    pairing_clear_parser = pairing_sub.add_parser("clear-pending", help="Clear all pending codes")
+    pairing_sub.add_parser("clear-pending", help="Clear all pending codes")
 
     def cmd_pairing(args):
         from hermes_cli.pairing import pairing_command
@@ -4855,7 +4898,7 @@ For more help on a command:
     memory_sub = memory_parser.add_subparsers(dest="memory_command")
     memory_sub.add_parser("setup", help="Interactive provider selection and configuration")
     memory_sub.add_parser("status", help="Show current memory provider config")
-    memory_off_p = memory_sub.add_parser("off", help="Disable external provider (built-in only)")
+    memory_sub.add_parser("off", help="Disable external provider (built-in only)")
 
     def cmd_memory(args):
         sub = getattr(args, "memory_command", None)
@@ -5019,7 +5062,7 @@ For more help on a command:
     sessions_prune.add_argument("--source", help="Only prune sessions from this source")
     sessions_prune.add_argument("--yes", "-y", action="store_true", help="Skip confirmation")
 
-    sessions_stats = sessions_subparsers.add_parser("stats", help="Show session store statistics")
+    sessions_subparsers.add_parser("stats", help="Show session store statistics")
 
     sessions_rename = sessions_subparsers.add_parser("rename", help="Set or change a session's title")
     sessions_rename.add_argument("session_id", help="Session ID to rename")
@@ -5379,7 +5422,7 @@ For more help on a command:
     )
     profile_subparsers = profile_parser.add_subparsers(dest="profile_action")
 
-    profile_list = profile_subparsers.add_parser("list", help="List all profiles")
+    profile_subparsers.add_parser("list", help="List all profiles")
     profile_use = profile_subparsers.add_parser("use", help="Set sticky default profile")
     profile_use.add_argument("profile_name", help="Profile name (or 'default')")
 

hermes_cli/memory_setup.py

@@ -12,6 +12,8 @@ import os
 import sys
 from pathlib import Path
 
+from hermes_constants import get_hermes_home
+
 
 # ---------------------------------------------------------------------------
 # Curses-based interactive picker (same pattern as hermes tools)
@@ -275,7 +277,7 @@ def cmd_setup_provider(provider_name: str) -> None:
         config["memory"] = {}
 
     if hasattr(provider, "post_setup"):
-        hermes_home = str(Path(os.environ.get("HERMES_HOME", os.path.expanduser("~/.hermes"))))
+        hermes_home = str(get_hermes_home())
         provider.post_setup(hermes_home, config)
         return
 
@@ -326,7 +328,7 @@ def cmd_setup(args) -> None:
     # If the provider has a post_setup hook, delegate entirely to it.
     # The hook handles its own config, connection test, and activation.
     if hasattr(provider, "post_setup"):
-        hermes_home = str(Path(os.environ.get("HERMES_HOME", os.path.expanduser("~/.hermes"))))
+        hermes_home = str(get_hermes_home())
         provider.post_setup(hermes_home, config)
         return
 
@@ -336,7 +338,7 @@ def cmd_setup(args) -> None:
     if not isinstance(provider_config, dict):
         provider_config = {}
 
-    env_path = Path(os.environ.get("HERMES_HOME", os.path.expanduser("~/.hermes"))) / ".env"
+    env_path = get_hermes_home() / ".env"
     env_writes = {}
 
     if schema:
@@ -400,7 +402,7 @@ def cmd_setup(args) -> None:
     save_config(config)
 
     # Write non-secret config to provider's native location
-    hermes_home = str(Path(os.environ.get("HERMES_HOME", os.path.expanduser("~/.hermes"))))
+    hermes_home = str(get_hermes_home())
     if provider_config and hasattr(provider, "save_config"):
         try:
             provider.save_config(provider_config, hermes_home)

hermes_cli/model_switch.py

@@ -21,22 +21,16 @@ OpenRouter variant suffixes (``:free``, ``:extended``, ``:fast``).
 from __future__ import annotations
 
 import logging
-from dataclasses import dataclass, field
+from dataclasses import dataclass
 from typing import List, NamedTuple, Optional
 
 from hermes_cli.providers import (
-    ALIASES,
-    LABELS,
-    TRANSPORT_TO_API_MODE,
     determine_api_mode,
     get_label,
-    get_provider,
     is_aggregator,
-    normalize_provider,
     resolve_provider_full,
 )
 from hermes_cli.model_normalize import (
-    detect_vendor,
     normalize_model_for_provider,
 )
 from agent.models_dev import (

hermes_cli/models.py

@@ -44,7 +44,7 @@ OPENROUTER_MODELS: list[tuple[str, str]] = [
     ("stepfun/step-3.5-flash",          ""),
     ("minimax/minimax-m2.7",            ""),
     ("minimax/minimax-m2.5",            ""),
-    ("z-ai/glm-5",                      ""),
+    ("z-ai/glm-5.1",                    ""),
     ("z-ai/glm-5-turbo",                ""),
     ("moonshotai/kimi-k2.5",            ""),
     ("x-ai/grok-4.20-beta",             ""),
@@ -75,7 +75,7 @@ _PROVIDER_MODELS: dict[str, list[str]] = {
         "stepfun/step-3.5-flash",
         "minimax/minimax-m2.7",
         "minimax/minimax-m2.5",
-        "z-ai/glm-5",
+        "z-ai/glm-5.1",
         "z-ai/glm-5-turbo",
         "moonshotai/kimi-k2.5",
         "x-ai/grok-4.20-beta",
@@ -265,6 +265,202 @@ _PROVIDER_MODELS: dict[str, list[str]] = {
     ],
 }
 
+# ---------------------------------------------------------------------------
+# Nous Portal free-model filtering
+# ---------------------------------------------------------------------------
+# Models that are ALLOWED to appear when priced as free on Nous Portal.
+# Any other free model is hidden — prevents promotional/temporary free models
+# from cluttering the selection when users are paying subscribers.
+# Models in this list are ALSO filtered out if they are NOT free (i.e. they
+# should only appear in the menu when they are genuinely free).
+_NOUS_ALLOWED_FREE_MODELS: frozenset[str] = frozenset({
+    "xiaomi/mimo-v2-pro",
+    "xiaomi/mimo-v2-omni",
+})
+
+
+def _is_model_free(model_id: str, pricing: dict[str, dict[str, str]]) -> bool:
+    """Return True if *model_id* has zero-cost prompt AND completion pricing."""
+    p = pricing.get(model_id)
+    if not p:
+        return False
+    try:
+        return float(p.get("prompt", "1")) == 0 and float(p.get("completion", "1")) == 0
+    except (TypeError, ValueError):
+        return False
+
+
+def filter_nous_free_models(
+    model_ids: list[str],
+    pricing: dict[str, dict[str, str]],
+) -> list[str]:
+    """Filter the Nous Portal model list according to free-model policy.
+
+    Rules:
+      • Paid models that are NOT in the allowlist → keep (normal case).
+      • Free models that are NOT in the allowlist → drop.
+      • Allowlist models that ARE free → keep.
+      • Allowlist models that are NOT free → drop.
+    """
+    if not pricing:
+        return model_ids  # no pricing data — can't filter, show everything
+
+    result: list[str] = []
+    for mid in model_ids:
+        free = _is_model_free(mid, pricing)
+        if mid in _NOUS_ALLOWED_FREE_MODELS:
+            # Allowlist model: only show when it's actually free
+            if free:
+                result.append(mid)
+        else:
+            # Regular model: keep only when it's NOT free
+            if not free:
+                result.append(mid)
+    return result
+
+
+# ---------------------------------------------------------------------------
+# Nous Portal account tier detection
+# ---------------------------------------------------------------------------
+
+def fetch_nous_account_tier(access_token: str, portal_base_url: str = "") -> dict[str, Any]:
+    """Fetch the user's Nous Portal account/subscription info.
+
+    Calls ``<portal>/api/oauth/account`` with the OAuth access token.
+
+    Returns the parsed JSON dict on success, e.g.::
+
+        {
+            "subscription": {
+                "plan": "Plus",
+                "tier": 2,
+                "monthly_charge": 20,
+                "credits_remaining": 1686.60,
+                ...
+            },
+            ...
+        }
+
+    Returns an empty dict on any failure (network, auth, parse).
+    """
+    base = (portal_base_url or "https://portal.nousresearch.com").rstrip("/")
+    url = f"{base}/api/oauth/account"
+    headers = {
+        "Authorization": f"Bearer {access_token}",
+        "Accept": "application/json",
+    }
+    try:
+        req = urllib.request.Request(url, headers=headers)
+        with urllib.request.urlopen(req, timeout=8) as resp:
+            return json.loads(resp.read().decode())
+    except Exception:
+        return {}
+
+
+def is_nous_free_tier(account_info: dict[str, Any]) -> bool:
+    """Return True if the account info indicates a free (unpaid) tier.
+
+    Checks ``subscription.monthly_charge == 0``.  Returns False when
+    the field is missing or unparseable (assumes paid — don't block users).
+    """
+    sub = account_info.get("subscription")
+    if not isinstance(sub, dict):
+        return False
+    charge = sub.get("monthly_charge")
+    if charge is None:
+        return False
+    try:
+        return float(charge) == 0
+    except (TypeError, ValueError):
+        return False
+
+
+def partition_nous_models_by_tier(
+    model_ids: list[str],
+    pricing: dict[str, dict[str, str]],
+    free_tier: bool,
+) -> tuple[list[str], list[str]]:
+    """Split Nous models into (selectable, unavailable) based on user tier.
+
+    For paid-tier users: all models are selectable, none unavailable
+    (free-model filtering is handled separately by ``filter_nous_free_models``).
+
+    For free-tier users: only free models are selectable; paid models
+    are returned as unavailable (shown grayed out in the menu).
+    """
+    if not free_tier:
+        return (model_ids, [])
+
+    if not pricing:
+        return (model_ids, [])  # can't determine, show everything
+
+    selectable: list[str] = []
+    unavailable: list[str] = []
+    for mid in model_ids:
+        if _is_model_free(mid, pricing):
+            selectable.append(mid)
+        else:
+            unavailable.append(mid)
+    return (selectable, unavailable)
+
+
+# ---------------------------------------------------------------------------
+# TTL cache for free-tier detection — avoids repeated API calls within a
+# session while still picking up upgrades quickly.
+# ---------------------------------------------------------------------------
+_FREE_TIER_CACHE_TTL: int = 180  # seconds (3 minutes)
+_free_tier_cache: tuple[bool, float] | None = None  # (result, timestamp)
+
+
+def clear_nous_free_tier_cache() -> None:
+    """Invalidate the cached free-tier result (e.g. after login/logout)."""
+    global _free_tier_cache
+    _free_tier_cache = None
+
+
+def check_nous_free_tier() -> bool:
+    """Check if the current Nous Portal user is on a free (unpaid) tier.
+
+    Results are cached for ``_FREE_TIER_CACHE_TTL`` seconds to avoid
+    hitting the Portal API on every call.  The cache is short-lived so
+    that an account upgrade is reflected within a few minutes.
+
+    Returns False (assume paid) on any error — never blocks paying users.
+    """
+    global _free_tier_cache
+    import time
+
+    now = time.monotonic()
+    if _free_tier_cache is not None:
+        cached_result, cached_at = _free_tier_cache
+        if now - cached_at < _FREE_TIER_CACHE_TTL:
+            return cached_result
+
+    try:
+        from hermes_cli.auth import get_provider_auth_state, resolve_nous_runtime_credentials
+
+        # Ensure we have a fresh token (triggers refresh if needed)
+        resolve_nous_runtime_credentials(min_key_ttl_seconds=60)
+
+        state = get_provider_auth_state("nous")
+        if not state:
+            _free_tier_cache = (False, now)
+            return False
+        access_token = state.get("access_token", "")
+        portal_url = state.get("portal_base_url", "")
+        if not access_token:
+            _free_tier_cache = (False, now)
+            return False
+
+        account_info = fetch_nous_account_tier(access_token, portal_url)
+        result = is_nous_free_tier(account_info)
+        _free_tier_cache = (result, now)
+        return result
+    except Exception:
+        _free_tier_cache = (False, now)
+        return False  # default to paid on error — don't block users
+
+
 _PROVIDER_LABELS = {
     "openrouter": "OpenRouter",
     "openai-codex": "OpenAI Codex",
@@ -935,10 +1131,6 @@ def _payload_items(payload: Any) -> list[dict[str, Any]]:
     return []
 
 
-def _extract_model_ids(payload: Any) -> list[str]:
-    return [item.get("id", "") for item in _payload_items(payload) if item.get("id")]
-
-
 def copilot_default_headers() -> dict[str, str]:
     """Standard headers for Copilot API requests.
 

hermes_cli/nous_subscription.py

@@ -167,20 +167,20 @@ def _resolve_browser_feature_state(
     if browser_provider_explicit:
         current_provider = browser_provider or "local"
         if current_provider == "browserbase":
-            provider_available = managed_browser_available or direct_browserbase
+            available = bool(browser_local_available and direct_browserbase)
+            active = bool(browser_tool_enabled and available)
+            return current_provider, available, active, False
+        if current_provider == "browser-use":
+            provider_available = managed_browser_available or direct_browser_use
             available = bool(browser_local_available and provider_available)
             managed = bool(
                 browser_tool_enabled
                 and browser_local_available
                 and managed_browser_available
-                and not direct_browserbase
+                and not direct_browser_use
             )
             active = bool(browser_tool_enabled and available)
             return current_provider, available, active, managed
-        if current_provider == "browser-use":
-            available = bool(browser_local_available and direct_browser_use)
-            active = bool(browser_tool_enabled and available)
-            return current_provider, available, active, False
         if current_provider == "firecrawl":
             available = bool(browser_local_available and direct_firecrawl)
             active = bool(browser_tool_enabled and available)
@@ -193,16 +193,21 @@ def _resolve_browser_feature_state(
         active = bool(browser_tool_enabled and available)
         return current_provider, available, active, False
 
-    if managed_browser_available or direct_browserbase:
+    if managed_browser_available or direct_browser_use:
         available = bool(browser_local_available)
         managed = bool(
             browser_tool_enabled
             and browser_local_available
             and managed_browser_available
-            and not direct_browserbase
+            and not direct_browser_use
         )
         active = bool(browser_tool_enabled and available)
-        return "browserbase", available, active, managed
+        return "browser-use", available, active, managed
+
+    if direct_browserbase:
+        available = bool(browser_local_available)
+        active = bool(browser_tool_enabled and available)
+        return "browserbase", available, active, False
 
     available = bool(browser_local_available)
     active = bool(browser_tool_enabled and available)
@@ -266,7 +271,7 @@ def get_nous_subscription_features(
     managed_web_available = managed_tools_flag and nous_auth_present and is_managed_tool_gateway_ready("firecrawl")
     managed_image_available = managed_tools_flag and nous_auth_present and is_managed_tool_gateway_ready("fal-queue")
     managed_tts_available = managed_tools_flag and nous_auth_present and is_managed_tool_gateway_ready("openai-audio")
-    managed_browser_available = managed_tools_flag and nous_auth_present and is_managed_tool_gateway_ready("browserbase")
+    managed_browser_available = managed_tools_flag and nous_auth_present and is_managed_tool_gateway_ready("browser-use")
     managed_modal_available = managed_tools_flag and nous_auth_present and is_managed_tool_gateway_ready("modal")
     modal_state = resolve_modal_backend_state(
         modal_mode,
@@ -512,10 +517,10 @@ def apply_nous_managed_defaults(
         changed.add("tts")
 
     if "browser" in selected_toolsets and not features.browser.explicit_configured and not (
-        get_env_value("BROWSERBASE_API_KEY")
-        or get_env_value("BROWSER_USE_API_KEY")
+        get_env_value("BROWSER_USE_API_KEY")
+        or get_env_value("BROWSERBASE_API_KEY")
     ):
-        browser_cfg["cloud_provider"] = "browserbase"
+        browser_cfg["cloud_provider"] = "browser-use"
         changed.add("browser")
 
     if "image_gen" in selected_toolsets and not get_env_value("FAL_KEY"):

hermes_cli/plugins.py

@@ -38,6 +38,7 @@ from dataclasses import dataclass, field
 from pathlib import Path
 from typing import Any, Callable, Dict, List, Optional, Set, Union
 
+from hermes_constants import get_hermes_home
 from utils import env_var_enabled
 
 try:
@@ -258,8 +259,7 @@ class PluginManager:
         manifests: List[PluginManifest] = []
 
         # 1. User plugins (~/.hermes/plugins/)
-        hermes_home = os.environ.get("HERMES_HOME", os.path.expanduser("~/.hermes"))
-        user_dir = Path(hermes_home) / "plugins"
+        user_dir = get_hermes_home() / "plugins"
         manifests.extend(self._scan_directory(user_dir, source="user"))
 
         # 2. Project plugins (./.hermes/plugins/)

hermes_cli/plugins_cmd.py

@@ -16,6 +16,8 @@ import subprocess
 import sys
 from pathlib import Path
 
+from hermes_constants import get_hermes_home
+
 logger = logging.getLogger(__name__)
 
 # Minimum manifest version this installer understands.
@@ -26,8 +28,7 @@ _SUPPORTED_MANIFEST_VERSION = 1
 
 def _plugins_dir() -> Path:
     """Return the user plugins directory, creating it if needed."""
-    hermes_home = os.environ.get("HERMES_HOME", os.path.expanduser("~/.hermes"))
-    plugins = Path(hermes_home) / "plugins"
+    plugins = get_hermes_home() / "plugins"
     plugins.mkdir(parents=True, exist_ok=True)
     return plugins
 
@@ -294,7 +295,7 @@ def cmd_install(identifier: str, force: bool = False) -> None:
         sys.exit(1)
 
     # Warn about insecure / local URL schemes
-    if git_url.startswith("http://") or git_url.startswith("file://"):
+    if git_url.startswith(("http://", "file://")):
         console.print(
             "[yellow]Warning:[/yellow] Using insecure/local URL scheme. "
             "Consider using https:// or git@ for production installs."

hermes_cli/profiles.py

@@ -26,7 +26,7 @@ import shutil
 import stat
 import subprocess
 import sys
-from dataclasses import dataclass, field
+from dataclasses import dataclass
 from pathlib import Path, PurePosixPath, PureWindowsPath
 from typing import List, Optional
 
@@ -517,7 +517,6 @@ def delete_profile(name: str, yes: bool = False) -> Path:
     ]
 
     # Check for service
-    from hermes_cli.gateway import _profile_suffix, get_service_name
     wrapper_path = _get_wrapper_dir() / name
     has_wrapper = wrapper_path.exists()
     if has_wrapper:

hermes_cli/providers.py

@@ -20,8 +20,7 @@ Other modules import from this file.  No parallel registries.
 from __future__ import annotations
 
 import logging
-import os
-from dataclasses import dataclass, field
+from dataclasses import dataclass
 from typing import Any, Dict, List, Optional, Tuple
 
 logger = logging.getLogger(__name__)
@@ -345,26 +344,6 @@ def get_label(provider_id: str) -> str:
     return canonical
 
 
-# Build LABELS dict for backward compat
-def _build_labels() -> Dict[str, str]:
-    """Build labels dict from overlays + overrides. Lazy, cached."""
-    labels: Dict[str, str] = {}
-    for pid in HERMES_OVERLAYS:
-        labels[pid] = get_label(pid)
-    labels.update(_LABEL_OVERRIDES)
-    return labels
-
-# Lazy-built on first access
-_labels_cache: Optional[Dict[str, str]] = None
-
-@property
-def LABELS() -> Dict[str, str]:
-    """Backward-compatible labels dict."""
-    global _labels_cache
-    if _labels_cache is None:
-        _labels_cache = _build_labels()
-    return _labels_cache
-
 # For direct import compat, expose as module-level dict
 # Built on demand by get_label() calls
 LABELS: Dict[str, str] = {

hermes_cli/runtime_provider.py

@@ -639,31 +639,47 @@ def resolve_runtime_provider(
             )
 
     if provider == "nous":
-        creds = resolve_nous_runtime_credentials(
-            min_key_ttl_seconds=max(60, int(os.getenv("HERMES_NOUS_MIN_KEY_TTL_SECONDS", "1800"))),
-            timeout_seconds=float(os.getenv("HERMES_NOUS_TIMEOUT_SECONDS", "15")),
-        )
-        return {
-            "provider": "nous",
-            "api_mode": "chat_completions",
-            "base_url": creds.get("base_url", "").rstrip("/"),
-            "api_key": creds.get("api_key", ""),
-            "source": creds.get("source", "portal"),
-            "expires_at": creds.get("expires_at"),
-            "requested_provider": requested_provider,
-        }
+        try:
+            creds = resolve_nous_runtime_credentials(
+                min_key_ttl_seconds=max(60, int(os.getenv("HERMES_NOUS_MIN_KEY_TTL_SECONDS", "1800"))),
+                timeout_seconds=float(os.getenv("HERMES_NOUS_TIMEOUT_SECONDS", "15")),
+            )
+            return {
+                "provider": "nous",
+                "api_mode": "chat_completions",
+                "base_url": creds.get("base_url", "").rstrip("/"),
+                "api_key": creds.get("api_key", ""),
+                "source": creds.get("source", "portal"),
+                "expires_at": creds.get("expires_at"),
+                "requested_provider": requested_provider,
+            }
+        except AuthError:
+            if requested_provider != "auto":
+                raise
+            # Auto-detected Nous but credentials are stale/revoked —
+            # fall through to env-var providers (e.g. OpenRouter).
+            logger.info("Auto-detected Nous provider but credentials failed; "
+                        "falling through to next provider.")
 
     if provider == "openai-codex":
-        creds = resolve_codex_runtime_credentials()
-        return {
-            "provider": "openai-codex",
-            "api_mode": "codex_responses",
-            "base_url": creds.get("base_url", "").rstrip("/"),
-            "api_key": creds.get("api_key", ""),
-            "source": creds.get("source", "hermes-auth-store"),
-            "last_refresh": creds.get("last_refresh"),
-            "requested_provider": requested_provider,
-        }
+        try:
+            creds = resolve_codex_runtime_credentials()
+            return {
+                "provider": "openai-codex",
+                "api_mode": "codex_responses",
+                "base_url": creds.get("base_url", "").rstrip("/"),
+                "api_key": creds.get("api_key", ""),
+                "source": creds.get("source", "hermes-auth-store"),
+                "last_refresh": creds.get("last_refresh"),
+                "requested_provider": requested_provider,
+            }
+        except AuthError:
+            if requested_provider != "auto":
+                raise
+            # Auto-detected Codex but credentials are stale/revoked —
+            # fall through to env-var providers (e.g. OpenRouter).
+            logger.info("Auto-detected Codex provider but credentials failed; "
+                        "falling through to next provider.")
 
     if provider == "copilot-acp":
         creds = resolve_external_process_provider_credentials(provider)

hermes_cli/setup.py

@@ -21,7 +21,6 @@ from typing import Optional, Dict, Any
 
 from hermes_cli.nous_subscription import (
     apply_nous_provider_defaults,
-    get_nous_subscription_explainer_lines,
     get_nous_subscription_features,
 )
 from tools.tool_backend_helpers import managed_nous_tools_enabled
@@ -43,18 +42,6 @@ def _model_config_dict(config: Dict[str, Any]) -> Dict[str, Any]:
     return {}
 
 
-def _set_model_provider(
-    config: Dict[str, Any], provider_id: str, base_url: str = ""
-) -> None:
-    model_cfg = _model_config_dict(config)
-    model_cfg["provider"] = provider_id
-    if base_url:
-        model_cfg["base_url"] = base_url.rstrip("/")
-    else:
-        model_cfg.pop("base_url", None)
-    config["model"] = model_cfg
-
-
 def _set_default_model(config: Dict[str, Any], model_name: str) -> None:
     if not model_name:
         return
@@ -327,16 +314,6 @@ def _setup_provider_model_selection(config, provider_id, current_model, prompt_c
         config["model"] = model_cfg
 
 
-def _sync_model_from_disk(config: Dict[str, Any]) -> None:
-    disk_model = load_config().get("model")
-    if isinstance(disk_model, dict):
-        model_cfg = _model_config_dict(config)
-        model_cfg.update(disk_model)
-        config["model"] = model_cfg
-    elif isinstance(disk_model, str) and disk_model.strip():
-        _set_default_model(config, disk_model.strip())
-
-
 # Import config helpers
 from hermes_cli.config import (
     get_hermes_home,
@@ -660,14 +637,14 @@ def _print_setup_summary(config: dict, hermes_home):
     # Browser tools (local Chromium, Camofox, Browserbase, Browser Use, or Firecrawl)
     browser_provider = subscription_features.browser.current_provider
     if subscription_features.browser.managed_by_nous:
-        tool_status.append(("Browser Automation (Nous Browserbase)", True, None))
+        tool_status.append(("Browser Automation (Nous Browser Use)", True, None))
     elif subscription_features.browser.available:
         label = "Browser Automation"
         if browser_provider:
             label = f"Browser Automation ({browser_provider})"
         tool_status.append((label, True, None))
     else:
-        missing_browser_hint = "npm install -g agent-browser, set CAMOFOX_URL, or configure Browserbase"
+        missing_browser_hint = "npm install -g agent-browser, set CAMOFOX_URL, or configure Browser Use or Browserbase"
         if browser_provider == "Browserbase":
             missing_browser_hint = (
                 "npm install -g agent-browser and set "
@@ -1348,8 +1325,6 @@ def setup_terminal_backend(config: dict):
     terminal_choices.append(f"Keep current ({current_backend})")
     idx_to_backend[keep_current_idx] = current_backend
 
-    default_terminal = backend_to_idx.get(current_backend, 0)
-
     terminal_idx = prompt_choice(
         "Select terminal backend:", terminal_choices, keep_current_idx
     )

hermes_cli/skin_engine.py

@@ -96,7 +96,6 @@ Activate with ``/skin <name>`` in the CLI or ``display.skin: <name>`` in config.
 """
 
 import logging
-import os
 from dataclasses import dataclass, field
 from pathlib import Path
 from typing import Any, Dict, List, Optional, Tuple

hermes_cli/status.py

@@ -123,7 +123,8 @@ def show_status(args):
         "MiniMax-CN": "MINIMAX_CN_API_KEY",
         "Firecrawl": "FIRECRAWL_API_KEY",
         "Tavily": "TAVILY_API_KEY",
-        "Browserbase": "BROWSERBASE_API_KEY",  # Optional — local browser works without this
+        "Browser Use": "BROWSER_USE_API_KEY",  # Optional — local browser works without this
+        "Browserbase": "BROWSERBASE_API_KEY",  # Optional — direct credentials only
         "FAL": "FAL_KEY",
         "Tinker": "TINKER_API_KEY",
         "WandB": "WANDB_API_KEY",

hermes_cli/tools_config.py

@@ -61,22 +61,6 @@ def _prompt(question: str, default: str = None, password: bool = False) -> str:
         print()
         return default or ""
 
-def _prompt_yes_no(question: str, default: bool = True) -> bool:
-    default_str = "Y/n" if default else "y/N"
-    while True:
-        try:
-            value = input(color(f"{question} [{default_str}]: ", Colors.YELLOW)).strip().lower()
-        except (KeyboardInterrupt, EOFError):
-            print()
-            return default
-        if not value:
-            return default
-        if value in ('y', 'yes'):
-            return True
-        if value in ('n', 'no'):
-            return False
-
-
 # ─── Toolset Registry ─────────────────────────────────────────────────────────
 
 # Toolsets shown in the configurator, grouped for display.
@@ -280,21 +264,21 @@ TOOL_CATEGORIES = {
         "icon": "🌐",
         "providers": [
             {
-                "name": "Nous Subscription (Browserbase cloud)",
-                "tag": "Managed Browserbase billed to your subscription",
+                "name": "Nous Subscription (Browser Use cloud)",
+                "tag": "Managed Browser Use billed to your subscription",
                 "env_vars": [],
-                "browser_provider": "browserbase",
+                "browser_provider": "browser-use",
                 "requires_nous_auth": True,
                 "managed_nous_feature": "browser",
-                "override_env_vars": ["BROWSERBASE_API_KEY", "BROWSERBASE_PROJECT_ID"],
-                "post_setup": "browserbase",
+                "override_env_vars": ["BROWSER_USE_API_KEY"],
+                "post_setup": "agent_browser",
             },
             {
                 "name": "Local Browser",
                 "tag": "Free headless Chromium (no API key needed)",
                 "env_vars": [],
                 "browser_provider": "local",
-                "post_setup": "browserbase",  # Same npm install for agent-browser
+                "post_setup": "agent_browser",
             },
             {
                 "name": "Browserbase",
@@ -304,7 +288,7 @@ TOOL_CATEGORIES = {
                     {"key": "BROWSERBASE_PROJECT_ID", "prompt": "Browserbase project ID"},
                 ],
                 "browser_provider": "browserbase",
-                "post_setup": "browserbase",
+                "post_setup": "agent_browser",
             },
             {
                 "name": "Browser Use",
@@ -313,7 +297,7 @@ TOOL_CATEGORIES = {
                     {"key": "BROWSER_USE_API_KEY", "prompt": "Browser Use API key", "url": "https://browser-use.com"},
                 ],
                 "browser_provider": "browser-use",
-                "post_setup": "browserbase",
+                "post_setup": "agent_browser",
             },
             {
                 "name": "Firecrawl",
@@ -322,7 +306,7 @@ TOOL_CATEGORIES = {
                     {"key": "FIRECRAWL_API_KEY", "prompt": "Firecrawl API key", "url": "https://firecrawl.dev"},
                 ],
                 "browser_provider": "firecrawl",
-                "post_setup": "browserbase",
+                "post_setup": "agent_browser",
             },
             {
                 "name": "Camofox",
@@ -381,7 +365,7 @@ TOOLSET_ENV_REQUIREMENTS = {
 def _run_post_setup(post_setup_key: str):
     """Run post-setup hooks for tools that need extra installation steps."""
     import shutil
-    if post_setup_key == "browserbase":
+    if post_setup_key in ("agent_browser", "browserbase"):
         node_modules = PROJECT_ROOT / "node_modules" / "agent-browser"
         if not node_modules.exists() and shutil.which("npm"):
             _print_info("    Installing Node.js dependencies for browser tools...")

hermes_cli/uninstall.py

@@ -6,7 +6,6 @@ Provides options for:
 - Keep data: Remove code but keep ~/.hermes/ (configs, sessions, logs)
 """
 
-import os
 import shutil
 import subprocess
 from pathlib import Path
@@ -24,10 +23,6 @@ def log_success(msg: str):
 def log_warn(msg: str):
     print(f"{color('⚠', Colors.YELLOW)} {msg}")
 
-def log_error(msg: str):
-    print(f"{color('✗', Colors.RED)} {msg}")
-
-
 def get_project_root() -> Path:
     """Get the project installation directory."""
     return Path(__file__).parent.parent.resolve()

hermes_cli/webhook.py

@@ -16,7 +16,7 @@ import re
 import secrets
 import time
 from pathlib import Path
-from typing import Dict, Optional
+from typing import Dict
 
 from hermes_constants import display_hermes_home
 
@@ -25,9 +25,8 @@ _SUBSCRIPTIONS_FILENAME = "webhook_subscriptions.json"
 
 
 def _hermes_home() -> Path:
-    return Path(
-        os.getenv("HERMES_HOME", str(Path.home() / ".hermes"))
-    ).expanduser()
+    from hermes_constants import get_hermes_home
+    return get_hermes_home()
 
 
 def _subscriptions_path() -> Path:

hermes_logging.py

@@ -13,7 +13,6 @@ secrets are never written to disk.
 """
 
 import logging
-import os
 from logging.handlers import RotatingFileHandler
 from pathlib import Path
 from typing import Optional

hermes_state.py

@@ -16,7 +16,6 @@ Key design decisions:
 
 import json
 import logging
-import os
 import random
 import re
 import sqlite3

hermes_time.py

@@ -16,7 +16,6 @@ crashes due to a bad timezone string.
 import logging
 import os
 from datetime import datetime
-from pathlib import Path
 from hermes_constants import get_hermes_home
 from typing import Optional
 
@@ -92,7 +91,6 @@ def get_timezone() -> Optional[ZoneInfo]:
 
 def get_timezone_name() -> str:
     """Return the IANA name of the configured timezone, or empty string."""
-    global _cached_tz_name, _cache_resolved
     if not _cache_resolved:
         get_timezone()  # populates cache
     return _cached_tz_name or ""

mcp_serve.py

@@ -37,9 +37,8 @@ import sys
 import threading
 import time
 from dataclasses import dataclass, field
-from datetime import datetime
 from pathlib import Path
-from typing import Any, Dict, List, Optional
+from typing import Dict, List, Optional
 
 logger = logging.getLogger("hermes.mcp_serve")
 

model_tools.py

@@ -211,7 +211,7 @@ _LEGACY_TOOLSET_MAP = {
     "browser_tools": [
         "browser_navigate", "browser_snapshot", "browser_click",
         "browser_type", "browser_scroll", "browser_back",
-        "browser_press", "browser_close", "browser_get_images",
+        "browser_press", "browser_get_images",
         "browser_vision", "browser_console"
     ],
     "cronjob_tools": ["cronjob"],

plugins/memory/byterover/__init__.py

@@ -23,11 +23,11 @@ import os
 import shutil
 import subprocess
 import threading
-import time
 from pathlib import Path
 from typing import Any, Dict, List, Optional
 
 from agent.memory_provider import MemoryProvider
+from tools.registry import tool_error
 
 logger = logging.getLogger(__name__)
 
@@ -321,7 +321,7 @@ class ByteRoverMemoryProvider(MemoryProvider):
             return self._tool_curate(args)
         elif tool_name == "brv_status":
             return self._tool_status()
-        return json.dumps({"error": f"Unknown tool: {tool_name}"})
+        return tool_error(f"Unknown tool: {tool_name}")
 
     def shutdown(self) -> None:
         if self._sync_thread and self._sync_thread.is_alive():
@@ -332,7 +332,7 @@ class ByteRoverMemoryProvider(MemoryProvider):
     def _tool_query(self, args: dict) -> str:
         query = args.get("query", "")
         if not query:
-            return json.dumps({"error": "query is required"})
+            return tool_error("query is required")
 
         result = _run_brv(
             ["query", "--", query.strip()[:5000]],
@@ -340,7 +340,7 @@ class ByteRoverMemoryProvider(MemoryProvider):
         )
 
         if not result["success"]:
-            return json.dumps({"error": result.get("error", "Query failed")})
+            return tool_error(result.get("error", "Query failed"))
 
         output = result.get("output", "").strip()
         if not output or len(output) < _MIN_OUTPUT_LEN:
@@ -355,7 +355,7 @@ class ByteRoverMemoryProvider(MemoryProvider):
     def _tool_curate(self, args: dict) -> str:
         content = args.get("content", "")
         if not content:
-            return json.dumps({"error": "content is required"})
+            return tool_error("content is required")
 
         result = _run_brv(
             ["curate", "--", content],
@@ -363,14 +363,14 @@ class ByteRoverMemoryProvider(MemoryProvider):
         )
 
         if not result["success"]:
-            return json.dumps({"error": result.get("error", "Curate failed")})
+            return tool_error(result.get("error", "Curate failed"))
 
         return json.dumps({"result": "Memory curated successfully."})
 
     def _tool_status(self) -> str:
         result = _run_brv(["status"], timeout=15, cwd=self._cwd)
         if not result["success"]:
-            return json.dumps({"error": result.get("error", "Status check failed")})
+            return tool_error(result.get("error", "Status check failed"))
         return json.dumps({"status": result.get("output", "")})
 
 

plugins/memory/hindsight/__init__.py

@@ -26,6 +26,7 @@ import threading
 from typing import Any, Dict, List
 
 from agent.memory_provider import MemoryProvider
+from tools.registry import tool_error
 
 logger = logging.getLogger(__name__)
 
@@ -290,8 +291,7 @@ class HindsightMemoryProvider(MemoryProvider):
         if self._mode == "local":
             def _start_daemon():
                 import traceback
-                from pathlib import Path
-                log_dir = Path(os.environ.get("HERMES_HOME", os.path.expanduser("~/.hermes"))) / "logs"
+                log_dir = get_hermes_home() / "logs"
                 log_dir.mkdir(parents=True, exist_ok=True)
                 log_path = log_dir / "hindsight-embed.log"
                 try:
@@ -434,12 +434,12 @@ class HindsightMemoryProvider(MemoryProvider):
             client = self._get_client()
         except Exception as e:
             logger.warning("Hindsight client init failed: %s", e)
-            return json.dumps({"error": f"Hindsight client unavailable: {e}"})
+            return tool_error(f"Hindsight client unavailable: {e}")
 
         if tool_name == "hindsight_retain":
             content = args.get("content", "")
             if not content:
-                return json.dumps({"error": "Missing required parameter: content"})
+                return tool_error("Missing required parameter: content")
             context = args.get("context")
             try:
                 _run_sync(client.aretain(
@@ -448,12 +448,12 @@ class HindsightMemoryProvider(MemoryProvider):
                 return json.dumps({"result": "Memory stored successfully."})
             except Exception as e:
                 logger.warning("hindsight_retain failed: %s", e)
-                return json.dumps({"error": f"Failed to store memory: {e}"})
+                return tool_error(f"Failed to store memory: {e}")
 
         elif tool_name == "hindsight_recall":
             query = args.get("query", "")
             if not query:
-                return json.dumps({"error": "Missing required parameter: query"})
+                return tool_error("Missing required parameter: query")
             try:
                 resp = _run_sync(client.arecall(
                     bank_id=self._bank_id, query=query, budget=self._budget
@@ -464,12 +464,12 @@ class HindsightMemoryProvider(MemoryProvider):
                 return json.dumps({"result": "\n".join(lines)})
             except Exception as e:
                 logger.warning("hindsight_recall failed: %s", e)
-                return json.dumps({"error": f"Failed to search memory: {e}"})
+                return tool_error(f"Failed to search memory: {e}")
 
         elif tool_name == "hindsight_reflect":
             query = args.get("query", "")
             if not query:
-                return json.dumps({"error": "Missing required parameter: query"})
+                return tool_error("Missing required parameter: query")
             try:
                 resp = _run_sync(client.areflect(
                     bank_id=self._bank_id, query=query, budget=self._budget
@@ -477,9 +477,9 @@ class HindsightMemoryProvider(MemoryProvider):
                 return json.dumps({"result": resp.text or "No relevant memories found."})
             except Exception as e:
                 logger.warning("hindsight_reflect failed: %s", e)
-                return json.dumps({"error": f"Failed to reflect: {e}"})
+                return tool_error(f"Failed to reflect: {e}")
 
-        return json.dumps({"error": f"Unknown tool: {tool_name}"})
+        return tool_error(f"Unknown tool: {tool_name}")
 
     def shutdown(self) -> None:
         global _loop, _loop_thread

plugins/memory/holographic/__init__.py

@@ -20,10 +20,10 @@ from __future__ import annotations
 import json
 import logging
 import re
-from pathlib import Path
 from typing import Any, Dict, List
 
 from agent.memory_provider import MemoryProvider
+from tools.registry import tool_error
 from .store import MemoryStore
 from .retrieval import FactRetriever
 
@@ -231,7 +231,7 @@ class HolographicMemoryProvider(MemoryProvider):
             return self._handle_fact_store(args)
         elif tool_name == "fact_feedback":
             return self._handle_fact_feedback(args)
-        return json.dumps({"error": f"Unknown tool: {tool_name}"})
+        return tool_error(f"Unknown tool: {tool_name}")
 
     def on_session_end(self, messages: List[Dict[str, Any]]) -> None:
         if not self._config.get("auto_extract", False):
@@ -297,7 +297,7 @@ class HolographicMemoryProvider(MemoryProvider):
             elif action == "reason":
                 entities = args.get("entities", [])
                 if not entities:
-                    return json.dumps({"error": "reason requires 'entities' list"})
+                    return tool_error("reason requires 'entities' list")
                 results = retriever.reason(
                     entities,
                     category=args.get("category"),
@@ -335,12 +335,12 @@ class HolographicMemoryProvider(MemoryProvider):
                 return json.dumps({"facts": facts, "count": len(facts)})
 
             else:
-                return json.dumps({"error": f"Unknown action: {action}"})
+                return tool_error(f"Unknown action: {action}")
 
         except KeyError as exc:
-            return json.dumps({"error": f"Missing required argument: {exc}"})
+            return tool_error(f"Missing required argument: {exc}")
         except Exception as exc:
-            return json.dumps({"error": str(exc)})
+            return tool_error(str(exc))
 
     def _handle_fact_feedback(self, args: dict) -> str:
         try:
@@ -349,9 +349,9 @@ class HolographicMemoryProvider(MemoryProvider):
             result = self._store.record_feedback(fact_id, helpful=helpful)
             return json.dumps(result)
         except KeyError as exc:
-            return json.dumps({"error": f"Missing required argument: {exc}"})
+            return tool_error(f"Missing required argument: {exc}")
         except Exception as exc:
-            return json.dumps({"error": str(exc)})
+            return tool_error(str(exc))
 
     # -- Auto-extraction (on_session_end) ------------------------------------
 

plugins/memory/holographic/store.py

@@ -6,7 +6,6 @@ Single-user Hermes memory store plugin.
 import re
 import sqlite3
 import threading
-from datetime import datetime
 from pathlib import Path
 
 try:

plugins/memory/honcho/__init__.py

@@ -18,10 +18,10 @@ from __future__ import annotations
 import json
 import logging
 import threading
-from pathlib import Path
 from typing import Any, Dict, List, Optional
 
 from agent.memory_provider import MemoryProvider
+from tools.registry import tool_error
 
 logger = logging.getLogger(__name__)
 
@@ -217,6 +217,12 @@ class HonchoMemoryProvider(MemoryProvider):
                 logger.debug("Honcho not configured — plugin inactive")
                 return
 
+            # Override peer_name with gateway user_id for per-user memory scoping.
+            # CLI sessions won't have user_id, so the config default is preserved.
+            _gw_user_id = kwargs.get("user_id")
+            if _gw_user_id:
+                cfg.peer_name = _gw_user_id
+
             self._config = cfg
 
             # ----- B1: recall_mode from config -----
@@ -633,15 +639,15 @@ class HonchoMemoryProvider(MemoryProvider):
     def handle_tool_call(self, tool_name: str, args: dict, **kwargs) -> str:
         """Handle a Honcho tool call, with lazy session init for tools-only mode."""
         if self._cron_skipped:
-            return json.dumps({"error": "Honcho is not active (cron context)."})
+            return tool_error("Honcho is not active (cron context).")
 
         # Port #1957: ensure session is initialized for tools-only mode
         if not self._session_initialized:
             if not self._ensure_session():
-                return json.dumps({"error": "Honcho session could not be initialized."})
+                return tool_error("Honcho session could not be initialized.")
 
         if not self._manager or not self._session_key:
-            return json.dumps({"error": "Honcho is not active for this session."})
+            return tool_error("Honcho is not active for this session.")
 
         try:
             if tool_name == "honcho_profile":
@@ -653,7 +659,7 @@ class HonchoMemoryProvider(MemoryProvider):
             elif tool_name == "honcho_search":
                 query = args.get("query", "")
                 if not query:
-                    return json.dumps({"error": "Missing required parameter: query"})
+                    return tool_error("Missing required parameter: query")
                 max_tokens = min(int(args.get("max_tokens", 800)), 2000)
                 result = self._manager.search_context(
                     self._session_key, query, max_tokens=max_tokens
@@ -665,7 +671,7 @@ class HonchoMemoryProvider(MemoryProvider):
             elif tool_name == "honcho_context":
                 query = args.get("query", "")
                 if not query:
-                    return json.dumps({"error": "Missing required parameter: query"})
+                    return tool_error("Missing required parameter: query")
                 peer = args.get("peer", "user")
                 result = self._manager.dialectic_query(
                     self._session_key, query, peer=peer
@@ -675,17 +681,17 @@ class HonchoMemoryProvider(MemoryProvider):
             elif tool_name == "honcho_conclude":
                 conclusion = args.get("conclusion", "")
                 if not conclusion:
-                    return json.dumps({"error": "Missing required parameter: conclusion"})
+                    return tool_error("Missing required parameter: conclusion")
                 ok = self._manager.create_conclusion(self._session_key, conclusion)
                 if ok:
                     return json.dumps({"result": f"Conclusion saved: {conclusion}"})
-                return json.dumps({"error": "Failed to save conclusion."})
+                return tool_error("Failed to save conclusion.")
 
-            return json.dumps({"error": f"Unknown tool: {tool_name}"})
+            return tool_error(f"Unknown tool: {tool_name}")
 
         except Exception as e:
             logger.error("Honcho tool %s failed: %s", tool_name, e)
-            return json.dumps({"error": f"Honcho {tool_name} failed: {e}"})
+            return tool_error(f"Honcho {tool_name} failed: {e}")
 
     def shutdown(self) -> None:
         for t in (self._prefetch_thread, self._sync_thread):

plugins/memory/honcho/cli.py

@@ -11,7 +11,7 @@ import sys
 from pathlib import Path
 
 from hermes_constants import get_hermes_home
-from plugins.memory.honcho.client import resolve_active_host, resolve_config_path, GLOBAL_CONFIG_PATH, HOST
+from plugins.memory.honcho.client import resolve_active_host, resolve_config_path, HOST
 
 
 def clone_honcho_for_profile(profile_name: str) -> bool:
@@ -1220,7 +1220,6 @@ def register_cli(subparser) -> None:
     Called by the plugin CLI registration system during argparse setup.
     The *subparser* is the parser for ``hermes honcho``.
     """
-    import argparse
 
     subparser.add_argument(
         "--target-profile", metavar="NAME", dest="target_profile",

plugins/memory/mem0/__init__.py

@@ -20,10 +20,10 @@ import logging
 import os
 import threading
 import time
-from pathlib import Path
 from typing import Any, Dict, List
 
 from agent.memory_provider import MemoryProvider
+from tools.registry import tool_error
 
 logger = logging.getLogger(__name__)
 
@@ -203,7 +203,9 @@ class Mem0MemoryProvider(MemoryProvider):
     def initialize(self, session_id: str, **kwargs) -> None:
         self._config = _load_config()
         self._api_key = self._config.get("api_key", "")
-        self._user_id = self._config.get("user_id", "hermes-user")
+        # Prefer gateway-provided user_id for per-user memory scoping;
+        # fall back to config/env default for CLI (single-user) sessions.
+        self._user_id = kwargs.get("user_id") or self._config.get("user_id", "hermes-user")
         self._agent_id = self._config.get("agent_id", "hermes")
         self._rerank = self._config.get("rerank", True)
 
@@ -304,7 +306,7 @@ class Mem0MemoryProvider(MemoryProvider):
         try:
             client = self._get_client()
         except Exception as e:
-            return json.dumps({"error": str(e)})
+            return tool_error(str(e))
 
         if tool_name == "mem0_profile":
             try:
@@ -316,12 +318,12 @@ class Mem0MemoryProvider(MemoryProvider):
                 return json.dumps({"result": "\n".join(lines), "count": len(lines)})
             except Exception as e:
                 self._record_failure()
-                return json.dumps({"error": f"Failed to fetch profile: {e}"})
+                return tool_error(f"Failed to fetch profile: {e}")
 
         elif tool_name == "mem0_search":
             query = args.get("query", "")
             if not query:
-                return json.dumps({"error": "Missing required parameter: query"})
+                return tool_error("Missing required parameter: query")
             rerank = args.get("rerank", False)
             top_k = min(int(args.get("top_k", 10)), 50)
             try:
@@ -338,12 +340,12 @@ class Mem0MemoryProvider(MemoryProvider):
                 return json.dumps({"results": items, "count": len(items)})
             except Exception as e:
                 self._record_failure()
-                return json.dumps({"error": f"Search failed: {e}"})
+                return tool_error(f"Search failed: {e}")
 
         elif tool_name == "mem0_conclude":
             conclusion = args.get("conclusion", "")
             if not conclusion:
-                return json.dumps({"error": "Missing required parameter: conclusion"})
+                return tool_error("Missing required parameter: conclusion")
             try:
                 client.add(
                     [{"role": "user", "content": conclusion}],
@@ -354,9 +356,9 @@ class Mem0MemoryProvider(MemoryProvider):
                 return json.dumps({"result": "Fact stored."})
             except Exception as e:
                 self._record_failure()
-                return json.dumps({"error": f"Failed to store: {e}"})
+                return tool_error(f"Failed to store: {e}")
 
-        return json.dumps({"error": f"Unknown tool: {tool_name}"})
+        return tool_error(f"Unknown tool: {tool_name}")
 
     def shutdown(self) -> None:
         for t in (self._prefetch_thread, self._sync_thread):

plugins/memory/openviking/__init__.py

@@ -31,6 +31,7 @@ import threading
 from typing import Any, Dict, List, Optional
 
 from agent.memory_provider import MemoryProvider
+from tools.registry import tool_error
 
 logger = logging.getLogger(__name__)
 
@@ -461,7 +462,7 @@ class OpenVikingMemoryProvider(MemoryProvider):
 
     def handle_tool_call(self, tool_name: str, args: dict, **kwargs) -> str:
         if not self._client:
-            return json.dumps({"error": "OpenViking server not connected"})
+            return tool_error("OpenViking server not connected")
 
         try:
             if tool_name == "viking_search":
@@ -474,9 +475,9 @@ class OpenVikingMemoryProvider(MemoryProvider):
                 return self._tool_remember(args)
             elif tool_name == "viking_add_resource":
                 return self._tool_add_resource(args)
-            return json.dumps({"error": f"Unknown tool: {tool_name}"})
+            return tool_error(f"Unknown tool: {tool_name}")
         except Exception as e:
-            return json.dumps({"error": str(e)})
+            return tool_error(str(e))
 
     def shutdown(self) -> None:
         # Wait for background threads to finish
@@ -493,7 +494,7 @@ class OpenVikingMemoryProvider(MemoryProvider):
     def _tool_search(self, args: dict) -> str:
         query = args.get("query", "")
         if not query:
-            return json.dumps({"error": "query is required"})
+            return tool_error("query is required")
 
         payload: Dict[str, Any] = {"query": query}
         mode = args.get("mode", "auto")
@@ -530,7 +531,7 @@ class OpenVikingMemoryProvider(MemoryProvider):
     def _tool_read(self, args: dict) -> str:
         uri = args.get("uri", "")
         if not uri:
-            return json.dumps({"error": "uri is required"})
+            return tool_error("uri is required")
 
         level = args.get("level", "overview")
         # Map our level names to OpenViking GET endpoints
@@ -582,7 +583,7 @@ class OpenVikingMemoryProvider(MemoryProvider):
     def _tool_remember(self, args: dict) -> str:
         content = args.get("content", "")
         if not content:
-            return json.dumps({"error": "content is required"})
+            return tool_error("content is required")
 
         # Store as a session message that will be extracted during commit.
         # The category hint helps OpenViking's extraction classify correctly.
@@ -606,7 +607,7 @@ class OpenVikingMemoryProvider(MemoryProvider):
     def _tool_add_resource(self, args: dict) -> str:
         url = args.get("url", "")
         if not url:
-            return json.dumps({"error": "url is required"})
+            return tool_error("url is required")
 
         payload: Dict[str, Any] = {"path": url}
         if args.get("reason"):

plugins/memory/retaindb/__init__.py

@@ -20,7 +20,6 @@ Config (env vars or hermes config.yaml under retaindb:):
 
 from __future__ import annotations
 
-import hashlib
 import json
 import logging
 import os
@@ -35,6 +34,7 @@ from typing import Any, Dict, List
 from urllib.parse import quote
 
 from agent.memory_provider import MemoryProvider
+from tools.registry import tool_error
 
 logger = logging.getLogger(__name__)
 
@@ -189,7 +189,7 @@ class _Client:
             "Content-Type": "application/json",
             "x-sdk-runtime": "hermes-plugin",
         }
-        if path.startswith("/v1/memory") or path.startswith("/v1/context"):
+        if path.startswith(("/v1/memory", "/v1/context")):
             h["X-API-Key"] = token
         return h
 
@@ -505,7 +505,8 @@ class RetainDBMemoryProvider(MemoryProvider):
         self._user_id = kwargs.get("user_id", "default") or "default"
         self._agent_id = kwargs.get("agent_id", "hermes") or "hermes"
 
-        hermes_home_path = Path(os.environ.get("HERMES_HOME", Path.home() / ".hermes"))
+        from hermes_constants import get_hermes_home
+        hermes_home_path = get_hermes_home()
         db_path = hermes_home_path / "retaindb_queue.db"
         self._queue = _WriteQueue(self._client, db_path)
 
@@ -649,11 +650,11 @@ class RetainDBMemoryProvider(MemoryProvider):
 
     def handle_tool_call(self, tool_name: str, args: dict, **kwargs) -> str:
         if not self._client:
-            return json.dumps({"error": "RetainDB not initialized"})
+            return tool_error("RetainDB not initialized")
         try:
             return json.dumps(self._dispatch(tool_name, args))
         except Exception as exc:
-            return json.dumps({"error": str(exc)})
+            return tool_error(str(exc))
 
     def _dispatch(self, tool_name: str, args: dict) -> Any:
         c = self._client

plugins/memory/supermemory/README.md

@@ -0,0 +1,54 @@
+# Supermemory Memory Provider
+
+Semantic long-term memory with profile recall, semantic search, explicit memory tools, and session-end conversation ingest.
+
+## Requirements
+
+- `pip install supermemory`
+- Supermemory API key from [supermemory.ai](https://supermemory.ai)
+
+## Setup
+
+```bash
+hermes memory setup    # select "supermemory"
+```
+
+Or manually:
+
+```bash
+hermes config set memory.provider supermemory
+echo 'SUPERMEMORY_API_KEY=your-key-here' >> ~/.hermes/.env
+```
+
+## Config
+
+Config file: `$HERMES_HOME/supermemory.json`
+
+| Key | Default | Description |
+|-----|---------|-------------|
+| `container_tag` | `hermes` | Container tag used for search and writes |
+| `auto_recall` | `true` | Inject relevant memory context before turns |
+| `auto_capture` | `true` | Store cleaned user-assistant turns after each response |
+| `max_recall_results` | `10` | Max recalled items to format into context |
+| `profile_frequency` | `50` | Include profile facts on first turn and every N turns |
+| `capture_mode` | `all` | Skip tiny or trivial turns by default |
+| `entity_context` | built-in default | Extraction guidance passed to Supermemory |
+| `api_timeout` | `5.0` | Timeout for SDK and ingest requests |
+
+## Tools
+
+| Tool | Description |
+|------|-------------|
+| `supermemory_store` | Store an explicit memory |
+| `supermemory_search` | Search memories by semantic similarity |
+| `supermemory_forget` | Forget a memory by ID or best-match query |
+| `supermemory_profile` | Retrieve persistent profile and recent context |
+
+## Behavior
+
+When enabled, Hermes can:
+
+- prefetch relevant memory context before each turn
+- store cleaned conversation turns after each completed response
+- ingest the full session on session end for richer graph updates
+- expose explicit tools for search, store, forget, and profile access

plugins/memory/supermemory/__init__.py

@@ -0,0 +1,672 @@
+"""Supermemory memory plugin using the MemoryProvider interface.
+
+Provides semantic long-term memory with profile recall, semantic search,
+explicit memory tools, cleaned turn capture, and session-end conversation ingest.
+"""
+
+from __future__ import annotations
+
+import json
+import logging
+import os
+import re
+import threading
+import urllib.error
+import urllib.request
+from datetime import datetime, timezone
+from pathlib import Path
+from typing import Any, Dict, List, Optional
+
+from agent.memory_provider import MemoryProvider
+from tools.registry import tool_error
+
+logger = logging.getLogger(__name__)
+
+_DEFAULT_CONTAINER_TAG = "hermes"
+_DEFAULT_MAX_RECALL_RESULTS = 10
+_DEFAULT_PROFILE_FREQUENCY = 50
+_DEFAULT_CAPTURE_MODE = "all"
+_DEFAULT_API_TIMEOUT = 5.0
+_MIN_CAPTURE_LENGTH = 10
+_MAX_ENTITY_CONTEXT_LENGTH = 1500
+_CONVERSATIONS_URL = "https://api.supermemory.ai/v4/conversations"
+_TRIVIAL_RE = re.compile(
+    r"^(ok|okay|thanks|thank you|got it|sure|yes|no|yep|nope|k|ty|thx|np)\.?$",
+    re.IGNORECASE,
+)
+_CONTEXT_STRIP_RE = re.compile(
+    r"<supermemory-context>[\s\S]*?</supermemory-context>\s*", re.DOTALL
+)
+_CONTAINERS_STRIP_RE = re.compile(
+    r"<supermemory-containers>[\s\S]*?</supermemory-containers>\s*", re.DOTALL
+)
+_DEFAULT_ENTITY_CONTEXT = (
+    "User-assistant conversation. Format: [role: user]...[user:end] and "
+    "[role: assistant]...[assistant:end].\n\n"
+    "Only extract things useful in future conversations. Most messages are not worth remembering.\n\n"
+    "Remember lasting personal facts, preferences, routines, tools, ongoing projects, working context, "
+    "and explicit requests to remember something.\n\n"
+    "Do not remember temporary intents, one-time tasks, assistant actions, implementation details, or in-progress status.\n\n"
+    "When in doubt, store less."
+)
+
+
+def _default_config() -> dict:
+    return {
+        "container_tag": _DEFAULT_CONTAINER_TAG,
+        "auto_recall": True,
+        "auto_capture": True,
+        "max_recall_results": _DEFAULT_MAX_RECALL_RESULTS,
+        "profile_frequency": _DEFAULT_PROFILE_FREQUENCY,
+        "capture_mode": _DEFAULT_CAPTURE_MODE,
+        "entity_context": _DEFAULT_ENTITY_CONTEXT,
+        "api_timeout": _DEFAULT_API_TIMEOUT,
+    }
+
+
+def _sanitize_tag(raw: str) -> str:
+    tag = re.sub(r"[^a-zA-Z0-9_]", "_", raw or "")
+    tag = re.sub(r"_+", "_", tag)
+    return tag.strip("_") or _DEFAULT_CONTAINER_TAG
+
+
+def _clamp_entity_context(text: str) -> str:
+    if not text:
+        return _DEFAULT_ENTITY_CONTEXT
+    text = text.strip()
+    return text[:_MAX_ENTITY_CONTEXT_LENGTH]
+
+
+def _as_bool(value: Any, default: bool) -> bool:
+    if isinstance(value, bool):
+        return value
+    if isinstance(value, str):
+        lowered = value.strip().lower()
+        if lowered in ("true", "1", "yes", "y", "on"):
+            return True
+        if lowered in ("false", "0", "no", "n", "off"):
+            return False
+    return default
+
+
+def _load_supermemory_config(hermes_home: str) -> dict:
+    config = _default_config()
+    config_path = Path(hermes_home) / "supermemory.json"
+    if config_path.exists():
+        try:
+            raw = json.loads(config_path.read_text(encoding="utf-8"))
+            if isinstance(raw, dict):
+                config.update({k: v for k, v in raw.items() if v is not None})
+        except Exception:
+            logger.debug("Failed to parse %s", config_path, exc_info=True)
+
+    config["container_tag"] = _sanitize_tag(str(config.get("container_tag", _DEFAULT_CONTAINER_TAG)))
+    config["auto_recall"] = _as_bool(config.get("auto_recall"), True)
+    config["auto_capture"] = _as_bool(config.get("auto_capture"), True)
+    try:
+        config["max_recall_results"] = max(1, min(20, int(config.get("max_recall_results", _DEFAULT_MAX_RECALL_RESULTS))))
+    except Exception:
+        config["max_recall_results"] = _DEFAULT_MAX_RECALL_RESULTS
+    try:
+        config["profile_frequency"] = max(1, min(500, int(config.get("profile_frequency", _DEFAULT_PROFILE_FREQUENCY))))
+    except Exception:
+        config["profile_frequency"] = _DEFAULT_PROFILE_FREQUENCY
+    config["capture_mode"] = "everything" if config.get("capture_mode") == "everything" else "all"
+    config["entity_context"] = _clamp_entity_context(str(config.get("entity_context", _DEFAULT_ENTITY_CONTEXT)))
+    try:
+        config["api_timeout"] = max(0.5, min(15.0, float(config.get("api_timeout", _DEFAULT_API_TIMEOUT))))
+    except Exception:
+        config["api_timeout"] = _DEFAULT_API_TIMEOUT
+    return config
+
+
+def _save_supermemory_config(values: dict, hermes_home: str) -> None:
+    config_path = Path(hermes_home) / "supermemory.json"
+    existing = {}
+    if config_path.exists():
+        try:
+            raw = json.loads(config_path.read_text(encoding="utf-8"))
+            if isinstance(raw, dict):
+                existing = raw
+        except Exception:
+            existing = {}
+    existing.update(values)
+    config_path.write_text(json.dumps(existing, indent=2, sort_keys=True) + "\n", encoding="utf-8")
+
+
+def _detect_category(text: str) -> str:
+    lowered = text.lower()
+    if re.search(r"prefer|like|love|hate|want", lowered):
+        return "preference"
+    if re.search(r"decided|will use|going with", lowered):
+        return "decision"
+    if re.search(r"\bis\b|\bare\b|\bhas\b|\bhave\b", lowered):
+        return "fact"
+    return "other"
+
+
+def _format_relative_time(iso_timestamp: str) -> str:
+    try:
+        dt = datetime.fromisoformat(iso_timestamp.replace("Z", "+00:00"))
+        now = datetime.now(timezone.utc)
+        seconds = (now - dt).total_seconds()
+        if seconds < 1800:
+            return "just now"
+        if seconds < 3600:
+            return f"{int(seconds / 60)}m ago"
+        if seconds < 86400:
+            return f"{int(seconds / 3600)}h ago"
+        if seconds < 604800:
+            return f"{int(seconds / 86400)}d ago"
+        if dt.year == now.year:
+            return dt.strftime("%d %b")
+        return dt.strftime("%d %b %Y")
+    except Exception:
+        return ""
+
+
+def _deduplicate_recall(static_facts: list, dynamic_facts: list, search_results: list) -> tuple[list, list, list]:
+    seen = set()
+    out_static, out_dynamic, out_search = [], [], []
+    for fact in static_facts or []:
+        if fact and fact not in seen:
+            seen.add(fact)
+            out_static.append(fact)
+    for fact in dynamic_facts or []:
+        if fact and fact not in seen:
+            seen.add(fact)
+            out_dynamic.append(fact)
+    for item in search_results or []:
+        memory = item.get("memory", "")
+        if memory and memory not in seen:
+            seen.add(memory)
+            out_search.append(item)
+    return out_static, out_dynamic, out_search
+
+
+def _format_prefetch_context(static_facts: list, dynamic_facts: list, search_results: list, max_results: int) -> str:
+    statics, dynamics, search = _deduplicate_recall(static_facts, dynamic_facts, search_results)
+    statics = statics[:max_results]
+    dynamics = dynamics[:max_results]
+    search = search[:max_results]
+    if not statics and not dynamics and not search:
+        return ""
+
+    sections = []
+    if statics:
+        sections.append("## User Profile (Persistent)\n" + "\n".join(f"- {item}" for item in statics))
+    if dynamics:
+        sections.append("## Recent Context\n" + "\n".join(f"- {item}" for item in dynamics))
+    if search:
+        lines = []
+        for item in search:
+            memory = item.get("memory", "")
+            if not memory:
+                continue
+            similarity = item.get("similarity")
+            updated = item.get("updated_at") or item.get("updatedAt") or ""
+            prefix_bits = []
+            rel = _format_relative_time(updated)
+            if rel:
+                prefix_bits.append(f"[{rel}]")
+            if similarity is not None:
+                try:
+                    prefix_bits.append(f"[{round(float(similarity) * 100)}%]")
+                except Exception:
+                    pass
+            prefix = " ".join(prefix_bits)
+            lines.append(f"- {prefix} {memory}".strip())
+        if lines:
+            sections.append("## Relevant Memories\n" + "\n".join(lines))
+    if not sections:
+        return ""
+
+    intro = (
+        "The following is background context from long-term memory. Use it silently when relevant. "
+        "Do not force memories into the conversation."
+    )
+    body = "\n\n".join(sections)
+    return f"<supermemory-context>\n{intro}\n\n{body}\n</supermemory-context>"
+
+
+def _clean_text_for_capture(text: str) -> str:
+    text = _CONTEXT_STRIP_RE.sub("", text or "")
+    text = _CONTAINERS_STRIP_RE.sub("", text)
+    return text.strip()
+
+
+def _is_trivial_message(text: str) -> bool:
+    return bool(_TRIVIAL_RE.match((text or "").strip()))
+
+
+class _SupermemoryClient:
+    def __init__(self, api_key: str, timeout: float, container_tag: str):
+        from supermemory import Supermemory
+
+        self._api_key = api_key
+        self._container_tag = container_tag
+        self._timeout = timeout
+        self._client = Supermemory(api_key=api_key, timeout=timeout, max_retries=0)
+
+    def add_memory(self, content: str, metadata: Optional[dict] = None, *, entity_context: str = "") -> dict:
+        kwargs = {
+            "content": content.strip(),
+            "container_tags": [self._container_tag],
+        }
+        if metadata:
+            kwargs["metadata"] = metadata
+        if entity_context:
+            kwargs["entity_context"] = _clamp_entity_context(entity_context)
+        result = self._client.documents.add(**kwargs)
+        return {"id": getattr(result, "id", "")}
+
+    def search_memories(self, query: str, *, limit: int = 5) -> list[dict]:
+        response = self._client.search.memories(q=query, container_tag=self._container_tag, limit=limit)
+        results = []
+        for item in (getattr(response, "results", None) or []):
+            results.append({
+                "id": getattr(item, "id", ""),
+                "memory": getattr(item, "memory", "") or "",
+                "similarity": getattr(item, "similarity", None),
+                "updated_at": getattr(item, "updated_at", None) or getattr(item, "updatedAt", None),
+                "metadata": getattr(item, "metadata", None),
+            })
+        return results
+
+    def get_profile(self, query: Optional[str] = None) -> dict:
+        kwargs = {"container_tag": self._container_tag}
+        if query:
+            kwargs["q"] = query
+        response = self._client.profile(**kwargs)
+        profile_data = getattr(response, "profile", None)
+        search_data = getattr(response, "search_results", None) or getattr(response, "searchResults", None)
+        static = getattr(profile_data, "static", []) or [] if profile_data else []
+        dynamic = getattr(profile_data, "dynamic", []) or [] if profile_data else []
+        raw_results = getattr(search_data, "results", None) or search_data or []
+        search_results = []
+        if isinstance(raw_results, list):
+            for item in raw_results:
+                if isinstance(item, dict):
+                    search_results.append(item)
+                else:
+                    search_results.append({
+                        "memory": getattr(item, "memory", ""),
+                        "updated_at": getattr(item, "updated_at", None) or getattr(item, "updatedAt", None),
+                        "similarity": getattr(item, "similarity", None),
+                    })
+        return {"static": static, "dynamic": dynamic, "search_results": search_results}
+
+    def forget_memory(self, memory_id: str) -> None:
+        self._client.memories.forget(container_tag=self._container_tag, id=memory_id)
+
+    def forget_by_query(self, query: str) -> dict:
+        results = self.search_memories(query, limit=5)
+        if not results:
+            return {"success": False, "message": "No matching memory found to forget."}
+        target = results[0]
+        memory_id = target.get("id", "")
+        if not memory_id:
+            return {"success": False, "message": "Best matching memory has no id."}
+        self.forget_memory(memory_id)
+        preview = (target.get("memory") or "")[:100]
+        return {"success": True, "message": f'Forgot: "{preview}"', "id": memory_id}
+
+    def ingest_conversation(self, session_id: str, messages: list[dict]) -> None:
+        payload = json.dumps({
+            "conversationId": session_id,
+            "messages": messages,
+            "containerTags": [self._container_tag],
+        }).encode("utf-8")
+        req = urllib.request.Request(
+            _CONVERSATIONS_URL,
+            data=payload,
+            headers={
+                "Authorization": f"Bearer {self._api_key}",
+                "Content-Type": "application/json",
+            },
+            method="POST",
+        )
+        with urllib.request.urlopen(req, timeout=self._timeout + 3):
+            return
+
+
+STORE_SCHEMA = {
+    "name": "supermemory_store",
+    "description": "Store an explicit memory for future recall.",
+    "parameters": {
+        "type": "object",
+        "properties": {
+            "content": {"type": "string", "description": "The memory content to store."},
+            "metadata": {"type": "object", "description": "Optional metadata attached to the memory."},
+        },
+        "required": ["content"],
+    },
+}
+
+SEARCH_SCHEMA = {
+    "name": "supermemory_search",
+    "description": "Search long-term memory by semantic similarity.",
+    "parameters": {
+        "type": "object",
+        "properties": {
+            "query": {"type": "string", "description": "What to search for."},
+            "limit": {"type": "integer", "description": "Maximum results to return, 1 to 20."},
+        },
+        "required": ["query"],
+    },
+}
+
+FORGET_SCHEMA = {
+    "name": "supermemory_forget",
+    "description": "Forget a memory by exact id or by best-match query.",
+    "parameters": {
+        "type": "object",
+        "properties": {
+            "id": {"type": "string", "description": "Exact memory id to delete."},
+            "query": {"type": "string", "description": "Query used to find the memory to forget."},
+        },
+    },
+}
+
+PROFILE_SCHEMA = {
+    "name": "supermemory_profile",
+    "description": "Retrieve persistent profile facts and recent memory context.",
+    "parameters": {
+        "type": "object",
+        "properties": {
+            "query": {"type": "string", "description": "Optional query to focus the profile response."},
+        },
+    },
+}
+
+
+class SupermemoryMemoryProvider(MemoryProvider):
+    def __init__(self):
+        self._config = _default_config()
+        self._api_key = ""
+        self._client: Optional[_SupermemoryClient] = None
+        self._container_tag = _DEFAULT_CONTAINER_TAG
+        self._session_id = ""
+        self._turn_count = 0
+        self._prefetch_result = ""
+        self._prefetch_lock = threading.Lock()
+        self._prefetch_thread: Optional[threading.Thread] = None
+        self._sync_thread: Optional[threading.Thread] = None
+        self._write_thread: Optional[threading.Thread] = None
+        self._auto_recall = True
+        self._auto_capture = True
+        self._max_recall_results = _DEFAULT_MAX_RECALL_RESULTS
+        self._profile_frequency = _DEFAULT_PROFILE_FREQUENCY
+        self._capture_mode = _DEFAULT_CAPTURE_MODE
+        self._entity_context = _DEFAULT_ENTITY_CONTEXT
+        self._api_timeout = _DEFAULT_API_TIMEOUT
+        self._hermes_home = ""
+        self._write_enabled = True
+        self._active = False
+
+    @property
+    def name(self) -> str:
+        return "supermemory"
+
+    def is_available(self) -> bool:
+        api_key = os.environ.get("SUPERMEMORY_API_KEY", "")
+        if not api_key:
+            return False
+        try:
+            __import__("supermemory")
+            return True
+        except Exception:
+            return False
+
+    def get_config_schema(self):
+        return [
+            {"key": "api_key", "description": "Supermemory API key", "secret": True, "required": True, "env_var": "SUPERMEMORY_API_KEY", "url": "https://supermemory.ai"},
+            {"key": "container_tag", "description": "Container tag for reads and writes", "default": _DEFAULT_CONTAINER_TAG},
+            {"key": "auto_recall", "description": "Enable automatic recall before each turn", "default": "true", "choices": ["true", "false"]},
+            {"key": "auto_capture", "description": "Enable automatic capture after each completed turn", "default": "true", "choices": ["true", "false"]},
+            {"key": "max_recall_results", "description": "Maximum recalled items to inject", "default": str(_DEFAULT_MAX_RECALL_RESULTS)},
+            {"key": "profile_frequency", "description": "Include profile facts on first turn and every N turns", "default": str(_DEFAULT_PROFILE_FREQUENCY)},
+            {"key": "capture_mode", "description": "Capture mode", "default": _DEFAULT_CAPTURE_MODE, "choices": ["all", "everything"]},
+            {"key": "entity_context", "description": "Extraction guidance passed to Supermemory", "default": _DEFAULT_ENTITY_CONTEXT},
+            {"key": "api_timeout", "description": "Timeout in seconds for SDK and ingest calls", "default": str(_DEFAULT_API_TIMEOUT)},
+        ]
+
+    def save_config(self, values, hermes_home):
+        sanitized = dict(values or {})
+        if "container_tag" in sanitized:
+            sanitized["container_tag"] = _sanitize_tag(str(sanitized["container_tag"]))
+        if "entity_context" in sanitized:
+            sanitized["entity_context"] = _clamp_entity_context(str(sanitized["entity_context"]))
+        _save_supermemory_config(sanitized, hermes_home)
+
+    def initialize(self, session_id: str, **kwargs) -> None:
+        from hermes_constants import get_hermes_home
+        self._hermes_home = kwargs.get("hermes_home") or str(get_hermes_home())
+        self._session_id = session_id
+        self._turn_count = 0
+        self._config = _load_supermemory_config(self._hermes_home)
+        self._api_key = os.environ.get("SUPERMEMORY_API_KEY", "")
+        self._container_tag = self._config["container_tag"]
+        self._auto_recall = self._config["auto_recall"]
+        self._auto_capture = self._config["auto_capture"]
+        self._max_recall_results = self._config["max_recall_results"]
+        self._profile_frequency = self._config["profile_frequency"]
+        self._capture_mode = self._config["capture_mode"]
+        self._entity_context = self._config["entity_context"]
+        self._api_timeout = self._config["api_timeout"]
+        agent_context = kwargs.get("agent_context", "")
+        self._write_enabled = agent_context not in ("cron", "flush", "subagent")
+        self._active = bool(self._api_key)
+        self._client = None
+        if self._active:
+            try:
+                self._client = _SupermemoryClient(
+                    api_key=self._api_key,
+                    timeout=self._api_timeout,
+                    container_tag=self._container_tag,
+                )
+            except Exception:
+                logger.warning("Supermemory initialization failed", exc_info=True)
+                self._active = False
+                self._client = None
+
+    def on_turn_start(self, turn_number: int, message: str, **kwargs) -> None:
+        self._turn_count = max(turn_number, 0)
+
+    def system_prompt_block(self) -> str:
+        if not self._active:
+            return ""
+        return (
+            "# Supermemory\n"
+            f"Active. Container: {self._container_tag}.\n"
+            "Use supermemory_search, supermemory_store, supermemory_forget, and supermemory_profile for explicit memory operations."
+        )
+
+    def prefetch(self, query: str, *, session_id: str = "") -> str:
+        if not self._active or not self._auto_recall or not self._client or not query.strip():
+            return ""
+        try:
+            profile = self._client.get_profile(query=query[:200])
+            include_profile = self._turn_count <= 1 or (self._turn_count % self._profile_frequency == 0)
+            context = _format_prefetch_context(
+                static_facts=profile["static"] if include_profile else [],
+                dynamic_facts=profile["dynamic"] if include_profile else [],
+                search_results=profile["search_results"],
+                max_results=self._max_recall_results,
+            )
+            return context
+        except Exception:
+            logger.debug("Supermemory prefetch failed", exc_info=True)
+            return ""
+
+    def sync_turn(self, user_content: str, assistant_content: str, *, session_id: str = "") -> None:
+        if not self._active or not self._auto_capture or not self._write_enabled or not self._client:
+            return
+
+        clean_user = _clean_text_for_capture(user_content)
+        clean_assistant = _clean_text_for_capture(assistant_content)
+        if not clean_user or not clean_assistant:
+            return
+        if self._capture_mode == "all":
+            if len(clean_user) < _MIN_CAPTURE_LENGTH or len(clean_assistant) < _MIN_CAPTURE_LENGTH:
+                return
+            if _is_trivial_message(clean_user):
+                return
+
+        content = (
+            f"[role: user]\n{clean_user}\n[user:end]\n\n"
+            f"[role: assistant]\n{clean_assistant}\n[assistant:end]"
+        )
+        metadata = {"source": "hermes", "type": "conversation_turn"}
+
+        def _run():
+            try:
+                self._client.add_memory(content, metadata=metadata, entity_context=self._entity_context)
+            except Exception:
+                logger.debug("Supermemory sync_turn failed", exc_info=True)
+
+        if self._sync_thread and self._sync_thread.is_alive():
+            self._sync_thread.join(timeout=2.0)
+        self._sync_thread = None
+        self._sync_thread = threading.Thread(target=_run, daemon=True, name="supermemory-sync")
+        self._sync_thread.start()
+
+    def on_session_end(self, messages: List[Dict[str, Any]]) -> None:
+        if not self._active or not self._write_enabled or not self._client or not self._session_id:
+            return
+        cleaned = []
+        for message in messages or []:
+            role = message.get("role")
+            if role not in ("user", "assistant"):
+                continue
+            content = _clean_text_for_capture(str(message.get("content", "")))
+            if content:
+                cleaned.append({"role": role, "content": content})
+        if not cleaned:
+            return
+        if len(cleaned) == 1 and len(cleaned[0].get("content", "")) < 20:
+            return
+        try:
+            self._client.ingest_conversation(self._session_id, cleaned)
+        except urllib.error.HTTPError:
+            logger.warning("Supermemory session ingest failed", exc_info=True)
+        except Exception:
+            logger.warning("Supermemory session ingest failed", exc_info=True)
+
+    def on_memory_write(self, action: str, target: str, content: str) -> None:
+        if not self._active or not self._write_enabled or not self._client:
+            return
+        if action != "add" or not (content or "").strip():
+            return
+
+        def _run():
+            try:
+                self._client.add_memory(
+                    content.strip(),
+                    metadata={"source": "hermes_memory", "target": target, "type": "explicit_memory"},
+                    entity_context=self._entity_context,
+                )
+            except Exception:
+                logger.debug("Supermemory on_memory_write failed", exc_info=True)
+
+        if self._write_thread and self._write_thread.is_alive():
+            self._write_thread.join(timeout=2.0)
+        self._write_thread = None
+        self._write_thread = threading.Thread(target=_run, daemon=False, name="supermemory-memory-write")
+        self._write_thread.start()
+
+    def shutdown(self) -> None:
+        for attr_name in ("_prefetch_thread", "_sync_thread", "_write_thread"):
+            thread = getattr(self, attr_name, None)
+            if thread and thread.is_alive():
+                thread.join(timeout=5.0)
+            setattr(self, attr_name, None)
+
+    def get_tool_schemas(self) -> List[Dict[str, Any]]:
+        return [STORE_SCHEMA, SEARCH_SCHEMA, FORGET_SCHEMA, PROFILE_SCHEMA]
+
+    def _tool_store(self, args: dict) -> str:
+        content = str(args.get("content") or "").strip()
+        if not content:
+            return tool_error("content is required")
+        metadata = args.get("metadata") or {}
+        if not isinstance(metadata, dict):
+            metadata = {}
+        metadata.setdefault("type", _detect_category(content))
+        metadata["source"] = "hermes_tool"
+        try:
+            result = self._client.add_memory(content, metadata=metadata, entity_context=self._entity_context)
+            preview = content[:80] + ("..." if len(content) > 80 else "")
+            return json.dumps({"saved": True, "id": result.get("id", ""), "preview": preview})
+        except Exception as exc:
+            return tool_error(f"Failed to store memory: {exc}")
+
+    def _tool_search(self, args: dict) -> str:
+        query = str(args.get("query") or "").strip()
+        if not query:
+            return tool_error("query is required")
+        try:
+            limit = max(1, min(20, int(args.get("limit", 5) or 5)))
+        except Exception:
+            limit = 5
+        try:
+            results = self._client.search_memories(query, limit=limit)
+            formatted = []
+            for item in results:
+                entry = {"id": item.get("id", ""), "content": item.get("memory", "")}
+                if item.get("similarity") is not None:
+                    try:
+                        entry["similarity"] = round(float(item["similarity"]) * 100)
+                    except Exception:
+                        pass
+                formatted.append(entry)
+            return json.dumps({"results": formatted, "count": len(formatted)})
+        except Exception as exc:
+            return tool_error(f"Search failed: {exc}")
+
+    def _tool_forget(self, args: dict) -> str:
+        memory_id = str(args.get("id") or "").strip()
+        query = str(args.get("query") or "").strip()
+        if not memory_id and not query:
+            return tool_error("Provide either id or query")
+        try:
+            if memory_id:
+                self._client.forget_memory(memory_id)
+                return json.dumps({"forgotten": True, "id": memory_id})
+            return json.dumps(self._client.forget_by_query(query))
+        except Exception as exc:
+            return tool_error(f"Forget failed: {exc}")
+
+    def _tool_profile(self, args: dict) -> str:
+        query = str(args.get("query") or "").strip() or None
+        try:
+            profile = self._client.get_profile(query=query)
+            sections = []
+            if profile["static"]:
+                sections.append("## User Profile (Persistent)\n" + "\n".join(f"- {item}" for item in profile["static"]))
+            if profile["dynamic"]:
+                sections.append("## Recent Context\n" + "\n".join(f"- {item}" for item in profile["dynamic"]))
+            return json.dumps({
+                "profile": "\n\n".join(sections),
+                "static_count": len(profile["static"]),
+                "dynamic_count": len(profile["dynamic"]),
+            })
+        except Exception as exc:
+            return tool_error(f"Profile failed: {exc}")
+
+    def handle_tool_call(self, tool_name: str, args: Dict[str, Any], **kwargs) -> str:
+        if not self._active or not self._client:
+            return tool_error("Supermemory is not configured")
+        if tool_name == "supermemory_store":
+            return self._tool_store(args)
+        if tool_name == "supermemory_search":
+            return self._tool_search(args)
+        if tool_name == "supermemory_forget":
+            return self._tool_forget(args)
+        if tool_name == "supermemory_profile":
+            return self._tool_profile(args)
+        return tool_error(f"Unknown tool: {tool_name}")
+
+
+def register(ctx):
+    ctx.register_memory_provider(SupermemoryMemoryProvider())

plugins/memory/supermemory/plugin.yaml

@@ -0,0 +1,5 @@
+name: supermemory
+version: 1.0.0
+description: "Supermemory semantic long-term memory with profile recall, semantic search, explicit memory tools, and session ingest."
+pip_dependencies:
+  - supermemory

run_agent.py

@@ -20,7 +20,6 @@ Usage:
     response = agent.run_conversation("Tell me about the latest Python updates")
 """
 
-import atexit
 import asyncio
 import base64
 import concurrent.futures
@@ -36,7 +35,6 @@ import sys
 import tempfile
 import time
 import threading
-import weakref
 from types import SimpleNamespace
 import uuid
 from typing import List, Dict, Any, Optional
@@ -528,6 +526,7 @@ class AIAgent:
         reasoning_config: Dict[str, Any] = None,
         prefill_messages: List[Dict[str, Any]] = None,
         platform: str = None,
+        user_id: str = None,
         skip_context_files: bool = False,
         skip_memory: bool = False,
         session_db=None,
@@ -592,6 +591,7 @@ class AIAgent:
         self.quiet_mode = quiet_mode
         self.ephemeral_system_prompt = ephemeral_system_prompt
         self.platform = platform  # "cli", "telegram", "discord", "whatsapp", etc.
+        self._user_id = user_id  # Platform user identifier (gateway sessions)
         # Pluggable print function — CLI replaces this with _cprint so that
         # raw ANSI status lines are routed through prompt_toolkit's renderer
         # instead of going directly to stdout where patch_stdout's StdoutProxy
@@ -654,7 +654,7 @@ class AIAgent:
         self.stream_delta_callback = stream_delta_callback
         self.status_callback = status_callback
         self.tool_gen_callback = tool_gen_callback
-        self._last_reported_tool = None  # Track for "new tool" mode
+
         
         # Tool execution state — allows _vprint during tool execution
         # even when stream consumers are registered (no tokens streaming then)
@@ -1094,6 +1094,9 @@ class AIAgent:
                             "hermes_home": str(_ghh()),
                             "agent_context": "primary",
                         }
+                        # Thread gateway user identity for per-user memory scoping
+                        if self._user_id:
+                            _init_kwargs["user_id"] = self._user_id
                         # Profile identity for per-profile provider scoping
                         try:
                             from hermes_cli.profiles import get_active_profile_name
@@ -1502,10 +1505,6 @@ class AIAgent:
         """Return True when the base URL targets OpenRouter."""
         return "openrouter" in self._base_url_lower
 
-    def _is_anthropic_url(self) -> bool:
-        """Return True when the base URL targets Anthropic (native or /anthropic proxy path)."""
-        return "api.anthropic.com" in self._base_url_lower or self._base_url_lower.rstrip("/").endswith("/anthropic")
-
     def _max_tokens_param(self, value: int) -> dict:
         """Return the correct max tokens kwarg for the current provider.
         
@@ -1691,74 +1690,6 @@ class AIAgent:
         
         return None
 
-    def _classify_empty_content_response(
-        self,
-        assistant_message,
-        *,
-        finish_reason: Optional[str],
-        approx_tokens: int,
-        api_messages: List[Dict[str, Any]],
-        conversation_history: Optional[List[Dict[str, Any]]],
-    ) -> Dict[str, Any]:
-        """Classify think-only/empty responses so we can retry, compress, or salvage.
-
-        We intentionally do NOT short-circuit all structured-reasoning responses.
-        Prior discussion/PR history shows some models recover on retry. Instead we:
-        - compress immediately when the pattern looks like implicit context pressure
-        - salvage reasoning early when the same reasoning-only payload repeats
-        - otherwise preserve the normal retry path
-        """
-        reasoning_text = self._extract_reasoning(assistant_message)
-        has_structured_reasoning = bool(
-            getattr(assistant_message, "reasoning", None)
-            or getattr(assistant_message, "reasoning_content", None)
-            or getattr(assistant_message, "reasoning_details", None)
-        )
-        content = getattr(assistant_message, "content", None) or ""
-        stripped_content = self._strip_think_blocks(content).strip()
-        signature = (
-            content,
-            reasoning_text or "",
-            bool(has_structured_reasoning),
-            finish_reason or "",
-        )
-        repeated_signature = signature == getattr(self, "_last_empty_content_signature", None)
-
-        compressor = getattr(self, "context_compressor", None)
-        ctx_len = getattr(compressor, "context_length", 0) or 0
-        threshold_tokens = getattr(compressor, "threshold_tokens", 0) or 0
-        is_large_session = bool(
-            (ctx_len and approx_tokens >= max(int(ctx_len * 0.4), threshold_tokens))
-            or len(api_messages) > 80
-        )
-        is_local_custom = is_local_endpoint(getattr(self, "base_url", "") or "")
-        is_resumed = bool(conversation_history)
-        context_pressure_signals = any(
-            [
-                finish_reason == "length",
-                getattr(compressor, "_context_probed", False),
-                is_large_session,
-                is_resumed,
-            ]
-        )
-        should_compress = bool(
-            self.compression_enabled
-            and is_local_custom
-            and context_pressure_signals
-            and not stripped_content
-        )
-
-        self._last_empty_content_signature = signature
-        return {
-            "reasoning_text": reasoning_text,
-            "has_structured_reasoning": has_structured_reasoning,
-            "repeated_signature": repeated_signature,
-            "should_compress": should_compress,
-            "is_local_custom": is_local_custom,
-            "is_large_session": is_large_session,
-            "is_resumed": is_resumed,
-        }
-    
     def _cleanup_task_resources(self, task_id: str) -> None:
         """Clean up VM and browser resources for a given task."""
         try:
@@ -2702,20 +2633,7 @@ class AIAgent:
 
         if not _soul_loaded:
             # Fallback to hardcoded identity
-            _ai_peer_name = (
-                None
-                if False
-                else None
-            )
-            if _ai_peer_name:
-                _identity = DEFAULT_AGENT_IDENTITY.replace(
-                    "You are Hermes Agent",
-                    f"You are {_ai_peer_name}",
-                    1,
-                )
-            else:
-                _identity = DEFAULT_AGENT_IDENTITY
-            prompt_parts = [_identity]
+            prompt_parts = [DEFAULT_AGENT_IDENTITY]
 
         # Tool-aware behavioral guidance: only inject when the tools are loaded
         tool_guidance = []
@@ -3400,7 +3318,7 @@ class AIAgent:
         elif "stream" in api_kwargs:
             raise ValueError("Codex Responses stream flag is only allowed in fallback streaming requests.")
 
-        unexpected = sorted(key for key in api_kwargs.keys() if key not in allowed_keys)
+        unexpected = sorted(key for key in api_kwargs if key not in allowed_keys)
         if unexpected:
             raise ValueError(
                 f"Codex Responses request has unsupported field(s): {', '.join(unexpected)}."
@@ -3444,7 +3362,22 @@ class AIAgent:
         """Normalize a Responses API object to an assistant_message-like object."""
         output = getattr(response, "output", None)
         if not isinstance(output, list) or not output:
-            raise RuntimeError("Responses API returned no output items")
+            # The Codex backend can return empty output when the answer was
+            # delivered entirely via stream events. Check output_text as a
+            # last-resort fallback before raising.
+            out_text = getattr(response, "output_text", None)
+            if isinstance(out_text, str) and out_text.strip():
+                logger.debug(
+                    "Codex response has empty output but output_text is present (%d chars); "
+                    "synthesizing output item.", len(out_text.strip()),
+                )
+                output = [SimpleNamespace(
+                    type="message", role="assistant", status="completed",
+                    content=[SimpleNamespace(type="output_text", text=out_text.strip())],
+                )]
+                response.output = output
+            else:
+                raise RuntimeError("Responses API returned no output items")
 
         response_status = getattr(response, "status", None)
         if isinstance(response_status, str):
@@ -3868,7 +3801,12 @@ class AIAgent:
         has_tool_calls = False
         first_delta_fired = False
         self._reasoning_deltas_fired = False
+        # Accumulate streamed text so we can recover if get_final_response()
+        # returns empty output (e.g. chatgpt.com backend-api sends
+        # response.incomplete instead of response.completed).
+        self._codex_streamed_text_parts: list = []
         for attempt in range(max_stream_retries + 1):
+            collected_output_items: list = []
             try:
                 with active_client.responses.stream(**api_kwargs) as stream:
                     for event in stream:
@@ -3878,6 +3816,8 @@ class AIAgent:
                         # Fire callbacks on text content deltas (suppress during tool calls)
                         if "output_text.delta" in event_type or event_type == "response.output_text.delta":
                             delta_text = getattr(event, "delta", "")
+                            if delta_text:
+                                self._codex_streamed_text_parts.append(delta_text)
                             if delta_text and not has_tool_calls:
                                 if not first_delta_fired:
                                     first_delta_fired = True
@@ -3895,7 +3835,51 @@ class AIAgent:
                             reasoning_text = getattr(event, "delta", "")
                             if reasoning_text:
                                 self._fire_reasoning_delta(reasoning_text)
-                    return stream.get_final_response()
+                        # Collect completed output items — some backends
+                        # (chatgpt.com/backend-api/codex) stream valid items
+                        # via response.output_item.done but the SDK's
+                        # get_final_response() returns an empty output list.
+                        elif event_type == "response.output_item.done":
+                            done_item = getattr(event, "item", None)
+                            if done_item is not None:
+                                collected_output_items.append(done_item)
+                        # Log non-completed terminal events for diagnostics
+                        elif event_type in ("response.incomplete", "response.failed"):
+                            resp_obj = getattr(event, "response", None)
+                            status = getattr(resp_obj, "status", None) if resp_obj else None
+                            incomplete_details = getattr(resp_obj, "incomplete_details", None) if resp_obj else None
+                            logger.warning(
+                                "Codex Responses stream received terminal event %s "
+                                "(status=%s, incomplete_details=%s, streamed_chars=%d). %s",
+                                event_type, status, incomplete_details,
+                                sum(len(p) for p in self._codex_streamed_text_parts),
+                                self._client_log_context(),
+                            )
+                    final_response = stream.get_final_response()
+                    # PATCH: ChatGPT Codex backend streams valid output items
+                    # but get_final_response() can return an empty output list.
+                    # Backfill from collected items or synthesize from deltas.
+                    _out = getattr(final_response, "output", None)
+                    if isinstance(_out, list) and not _out:
+                        if collected_output_items:
+                            final_response.output = list(collected_output_items)
+                            logger.debug(
+                                "Codex stream: backfilled %d output items from stream events",
+                                len(collected_output_items),
+                            )
+                        elif self._codex_streamed_text_parts and not has_tool_calls:
+                            assembled = "".join(self._codex_streamed_text_parts)
+                            final_response.output = [SimpleNamespace(
+                                type="message",
+                                role="assistant",
+                                status="completed",
+                                content=[SimpleNamespace(type="output_text", text=assembled)],
+                            )]
+                            logger.debug(
+                                "Codex stream: synthesized output from %d text deltas (%d chars)",
+                                len(self._codex_streamed_text_parts), len(assembled),
+                            )
+                    return final_response
             except (_httpx.RemoteProtocolError, _httpx.ReadTimeout, _httpx.ConnectError, ConnectionError) as exc:
                 if attempt < max_stream_retries:
                     logger.debug(
@@ -3946,11 +3930,28 @@ class AIAgent:
             return stream_or_response
 
         terminal_response = None
+        collected_output_items: list = []
+        collected_text_deltas: list = []
         try:
             for event in stream_or_response:
                 event_type = getattr(event, "type", None)
                 if not event_type and isinstance(event, dict):
                     event_type = event.get("type")
+
+                # Collect output items and text deltas for backfill
+                if event_type == "response.output_item.done":
+                    done_item = getattr(event, "item", None)
+                    if done_item is None and isinstance(event, dict):
+                        done_item = event.get("item")
+                    if done_item is not None:
+                        collected_output_items.append(done_item)
+                elif event_type in ("response.output_text.delta",):
+                    delta = getattr(event, "delta", "")
+                    if not delta and isinstance(event, dict):
+                        delta = event.get("delta", "")
+                    if delta:
+                        collected_text_deltas.append(delta)
+
                 if event_type not in {"response.completed", "response.incomplete", "response.failed"}:
                     continue
 
@@ -3958,6 +3959,26 @@ class AIAgent:
                 if terminal_response is None and isinstance(event, dict):
                     terminal_response = event.get("response")
                 if terminal_response is not None:
+                    # Backfill empty output from collected stream events
+                    _out = getattr(terminal_response, "output", None)
+                    if isinstance(_out, list) and not _out:
+                        if collected_output_items:
+                            terminal_response.output = list(collected_output_items)
+                            logger.debug(
+                                "Codex fallback stream: backfilled %d output items",
+                                len(collected_output_items),
+                            )
+                        elif collected_text_deltas:
+                            assembled = "".join(collected_text_deltas)
+                            terminal_response.output = [SimpleNamespace(
+                                type="message", role="assistant",
+                                status="completed",
+                                content=[SimpleNamespace(type="output_text", text=assembled)],
+                            )]
+                            logger.debug(
+                                "Codex fallback stream: synthesized from %d deltas (%d chars)",
+                                len(collected_text_deltas), len(assembled),
+                            )
                     return terminal_response
         finally:
             close_fn = getattr(stream_or_response, "close", None)
@@ -5720,6 +5741,7 @@ class AIAgent:
                 api_msg.pop("reasoning", None)
                 api_msg.pop("finish_reason", None)
                 api_msg.pop("_flush_sentinel", None)
+                api_msg.pop("_thinking_prefill", None)
                 if _needs_sanitize:
                     self._sanitize_tool_calls_for_strict_api(api_msg)
                 api_messages.append(api_msg)
@@ -5805,7 +5827,7 @@ class AIAgent:
                         args = json.loads(tc.function.arguments)
                         flush_target = args.get("target", "memory")
                         from tools.memory_tool import memory_tool as _memory_tool
-                        result = _memory_tool(
+                        _memory_tool(
                             action=args.get("action"),
                             target=flush_target,
                             content=args.get("content"),
@@ -6643,7 +6665,7 @@ class AIAgent:
             api_messages = []
             for msg in messages:
                 api_msg = msg.copy()
-                for internal_field in ("reasoning", "finish_reason"):
+                for internal_field in ("reasoning", "finish_reason", "_thinking_prefill"):
                     api_msg.pop(internal_field, None)
                 if _needs_sanitize:
                     self._sanitize_tool_calls_for_strict_api(api_msg)
@@ -6835,6 +6857,7 @@ class AIAgent:
         self._empty_content_retries = 0
         self._incomplete_scratchpad_retries = 0
         self._codex_incomplete_retries = 0
+        self._thinking_prefill_retries = 0
         self._last_content_with_tools = None
         self._mute_post_response = False
         self._surrogate_sanitized = False
@@ -7180,6 +7203,8 @@ class AIAgent:
                 # Remove finish_reason - not accepted by strict APIs (e.g. Mistral)
                 if "finish_reason" in api_msg:
                     api_msg.pop("finish_reason")
+                # Strip internal thinking-prefill marker
+                api_msg.pop("_thinking_prefill", None)
                 # Strip Codex Responses API fields (call_id, response_item_id) for
                 # strict providers like Mistral, Fireworks, etc. that reject unknown fields.
                 # Uses new dicts so the internal messages list retains the fields
@@ -7365,7 +7390,19 @@ class AIAgent:
                         elif not isinstance(output_items, list):
                             response_invalid = True
                             error_details.append("response.output is not a list")
-                        elif len(output_items) == 0:
+                        elif not output_items:
+                            # If we reach here, _run_codex_stream's backfill
+                            # from output_item.done events and text-delta
+                            # synthesis both failed to populate output.
+                            _resp_status = getattr(response, "status", None)
+                            _resp_incomplete = getattr(response, "incomplete_details", None)
+                            logging.warning(
+                                "Codex response.output is empty after stream backfill "
+                                "(status=%s, incomplete_details=%s, model=%s). %s",
+                                _resp_status, _resp_incomplete,
+                                getattr(response, "model", None),
+                                f"api_mode={self.api_mode} provider={self.provider}",
+                            )
                             response_invalid = True
                             error_details.append("response.output is empty")
                     elif self.api_mode == "anthropic_messages":
@@ -7376,11 +7413,11 @@ class AIAgent:
                         elif not isinstance(content_blocks, list):
                             response_invalid = True
                             error_details.append("response.content is not a list")
-                        elif len(content_blocks) == 0:
+                        elif not content_blocks:
                             response_invalid = True
                             error_details.append("response.content is empty")
                     else:
-                        if response is None or not hasattr(response, 'choices') or response.choices is None or len(response.choices) == 0:
+                        if response is None or not hasattr(response, 'choices') or response.choices is None or not response.choices:
                             response_invalid = True
                             if response is None:
                                 error_details.append("response is None")
@@ -8191,7 +8228,7 @@ class AIAgent:
                                 self._vprint(f"{self.log_prefix}   💡 Codex OAuth token was rejected (HTTP 401). Your token may have been", force=True)
                                 self._vprint(f"{self.log_prefix}      refreshed by another client (Codex CLI, VS Code). To fix:", force=True)
                                 self._vprint(f"{self.log_prefix}      1. Run `codex` in your terminal to generate fresh tokens.", force=True)
-                                self._vprint(f"{self.log_prefix}      2. Then run `hermes login --provider openai-codex` to re-authenticate.", force=True)
+                                self._vprint(f"{self.log_prefix}      2. Then run `hermes auth` to re-authenticate.", force=True)
                             else:
                                 self._vprint(f"{self.log_prefix}   💡 Your API key was rejected by the provider. Check:", force=True)
                                 self._vprint(f"{self.log_prefix}      • Is the key valid? Run: hermes setup", force=True)
@@ -8702,6 +8739,15 @@ class AIAgent:
                             if clean:
                                 self._vprint(f"  ┊ 💬 {clean}")
                     
+                    # Pop thinking-only prefill message(s) before appending
+                    # (tool-call path — same rationale as the final-response path).
+                    while (
+                        messages
+                        and isinstance(messages[-1], dict)
+                        and messages[-1].get("_thinking_prefill")
+                    ):
+                        messages.pop()
+
                     messages.append(assistant_msg)
 
                     # Close any open streaming display (response box, reasoning
@@ -8815,11 +8861,36 @@ class AIAgent:
                             self._response_was_previewed = True
                             break
 
-                        # Reasoning-only response: the model produced thinking
-                        # but no visible content.  This is a valid response —
-                        # keep reasoning in its own field and set content to
-                        # "(empty)" so every provider accepts the message.
-                        # No retries needed.
+                        # ── Thinking-only prefill continuation ──────────
+                        # The model produced structured reasoning (via API
+                        # fields) but no visible text content.  Rather than
+                        # giving up, append the assistant message as-is and
+                        # continue — the model will see its own reasoning
+                        # on the next turn and produce the text portion.
+                        # Inspired by clawdbot's "incomplete-text" recovery.
+                        _has_structured = bool(
+                            getattr(assistant_message, "reasoning", None)
+                            or getattr(assistant_message, "reasoning_content", None)
+                            or getattr(assistant_message, "reasoning_details", None)
+                        )
+                        if _has_structured and self._thinking_prefill_retries < 2:
+                            self._thinking_prefill_retries += 1
+                            self._vprint(
+                                f"{self.log_prefix}↻ Thinking-only response — "
+                                f"prefilling to continue "
+                                f"({self._thinking_prefill_retries}/2)"
+                            )
+                            interim_msg = self._build_assistant_message(
+                                assistant_message, "incomplete"
+                            )
+                            interim_msg["_thinking_prefill"] = True
+                            messages.append(interim_msg)
+                            self._session_messages = messages
+                            self._save_session_log(messages)
+                            continue
+
+                        # Exhausted prefill attempts or no structured
+                        # reasoning — fall through to "(empty)" terminal.
                         reasoning_text = self._extract_reasoning(assistant_message)
                         assistant_msg = self._build_assistant_message(assistant_message, finish_reason)
                         assistant_msg["content"] = "(empty)"
@@ -8838,6 +8909,7 @@ class AIAgent:
                     if hasattr(self, '_empty_content_retries'):
                         self._empty_content_retries = 0
                     self._last_empty_content_signature = None
+                    self._thinking_prefill_retries = 0
 
                     if (
                         self.api_mode == "codex_responses"
@@ -8876,7 +8948,18 @@ class AIAgent:
                     final_response = self._strip_think_blocks(final_response).strip()
                     
                     final_msg = self._build_assistant_message(assistant_message, finish_reason)
-                    
+
+                    # Pop thinking-only prefill message(s) before appending
+                    # the final response.  This avoids consecutive assistant
+                    # messages which break strict-alternation providers
+                    # (Anthropic Messages API) and keeps history clean.
+                    while (
+                        messages
+                        and isinstance(messages[-1], dict)
+                        and messages[-1].get("_thinking_prefill")
+                    ):
+                        messages.pop()
+
                     messages.append(final_msg)
                     
                     if not self.quiet_mode:
@@ -8918,7 +9001,6 @@ class AIAgent:
                                     "content": f"Error executing tool: {error_msg}",
                                 }
                                 messages.append(err_msg)
-                        pending_handled = True
                     break
                 
                 # Non-tool errors don't need a synthetic message injected.

scripts/release.py

@@ -21,8 +21,6 @@ Usage:
 """
 
 import argparse
-import json
-import os
 import re
 import shutil
 import subprocess

scripts/sample_and_compress.py

@@ -17,7 +17,6 @@ Usage:
 
 import json
 import random
-import os
 from pathlib import Path
 from typing import List, Dict, Any, Tuple
 import fire
@@ -138,7 +137,6 @@ def sample_from_datasets(
         List of sampled trajectory entries
     """
     from multiprocessing import Pool
-    from functools import partial
     
     random.seed(seed)
     

skills/creative/manim-video/SKILL.md

@@ -24,7 +24,7 @@ This is educational cinema. Every frame teaches. Every animation reveals structu
 
 ## Prerequisites
 
-Run `scripts/setup.sh` to verify all dependencies. Requires: Python 3.10+, Manim Community Edition (`pip install manim`), LaTeX (`texlive-full` on Linux, `mactex` on macOS), and ffmpeg.
+Run `scripts/setup.sh` to verify all dependencies. Requires: Python 3.10+, Manim Community Edition v0.20+ (`pip install manim`), LaTeX (`texlive-full` on Linux, `mactex` on macOS), and ffmpeg. Reference docs tested against Manim CE v0.20.1.
 
 ## Modes
 

skills/creative/manim-video/references/animations.md

@@ -50,6 +50,31 @@ self.play(circle.animate.set_color(RED))
 self.play(circle.animate.shift(RIGHT * 2).scale(0.5))  # chain multiple
 ```
 
+## Additional Creation Animations
+
+```python
+self.play(GrowFromPoint(circle, LEFT * 3))     # scale 0 -> 1 from a specific point
+self.play(GrowFromEdge(rect, DOWN))             # grow from one edge
+self.play(SpinInFromNothing(square))            # scale up while rotating (default PI/2)
+self.play(GrowArrow(arrow))                     # grows arrow from start to tip
+```
+
+## Movement Animations
+
+```python
+# Move a mobject along an arbitrary path
+path = Arc(radius=2, angle=PI)
+self.play(MoveAlongPath(dot, path), run_time=2)
+
+# Rotate (as a Transform, not .animate — supports about_point)
+self.play(Rotate(square, angle=PI / 2, about_point=ORIGIN), run_time=1.5)
+
+# Rotating (continuous rotation, updater-style — good for spinning objects)
+self.play(Rotating(gear, angle=TAU, run_time=4, rate_func=linear))
+```
+
+`MoveAlongPath` takes any `VMobject` as the path — use `Arc`, `CubicBezier`, `Line`, or a custom `VMobject`. Position is computed via `path.point_from_proportion()`.
+
 ## Emphasis Animations
 
 ```python

skills/creative/manim-video/references/equations.md

@@ -65,6 +65,57 @@ MathTex(r"\vec{v}")                       # vector
 MathTex(r"\lim_{x \to \infty} f(x)")    # limit
 ```
 
+## Matrices
+
+`MathTex` supports standard LaTeX matrix environments via `amsmath` (loaded by default):
+
+```python
+# Bracketed matrix
+MathTex(r"\begin{bmatrix} 1 & 0 \\ 0 & 1 \end{bmatrix}")
+
+# Parenthesized matrix
+MathTex(r"\begin{pmatrix} a & b \\ c & d \end{pmatrix}")
+
+# Determinant (vertical bars)
+MathTex(r"\begin{vmatrix} a & b \\ c & d \end{vmatrix}")
+
+# Plain (no delimiters)
+MathTex(r"\begin{matrix} x_1 \\ x_2 \\ x_3 \end{matrix}")
+```
+
+For matrices you need to animate element-by-element or color individual entries, use the `IntegerMatrix`, `DecimalMatrix`, or `MobjectMatrix` mobjects instead — see `mobjects.md`.
+
+## Cases and Piecewise Functions
+
+```python
+MathTex(r"""
+    f(x) = \begin{cases}
+        x^2    & \text{if } x \geq 0 \\
+        -x^2   & \text{if } x < 0
+    \end{cases}
+""")
+```
+
+## Aligned Environments
+
+For multi-line derivations with alignment, use `aligned` inside `MathTex`:
+
+```python
+MathTex(r"""
+    \begin{aligned}
+        \nabla \cdot \mathbf{E} &= \frac{\rho}{\epsilon_0} \\
+        \nabla \cdot \mathbf{B} &= 0 \\
+        \nabla \times \mathbf{E} &= -\frac{\partial \mathbf{B}}{\partial t} \\
+        \nabla \times \mathbf{B} &= \mu_0 \mathbf{J} + \mu_0 \epsilon_0 \frac{\partial \mathbf{E}}{\partial t}
+    \end{aligned}
+""")
+```
+
+Note: `MathTex` wraps content in `align*` by default. Override with `tex_environment` if needed:
+```python
+MathTex(r"...", tex_environment="gather*")
+```
+
 ## Derivation Pattern
 
 ```python

skills/creative/manim-video/references/mobjects.md

@@ -35,6 +35,52 @@ rrect = RoundedRectangle(corner_radius=0.3, width=4, height=2)
 brace = Brace(rect, DOWN, color=YELLOW)
 ```
 
+## Polygons and Arcs
+
+```python
+# Arbitrary polygon from vertices
+poly = Polygon(LEFT, UP * 2, RIGHT, color=GREEN, fill_opacity=0.3)
+
+# Regular n-sided polygon
+hexagon = RegularPolygon(n=6, color=TEAL, fill_opacity=0.4)
+
+# Triangle (shorthand for RegularPolygon(n=3))
+tri = Triangle(color=YELLOW, fill_opacity=0.5)
+
+# Arc (portion of a circle)
+arc = Arc(radius=2, start_angle=0, angle=PI / 2, color=BLUE)
+
+# Arc between two points
+arc_between = ArcBetweenPoints(LEFT * 2, RIGHT * 2, angle=TAU / 4, color=RED)
+
+# Curved arrow (arc with tip)
+curved_arrow = CurvedArrow(LEFT * 2, RIGHT * 2, color=ORANGE)
+```
+
+## Sectors and Annuli
+
+```python
+# Sector (pie slice)
+sector = Sector(outer_radius=2, start_angle=0, angle=PI / 3, fill_opacity=0.7, color=BLUE)
+
+# Annulus (ring)
+ring = Annulus(inner_radius=1, outer_radius=2, fill_opacity=0.5, color=GREEN)
+
+# Annular sector (partial ring)
+partial_ring = AnnularSector(
+    inner_radius=1, outer_radius=2,
+    angle=PI / 2, start_angle=0,
+    fill_opacity=0.7, color=TEAL
+)
+
+# Cutout (punch holes in a shape)
+background = Square(side_length=4, fill_opacity=1, color=BLUE)
+hole = Circle(radius=0.5)
+cutout = Cutout(background, hole, fill_opacity=1, color=BLUE)
+```
+
+Use cases: pie charts, ring progress indicators, Venn diagrams with arcs, geometric proofs.
+
 ## Positioning
 
 ```python
@@ -99,6 +145,29 @@ class NetworkNode(Group):
         self.add(self.circle, self.label)
 ```
 
+## Matrix Mobjects
+
+Display matrices as grids of numbers or mobjects:
+
+```python
+# Integer matrix
+m = IntegerMatrix([[1, 2], [3, 4]])
+
+# Decimal matrix (control decimal places)
+m = DecimalMatrix([[1.5, 2.7], [3.1, 4.9]], element_to_mobject_config={"num_decimal_places": 2})
+
+# Mobject matrix (any mobject in each cell)
+m = MobjectMatrix([
+    [MathTex(r"\pi"), MathTex(r"e")],
+    [MathTex(r"\phi"), MathTex(r"\tau")]
+])
+
+# Bracket types: "(" "[" "|" or "\\{"
+m = IntegerMatrix([[1, 0], [0, 1]], left_bracket="[", right_bracket="]")
+```
+
+Use cases: linear algebra, transformation matrices, system-of-equations coefficient display.
+
 ## Constants
 
 Directions: `UP, DOWN, LEFT, RIGHT, ORIGIN, UL, UR, DL, DR`

skills/dogfood/SKILL.md

@@ -16,7 +16,7 @@ This skill guides you through systematic exploratory QA testing of web applicati
 
 ## Prerequisites
 
-- Browser toolset must be available (`browser_navigate`, `browser_snapshot`, `browser_click`, `browser_type`, `browser_vision`, `browser_console`, `browser_scroll`, `browser_back`, `browser_press`, `browser_close`)
+- Browser toolset must be available (`browser_navigate`, `browser_snapshot`, `browser_click`, `browser_type`, `browser_vision`, `browser_console`, `browser_scroll`, `browser_back`, `browser_press`)
 - A target URL and testing scope from the user
 
 ## Inputs
@@ -148,7 +148,6 @@ Save the report to `{output_dir}/report.md`.
 | `browser_press` | Press a keyboard key |
 | `browser_vision` | Screenshot + AI analysis; use `annotate=true` for element labels |
 | `browser_console` | Get JS console output and errors |
-| `browser_close` | Close the browser session |
 
 ## Tips
 

skills/mlops/training/grpo-rl-training/templates/basic_grpo_training.py

@@ -12,7 +12,7 @@ Adapt this for your specific task by modifying:
 
 import torch
 import re
-from datasets import load_dataset, Dataset
+from datasets import load_dataset
 from transformers import AutoModelForCausalLM, AutoTokenizer
 from peft import LoraConfig
 from trl import GRPOTrainer, GRPOConfig

skills/red-teaming/godmode/scripts/auto_jailbreak.py

@@ -16,13 +16,10 @@ Usage in execute_code:
 """
 
 import os
-import sys
 import json
 import time
-import re
 import yaml
 from pathlib import Path
-from concurrent.futures import ThreadPoolExecutor, as_completed
 
 try:
     from openai import OpenAI

skills/red-teaming/godmode/scripts/godmode_race.py

@@ -20,7 +20,6 @@ Usage in execute_code:
 
 import os
 import re
-import json
 import time
 from concurrent.futures import ThreadPoolExecutor, as_completed
 
@@ -404,7 +403,6 @@ def race_godmode_classic(query, api_key=None, timeout=60):
     Each combo uses a different model paired with its best-performing jailbreak prompt.
     Returns the best result across all combos.
     """
-    from collections import namedtuple
     
     HALL_OF_FAME = [
         {

skills/red-teaming/godmode/scripts/parseltongue.py

@@ -17,7 +17,6 @@ Usage:
 
 import re
 import base64
-import sys
 
 # ═══════════════════════════════════════════════════════════════════
 # Trigger words that commonly trip safety classifiers

skills/research/blogwatcher/SKILL.md

@@ -1,48 +1,106 @@
 ---
 name: blogwatcher
-description: Monitor blogs and RSS/Atom feeds for updates using the blogwatcher CLI. Add blogs, scan for new articles, and track what you've read.
-version: 1.0.0
-author: community
+description: Monitor blogs and RSS/Atom feeds for updates using the blogwatcher-cli tool. Add blogs, scan for new articles, track read status, and filter by category.
+version: 2.0.0
+author: JulienTant (fork of Hyaxia/blogwatcher)
 license: MIT
 metadata:
   hermes:
     tags: [RSS, Blogs, Feed-Reader, Monitoring]
-    homepage: https://github.com/Hyaxia/blogwatcher
+    homepage: https://github.com/JulienTant/blogwatcher-cli
 prerequisites:
-  commands: [blogwatcher]
+  commands: [blogwatcher-cli]
 ---
 
 # Blogwatcher
 
-Track blog and RSS/Atom feed updates with the `blogwatcher` CLI.
+Track blog and RSS/Atom feed updates with the `blogwatcher-cli` tool. Supports automatic feed discovery, HTML scraping fallback, OPML import, and read/unread article management.
 
-## Prerequisites
+## Installation
 
-- Go installed (`go version` to check)
-- Install: `go install github.com/Hyaxia/blogwatcher/cmd/blogwatcher@latest`
+Pick one method:
+
+- **Go:** `go install github.com/JulienTant/blogwatcher-cli/cmd/blogwatcher-cli@latest`
+- **Docker:** `docker run --rm -v blogwatcher-cli:/data ghcr.io/julientant/blogwatcher-cli`
+- **Binary (Linux amd64):** `curl -sL https://github.com/JulienTant/blogwatcher-cli/releases/latest/download/blogwatcher-cli_linux_amd64.tar.gz | tar xz -C /usr/local/bin blogwatcher-cli`
+- **Binary (Linux arm64):** `curl -sL https://github.com/JulienTant/blogwatcher-cli/releases/latest/download/blogwatcher-cli_linux_arm64.tar.gz | tar xz -C /usr/local/bin blogwatcher-cli`
+- **Binary (macOS Apple Silicon):** `curl -sL https://github.com/JulienTant/blogwatcher-cli/releases/latest/download/blogwatcher-cli_darwin_arm64.tar.gz | tar xz -C /usr/local/bin blogwatcher-cli`
+- **Binary (macOS Intel):** `curl -sL https://github.com/JulienTant/blogwatcher-cli/releases/latest/download/blogwatcher-cli_darwin_amd64.tar.gz | tar xz -C /usr/local/bin blogwatcher-cli`
+
+All releases: https://github.com/JulienTant/blogwatcher-cli/releases
+
+### Docker with persistent storage
+
+By default the database lives at `~/.blogwatcher-cli/blogwatcher-cli.db`. In Docker this is lost on container restart. Use `BLOGWATCHER_DB` or a volume mount to persist it:
+
+```bash
+# Named volume (simplest)
+docker run --rm -v blogwatcher-cli:/data -e BLOGWATCHER_DB=/data/blogwatcher-cli.db ghcr.io/julientant/blogwatcher-cli scan
+
+# Host bind mount
+docker run --rm -v /path/on/host:/data -e BLOGWATCHER_DB=/data/blogwatcher-cli.db ghcr.io/julientant/blogwatcher-cli scan
+```
+
+### Migrating from the original blogwatcher
+
+If upgrading from `Hyaxia/blogwatcher`, move your database:
+
+```bash
+mv ~/.blogwatcher/blogwatcher.db ~/.blogwatcher-cli/blogwatcher-cli.db
+```
+
+The binary name changed from `blogwatcher` to `blogwatcher-cli`.
 
 ## Common Commands
 
-- Add a blog: `blogwatcher add "My Blog" https://example.com`
-- List blogs: `blogwatcher blogs`
-- Scan for updates: `blogwatcher scan`
-- List articles: `blogwatcher articles`
-- Mark an article read: `blogwatcher read 1`
-- Mark all articles read: `blogwatcher read-all`
-- Remove a blog: `blogwatcher remove "My Blog"`
+### Managing blogs
+
+- Add a blog: `blogwatcher-cli add "My Blog" https://example.com`
+- Add with explicit feed: `blogwatcher-cli add "My Blog" https://example.com --feed-url https://example.com/feed.xml`
+- Add with HTML scraping: `blogwatcher-cli add "My Blog" https://example.com --scrape-selector "article h2 a"`
+- List tracked blogs: `blogwatcher-cli blogs`
+- Remove a blog: `blogwatcher-cli remove "My Blog" --yes`
+- Import from OPML: `blogwatcher-cli import subscriptions.opml`
+
+### Scanning and reading
+
+- Scan all blogs: `blogwatcher-cli scan`
+- Scan one blog: `blogwatcher-cli scan "My Blog"`
+- List unread articles: `blogwatcher-cli articles`
+- List all articles: `blogwatcher-cli articles --all`
+- Filter by blog: `blogwatcher-cli articles --blog "My Blog"`
+- Filter by category: `blogwatcher-cli articles --category "Engineering"`
+- Mark article read: `blogwatcher-cli read 1`
+- Mark article unread: `blogwatcher-cli unread 1`
+- Mark all read: `blogwatcher-cli read-all`
+- Mark all read for a blog: `blogwatcher-cli read-all --blog "My Blog" --yes`
+
+## Environment Variables
+
+All flags can be set via environment variables with the `BLOGWATCHER_` prefix:
+
+| Variable | Description |
+|---|---|
+| `BLOGWATCHER_DB` | Path to SQLite database file |
+| `BLOGWATCHER_WORKERS` | Number of concurrent scan workers (default: 8) |
+| `BLOGWATCHER_SILENT` | Only output "scan done" when scanning |
+| `BLOGWATCHER_YES` | Skip confirmation prompts |
+| `BLOGWATCHER_CATEGORY` | Default filter for articles by category |
 
 ## Example Output
 
 ```
-$ blogwatcher blogs
+$ blogwatcher-cli blogs
 Tracked blogs (1):
 
   xkcd
     URL: https://xkcd.com
+    Feed: https://xkcd.com/atom.xml
+    Last scanned: 2026-04-03 10:30
 ```
 
 ```
-$ blogwatcher scan
+$ blogwatcher-cli scan
 Scanning 1 blog(s)...
 
   xkcd
@@ -51,6 +109,28 @@ Scanning 1 blog(s)...
 Found 4 new article(s) total!
 ```
 
+```
+$ blogwatcher-cli articles
+Unread articles (2):
+
+  [1] [new] Barrel - Part 13
+       Blog: xkcd
+       URL: https://xkcd.com/3095/
+       Published: 2026-04-02
+       Categories: Comics, Science
+
+  [2] [new] Volcano Fact
+       Blog: xkcd
+       URL: https://xkcd.com/3094/
+       Published: 2026-04-01
+       Categories: Comics
+```
+
 ## Notes
 
-- Use `blogwatcher <command> --help` to discover flags and options.
+- Auto-discovers RSS/Atom feeds from blog homepages when no `--feed-url` is provided.
+- Falls back to HTML scraping if RSS fails and `--scrape-selector` is configured.
+- Categories from RSS/Atom feeds are stored and can be used to filter articles.
+- Import blogs in bulk from OPML files exported by Feedly, Inoreader, NewsBlur, etc.
+- Database stored at `~/.blogwatcher-cli/blogwatcher-cli.db` by default (override with `--db` or `BLOGWATCHER_DB`).
+- Use `blogwatcher-cli <command> --help` to discover all flags and options.

tests/agent/test_auxiliary_config_bridge.py

@@ -13,7 +13,7 @@ from unittest.mock import patch, MagicMock
 import pytest
 import yaml
 
-sys.path.insert(0, os.path.join(os.path.dirname(__file__), ".."))
+sys.path.insert(0, os.path.join(os.path.dirname(__file__), "..", ".."))
 
 
 def _run_auxiliary_bridge(config_dict, monkeypatch):
@@ -199,7 +199,7 @@ class TestGatewayBridgeCodeParity:
 
     def test_gateway_has_auxiliary_bridge(self):
         """The gateway config bridge must include auxiliary.* bridging."""
-        gateway_path = Path(__file__).parent.parent / "gateway" / "run.py"
+        gateway_path = Path(__file__).parent.parent.parent / "gateway" / "run.py"
         content = gateway_path.read_text()
         # Check for key patterns that indicate the bridge is present
         assert "AUXILIARY_VISION_PROVIDER" in content
@@ -213,7 +213,7 @@ class TestGatewayBridgeCodeParity:
 
     def test_gateway_no_compression_env_bridge(self):
         """Gateway should NOT bridge compression config to env vars (config-only)."""
-        gateway_path = Path(__file__).parent.parent / "gateway" / "run.py"
+        gateway_path = Path(__file__).parent.parent.parent / "gateway" / "run.py"
         content = gateway_path.read_text()
         assert "CONTEXT_COMPRESSION_PROVIDER" not in content
         assert "CONTEXT_COMPRESSION_MODEL" not in content

tests/agent/test_credential_pool.py

@@ -947,7 +947,7 @@ def test_list_custom_pool_providers(tmp_path, monkeypatch):
                         "auth_type": "api_key",
                         "priority": 0,
                         "source": "manual",
-                        "access_token": "sk-ant-xxx",
+                        "access_token": "***",
                     }
                 ],
                 "custom:together.ai": [
@@ -957,7 +957,7 @@ def test_list_custom_pool_providers(tmp_path, monkeypatch):
                         "auth_type": "api_key",
                         "priority": 0,
                         "source": "manual",
-                        "access_token": "sk-tog-xxx",
+                        "access_token": "***",
                     }
                 ],
                 "custom:fireworks": [
@@ -967,7 +967,7 @@ def test_list_custom_pool_providers(tmp_path, monkeypatch):
                         "auth_type": "api_key",
                         "priority": 0,
                         "source": "manual",
-                        "access_token": "sk-fw-xxx",
+                        "access_token": "***",
                     }
                 ],
                 "custom:empty": [],
@@ -980,3 +980,78 @@ def test_list_custom_pool_providers(tmp_path, monkeypatch):
     result = list_custom_pool_providers()
     assert result == ["custom:fireworks", "custom:together.ai"]
     # "custom:empty" not included because it's empty
+
+
+
+def test_acquire_lease_prefers_unleased_entry(tmp_path, monkeypatch):
+    monkeypatch.setenv("HERMES_HOME", str(tmp_path / "hermes"))
+    _write_auth_store(
+        tmp_path,
+        {
+            "version": 1,
+            "credential_pool": {
+                "openrouter": [
+                    {
+                        "id": "cred-1",
+                        "label": "primary",
+                        "auth_type": "api_key",
+                        "priority": 0,
+                        "source": "manual",
+                        "access_token": "***",
+                    },
+                    {
+                        "id": "cred-2",
+                        "label": "secondary",
+                        "auth_type": "api_key",
+                        "priority": 1,
+                        "source": "manual",
+                        "access_token": "***",
+                    },
+                ]
+            },
+        },
+    )
+
+    from agent.credential_pool import load_pool
+
+    pool = load_pool("openrouter")
+    first = pool.acquire_lease()
+    second = pool.acquire_lease()
+
+    assert first == "cred-1"
+    assert second == "cred-2"
+    assert pool.active_lease_count("cred-1") == 1
+    assert pool.active_lease_count("cred-2") == 1
+
+
+
+def test_release_lease_decrements_counter(tmp_path, monkeypatch):
+    monkeypatch.setenv("HERMES_HOME", str(tmp_path / "hermes"))
+    _write_auth_store(
+        tmp_path,
+        {
+            "version": 1,
+            "credential_pool": {
+                "openrouter": [
+                    {
+                        "id": "cred-1",
+                        "label": "primary",
+                        "auth_type": "api_key",
+                        "priority": 0,
+                        "source": "manual",
+                        "access_token": "***",
+                    }
+                ]
+            },
+        },
+    )
+
+    from agent.credential_pool import load_pool
+
+    pool = load_pool("openrouter")
+    leased = pool.acquire_lease()
+    assert leased == "cred-1"
+    assert pool.active_lease_count("cred-1") == 1
+
+    pool.release_lease("cred-1")
+    assert pool.active_lease_count("cred-1") == 0