fix(terminal): suppress declare -x stdout leak when sourcing snapshot on macOS

On macOS (bash 3.2 and certain Homebrew builds), sourcing a file
containing declare -x statements prints each declaration to stdout.
The session snapshot (written by export -p) is full of these, so
every command execution prepends ~60 lines of env vars into the LLM's
context window, wasting tokens and potentially exceeding context
limits (issue #15459).

Fix: redirect both stdout and stderr to /dev/null when sourcing the
snapshot. The source builtin still sets the variables in the current
shell — only the print output is suppressed. On Linux this is already
silent, so the redirect is harmless.

Fixes #15459

agent/context_compressor.py

@@ -484,7 +484,18 @@ class ContextCompressor(ContextEngine):
             for i in range(len(result) - 1, -1, -1):
                 msg = result[i]
                 raw_content = msg.get("content") or ""
-                content_len = sum(len(p.get("text", "")) for p in raw_content) if isinstance(raw_content, list) else len(raw_content)
+                content_len = (
+                    sum(
+                        len(p.get("text", ""))
+                        if isinstance(p, dict)
+                        else len(p)
+                        if isinstance(p, str)
+                        else len(str(p))
+                        for p in raw_content
+                    )
+                    if isinstance(raw_content, list)
+                    else len(raw_content)
+                )
                 msg_tokens = content_len // _CHARS_PER_TOKEN + 10
                 for tc in msg.get("tool_calls") or []:
                     if isinstance(tc, dict):
@@ -1082,8 +1093,20 @@ The user has requested that this compaction PRIORITISE preserving all informatio
 
         for i in range(n - 1, head_end - 1, -1):
             msg = messages[i]
-            content = msg.get("content") or ""
-            msg_tokens = len(content) // _CHARS_PER_TOKEN + 10  # +10 for role/metadata
+            raw_content = msg.get("content") or ""
+            content_len = (
+                sum(
+                    len(p.get("text", ""))
+                    if isinstance(p, dict)
+                    else len(p)
+                    if isinstance(p, str)
+                    else len(str(p))
+                    for p in raw_content
+                )
+                if isinstance(raw_content, list)
+                else len(raw_content)
+            )
+            msg_tokens = content_len // _CHARS_PER_TOKEN + 10  # +10 for role/metadata
             # Include tool call arguments in estimate
             for tc in msg.get("tool_calls") or []:
                 if isinstance(tc, dict):

agent/credential_pool.py

@@ -14,6 +14,7 @@ from datetime import datetime
 from typing import Any, Dict, List, Optional, Set, Tuple
 
 from hermes_constants import OPENROUTER_BASE_URL
+from hermes_cli.config import get_env_value
 import hermes_cli.auth as auth_mod
 from hermes_cli.auth import (
     CODEX_ACCESS_TOKEN_REFRESH_SKEW_SECONDS,
@@ -1273,7 +1274,8 @@ def _seed_from_env(provider: str, entries: List[PooledCredential]) -> Tuple[bool
         def _is_source_suppressed(_p, _s):  # type: ignore[misc]
             return False
     if provider == "openrouter":
-        token = os.getenv("OPENROUTER_API_KEY", "").strip()
+        # Check both os.environ and ~/.hermes/.env file
+        token = (get_env_value("OPENROUTER_API_KEY") or "").strip()
         if token:
             source = "env:OPENROUTER_API_KEY"
             if _is_source_suppressed(provider, source):
@@ -1299,7 +1301,7 @@ def _seed_from_env(provider: str, entries: List[PooledCredential]) -> Tuple[bool
 
     env_url = ""
     if pconfig.base_url_env_var:
-        env_url = os.getenv(pconfig.base_url_env_var, "").strip().rstrip("/")
+        env_url = (get_env_value(pconfig.base_url_env_var) or "").strip().rstrip("/")
 
     env_vars = list(pconfig.api_key_env_vars)
     if provider == "anthropic":
@@ -1310,7 +1312,8 @@ def _seed_from_env(provider: str, entries: List[PooledCredential]) -> Tuple[bool
         ]
 
     for env_var in env_vars:
-        token = os.getenv(env_var, "").strip()
+        # Check both os.environ and ~/.hermes/.env file
+        token = (get_env_value(env_var) or "").strip()
         if not token:
             continue
         source = f"env:{env_var}"

agent/onboarding.py

@@ -25,6 +25,7 @@ logger = logging.getLogger(__name__)
 
 BUSY_INPUT_FLAG = "busy_input_prompt"
 TOOL_PROGRESS_FLAG = "tool_progress_prompt"
+OPENCLAW_RESIDUE_FLAG = "openclaw_residue_cleanup"
 
 
 # -------------------------------------------------------------------------
@@ -43,10 +44,18 @@ def busy_input_hint_gateway(mode: str) -> str:
             "Send `/busy interrupt` to make new messages stop the current task "
             "immediately, or `/busy status` to check. This notice won't appear again."
         )
+    if mode == "steer":
+        return (
+            "💡 First-time tip — I steered your message into the current run; "
+            "it will arrive after the next tool call instead of interrupting. "
+            "Send `/busy interrupt` or `/busy queue` to change this, or "
+            "`/busy status` to check. This notice won't appear again."
+        )
     return (
         "💡 First-time tip — I just interrupted my current task to answer you. "
         "Send `/busy queue` to queue follow-ups for after the current task instead, "
-        "or `/busy status` to check. This notice won't appear again."
+        "`/busy steer` to inject them mid-run without interrupting, or "
+        "`/busy status` to check. This notice won't appear again."
     )
 
 
@@ -55,13 +64,19 @@ def busy_input_hint_cli(mode: str) -> str:
     if mode == "queue":
         return (
             "(tip) Your message was queued for the next turn. "
-            "Use /busy interrupt to make Enter stop the current run instead. "
-            "This tip only shows once."
+            "Use /busy interrupt to make Enter stop the current run instead, "
+            "or /busy steer to inject mid-run. This tip only shows once."
+        )
+    if mode == "steer":
+        return (
+            "(tip) Your message was steered into the current run; it arrives "
+            "after the next tool call. Use /busy interrupt or /busy queue to "
+            "change this. This tip only shows once."
         )
     return (
         "(tip) Your message interrupted the current run. "
-        "Use /busy queue to queue messages for the next turn instead. "
-        "This tip only shows once."
+        "Use /busy queue to queue messages for the next turn instead, "
+        "or /busy steer to inject mid-run. This tip only shows once."
     )
 
 
@@ -80,6 +95,35 @@ def tool_progress_hint_cli() -> str:
     )
 
 
+def openclaw_residue_hint_cli() -> str:
+    """Banner shown the first time Hermes starts and finds ``~/.openclaw/``.
+
+    OpenClaw-era config, memory, and skill paths in ``~/.openclaw/`` will
+    otherwise attract the agent (memory entries like ``~/.openclaw/config.yaml``
+    get carried forward and the agent dutifully reads them). ``hermes claw
+    cleanup`` renames the directory so the agent stops finding it.
+    """
+    return (
+        "Heads up — an OpenClaw workspace was detected at ~/.openclaw/.\n"
+        "After migrating, the agent can still get confused and read that "
+        "directory's config/memory instead of Hermes's.\n"
+        "Run `hermes claw cleanup` to archive it (rename → .openclaw.pre-migration). "
+        "This tip only shows once; rerun it any time with `hermes claw cleanup`."
+    )
+
+
+def detect_openclaw_residue(home: Optional[Path] = None) -> bool:
+    """Return True if an OpenClaw workspace directory is present in ``$HOME``.
+
+    Pure filesystem check — no side effects. ``home`` override exists for tests.
+    """
+    base = home or Path.home()
+    try:
+        return (base / ".openclaw").is_dir()
+    except OSError:
+        return False
+
+
 # -------------------------------------------------------------------------
 # State read / write
 # -------------------------------------------------------------------------
@@ -135,10 +179,13 @@ def mark_seen(config_path: Path, flag: str) -> bool:
 __all__ = [
     "BUSY_INPUT_FLAG",
     "TOOL_PROGRESS_FLAG",
+    "OPENCLAW_RESIDUE_FLAG",
     "busy_input_hint_gateway",
     "busy_input_hint_cli",
     "tool_progress_hint_gateway",
     "tool_progress_hint_cli",
+    "openclaw_residue_hint_cli",
+    "detect_openclaw_residue",
     "is_seen",
     "mark_seen",
 ]

agent/prompt_builder.py

@@ -422,6 +422,29 @@ PLATFORM_HINTS = {
         "your response. Images are sent as native photos, and other files arrive as downloadable "
         "documents."
     ),
+    "yuanbao": (
+        "You are on Yuanbao (腾讯元宝), a Chinese AI assistant platform. "
+        "Markdown formatting is supported (code blocks, tables, bold/italic). "
+        "You CAN send media files natively — to deliver a file to the user, include "
+        "MEDIA:/absolute/path/to/file in your response. The file will be sent as a native "
+        "Yuanbao attachment: images (.jpg, .png, .webp, .gif) are sent as photos, "
+        "and other files (.pdf, .docx, .txt, .zip, etc.) arrive as downloadable documents "
+        "(max 50 MB). You can also include image URLs in markdown format ![alt](url) and "
+        "they will be downloaded and sent as native photos. "
+        "Do NOT tell the user you lack file-sending capability — use MEDIA: syntax "
+        "whenever a file delivery is appropriate.\n\n"
+        "Stickers (贴纸 / 表情包 / TIM face): Yuanbao has a built-in sticker catalogue. "
+        "When the user sends a sticker (you see '[emoji: 名称]' in their message) or asks "
+        "you to send/reply-with a 贴纸/表情/表情包, you MUST use the sticker tools:\n"
+        "  1. Call yb_search_sticker with a Chinese keyword (e.g. '666', '比心', '吃瓜', "
+        "     '捂脸', '合十') to discover matching sticker_ids.\n"
+        "  2. Call yb_send_sticker with the chosen sticker_id or name — this sends a real "
+        "     TIMFaceElem that renders as a native sticker in the chat.\n"
+        "DO NOT draw sticker-like PNGs with execute_code/Pillow/matplotlib and then send "
+        "them via MEDIA: or send_image_file. That produces a fake low-quality 'sticker' "
+        "image and is the WRONG path. Bare Unicode emoji in text is also not a substitute "
+        "— when a sticker is the right response, use yb_send_sticker."
+    ),
 }
 
 # ---------------------------------------------------------------------------
@@ -825,6 +848,11 @@ def build_skills_system_prompt(
             "Skills also encode the user's preferred approach, conventions, and quality standards "
             "for tasks like code review, planning, and testing — load them even for tasks you "
             "already know how to do, because the skill defines how it should be done here.\n"
+            "Whenever the user asks you to configure, set up, install, enable, disable, modify, "
+            "or troubleshoot Hermes Agent itself — its CLI, config, models, providers, tools, "
+            "skills, voice, gateway, plugins, or any feature — load the `hermes-agent` skill "
+            "first. It has the actual commands (e.g. `hermes config set …`, `hermes tools`, "
+            "`hermes setup`) so you don't have to guess or invent workarounds.\n"
             "If a skill has issues, fix it with skill_manage(action='patch').\n"
             "After difficult/iterative tasks, offer to save as a skill. "
             "If a skill you loaded was missing steps, had wrong commands, or needed "

agent/shell_hooks.py

@@ -754,7 +754,11 @@ def _resolve_effective_accept(
     if env in ("1", "true", "yes", "on"):
         return True
     cfg_val = cfg.get("hooks_auto_accept", False)
-    return bool(cfg_val)
+    if isinstance(cfg_val, bool):
+        return cfg_val
+    if isinstance(cfg_val, str):
+        return cfg_val.strip().lower() in ("1", "true", "yes", "on")
+    return False
 
 
 # ---------------------------------------------------------------------------

agent/skill_commands.py

@@ -329,7 +329,7 @@ def build_skill_invocation_message(
 
     loaded_skill, skill_dir, skill_name = loaded
     activation_note = (
-        f'[SYSTEM: The user has invoked the "{skill_name}" skill, indicating they want '
+        f'[IMPORTANT: The user has invoked the "{skill_name}" skill, indicating they want '
         "you to follow its instructions. The full skill content is loaded below.]"
     )
     return _build_skill_message(
@@ -368,7 +368,7 @@ def build_preloaded_skills_prompt(
 
         loaded_skill, skill_dir, skill_name = loaded
         activation_note = (
-            f'[SYSTEM: The user launched this CLI session with the "{skill_name}" skill '
+            f'[IMPORTANT: The user launched this CLI session with the "{skill_name}" skill '
             "preloaded. Treat its instructions as active guidance for the duration of this "
             "session unless the user overrides them.]"
         )

agent/title_generator.py

@@ -6,12 +6,18 @@ adds latency to the user-facing reply.
 
 import logging
 import threading
-from typing import Optional
+from typing import Callable, Optional
 
 from agent.auxiliary_client import call_llm
 
 logger = logging.getLogger(__name__)
 
+# Callback signature: (task_name, exception) -> None. Used to surface
+# auxiliary failures to the user through AIAgent._emit_auxiliary_failure
+# so silent-drops (e.g. OpenRouter 402 exhausting the fallback chain)
+# become visible instead of piling up as NULL session titles.
+FailureCallback = Callable[[str, BaseException], None]
+
 _TITLE_PROMPT = (
     "Generate a short, descriptive title (3-7 words) for a conversation that starts with the "
     "following exchange. The title should capture the main topic or intent. "
@@ -19,11 +25,21 @@ _TITLE_PROMPT = (
 )
 
 
-def generate_title(user_message: str, assistant_response: str, timeout: float = 30.0) -> Optional[str]:
+def generate_title(
+    user_message: str,
+    assistant_response: str,
+    timeout: float = 30.0,
+    failure_callback: Optional[FailureCallback] = None,
+) -> Optional[str]:
     """Generate a session title from the first exchange.
 
     Uses the auxiliary LLM client (cheapest/fastest available model).
     Returns the title string or None on failure.
+
+    ``failure_callback`` is invoked with ``(task, exception)`` when the
+    auxiliary call raises — the caller typically wires this to
+    ``AIAgent._emit_auxiliary_failure`` so the user sees a warning instead
+    of silently accumulating untitled sessions.
     """
     # Truncate long messages to keep the request small
     user_snippet = user_message[:500] if user_message else ""
@@ -52,7 +68,15 @@ def generate_title(user_message: str, assistant_response: str, timeout: float =
             title = title[:77] + "..."
         return title if title else None
     except Exception as e:
-        logger.debug("Title generation failed: %s", e)
+        # Log at WARNING so this shows up in agent.log without debug mode.
+        # Full detail at debug level for operators who need the stack.
+        logger.warning("Title generation failed: %s", e)
+        logger.debug("Title generation traceback", exc_info=True)
+        if failure_callback is not None:
+            try:
+                failure_callback("title generation", e)
+            except Exception:
+                logger.debug("Title generation failure_callback raised", exc_info=True)
         return None
 
 
@@ -61,6 +85,7 @@ def auto_title_session(
     session_id: str,
     user_message: str,
     assistant_response: str,
+    failure_callback: Optional[FailureCallback] = None,
 ) -> None:
     """Generate and set a session title if one doesn't already exist.
 
@@ -81,7 +106,9 @@ def auto_title_session(
     except Exception:
         return
 
-    title = generate_title(user_message, assistant_response)
+    title = generate_title(
+        user_message, assistant_response, failure_callback=failure_callback
+    )
     if not title:
         return
 
@@ -98,6 +125,7 @@ def maybe_auto_title(
     user_message: str,
     assistant_response: str,
     conversation_history: list,
+    failure_callback: Optional[FailureCallback] = None,
 ) -> None:
     """Fire-and-forget title generation after the first exchange.
 
@@ -119,6 +147,7 @@ def maybe_auto_title(
     thread = threading.Thread(
         target=auto_title_session,
         args=(session_db, session_id, user_message, assistant_response),
+        kwargs={"failure_callback": failure_callback},
         daemon=True,
         name="auto-title",
     )

cli-config.yaml.example

@@ -606,6 +606,7 @@ platform_toolsets:
   signal: [hermes-signal]
   homeassistant: [hermes-homeassistant]
   qqbot: [hermes-qqbot]
+  yuanbao: [hermes-yuanbao]
 
 # =============================================================================
 # Gateway Platform Settings
@@ -847,8 +848,12 @@ display:
   # What Enter does when Hermes is already busy (CLI and gateway platforms).
   #   interrupt: Interrupt the current run and redirect Hermes (default)
   #   queue:     Queue your message for the next turn
+  #   steer:     Inject your message mid-run via /steer, arriving at the agent
+  #              after the next tool call — no interrupt, no role violation.
+  #              Falls back to 'queue' if the agent isn't running yet or if
+  #              images are attached (steer only carries text).
   # Ctrl+C (or /stop in gateway) always interrupts regardless of this setting.
-  # Toggle at runtime with /busy_input_mode <interrupt|queue>.
+  # Toggle at runtime with /busy <interrupt|queue|steer>.
   busy_input_mode: interrupt
 
   # Background process notifications (gateway/messaging only).

cli.py

@@ -15,6 +15,7 @@ Usage:
 
 import logging
 import os
+import re
 import shutil
 import sys
 import json
@@ -974,6 +975,7 @@ def _run_state_db_auto_maintenance(session_db) -> None:
         return
     try:
         from hermes_cli.config import load_config as _load_full_config
+        from hermes_constants import get_hermes_home as _get_hermes_home
         cfg = (_load_full_config().get("sessions") or {})
         if not cfg.get("auto_prune", False):
             return
@@ -981,11 +983,35 @@ def _run_state_db_auto_maintenance(session_db) -> None:
             retention_days=int(cfg.get("retention_days", 90)),
             min_interval_hours=int(cfg.get("min_interval_hours", 24)),
             vacuum=bool(cfg.get("vacuum_after_prune", True)),
+            sessions_dir=_get_hermes_home() / "sessions",
         )
     except Exception as exc:
         logger.debug("state.db auto-maintenance skipped: %s", exc)
 
 
+def _run_checkpoint_auto_maintenance() -> None:
+    """Call ``checkpoint_manager.maybe_auto_prune_checkpoints`` using current config.
+
+    Reads the ``checkpoints:`` section from config.yaml via
+    :func:`hermes_cli.config.load_config`. Honours ``auto_prune`` /
+    ``retention_days`` / ``delete_orphans`` / ``min_interval_hours``.
+    Never raises — maintenance must never block interactive startup.
+    """
+    try:
+        from hermes_cli.config import load_config as _load_full_config
+        cfg = (_load_full_config().get("checkpoints") or {})
+        if not cfg.get("auto_prune", False):
+            return
+        from tools.checkpoint_manager import maybe_auto_prune_checkpoints
+        maybe_auto_prune_checkpoints(
+            retention_days=int(cfg.get("retention_days", 7)),
+            min_interval_hours=int(cfg.get("min_interval_hours", 24)),
+            delete_orphans=bool(cfg.get("delete_orphans", True)),
+        )
+    except Exception as exc:
+        logger.debug("checkpoint auto-maintenance skipped: %s", exc)
+
+
 def _prune_stale_worktrees(repo_root: str, max_age_hours: int = 24) -> None:
     """Remove stale worktrees and orphaned branches on startup.
 
@@ -1378,7 +1404,7 @@ def _resolve_attachment_path(raw_path: str) -> Path | None:
 
 
 def _format_process_notification(evt: dict) -> "str | None":
-    """Format a process notification event into a [SYSTEM: ...] message.
+    """Format a process notification event into a [IMPORTANT: ...] message.
 
     Handles both completion events (notify_on_complete) and watch pattern
     match events from the unified completion_queue.
@@ -1388,14 +1414,14 @@ def _format_process_notification(evt: dict) -> "str | None":
     _cmd = evt.get("command", "unknown")
 
     if evt_type == "watch_disabled":
-        return f"[SYSTEM: {evt.get('message', '')}]"
+        return f"[IMPORTANT: {evt.get('message', '')}]"
 
     if evt_type == "watch_match":
         _pat = evt.get("pattern", "?")
         _out = evt.get("output", "")
         _sup = evt.get("suppressed", 0)
         text = (
-            f"[SYSTEM: Background process {_sid} matched "
+            f"[IMPORTANT: Background process {_sid} matched "
             f"watch pattern \"{_pat}\".\n"
             f"Command: {_cmd}\n"
             f"Matched output:\n{_out}"
@@ -1409,7 +1435,7 @@ def _format_process_notification(evt: dict) -> "str | None":
     _exit = evt.get("exit_code", "?")
     _out = evt.get("output", "")
     return (
-        f"[SYSTEM: Background process {_sid} completed "
+        f"[IMPORTANT: Background process {_sid} completed "
         f"(exit code {_exit}).\n"
         f"Command: {_cmd}\n"
         f"Output:\n{_out}]"
@@ -1522,6 +1548,32 @@ def _should_auto_attach_clipboard_image_on_paste(pasted_text: str) -> bool:
     return not pasted_text.strip()
 
 
+def _strip_leaked_bracketed_paste_wrappers(text: str) -> str:
+    """Strip leaked bracketed-paste wrapper markers from user-visible text.
+
+    Defensive normalization for cases where terminal/prompt_toolkit parsing
+    fails and bracketed-paste markers end up in the buffer as literal text.
+
+    We strip canonical wrappers unconditionally and also handle degraded
+    visible forms like ``[200~`` / ``[201~`` and ``00~`` / ``01~`` when they
+    look like wrapper boundaries, not arbitrary user content.
+    """
+    if not text:
+        return text
+
+    text = (
+        text.replace("\x1b[200~", "")
+        .replace("\x1b[201~", "")
+        .replace("^[[200~", "")
+        .replace("^[[201~", "")
+    )
+    text = re.sub(r"(^|[\s\n>:\]\)])\[200~", r"\1", text)
+    text = re.sub(r"\[201~(?=$|[\s\n<\[\(\):;.,!?])", "", text)
+    text = re.sub(r"(^|[\s\n>:\]\)])00~", r"\1", text)
+    text = re.sub(r"01~(?=$|[\s\n<\[\(\):;.,!?])", "", text)
+    return text
+
+
 def _collect_query_images(query: str | None, image_arg: str | None = None) -> tuple[str, list[Path]]:
     """Collect local image attachments for single-query CLI flows."""
     message = query or ""
@@ -1848,9 +1900,16 @@ class HermesCLI:
         self.bell_on_complete = CLI_CONFIG["display"].get("bell_on_complete", False)
         # show_reasoning: display model thinking/reasoning before the response
         self.show_reasoning = CLI_CONFIG["display"].get("show_reasoning", False)
-        # busy_input_mode: "interrupt" (Enter interrupts current run) or "queue" (Enter queues for next turn)
-        _bim = CLI_CONFIG["display"].get("busy_input_mode", "interrupt")
-        self.busy_input_mode = "queue" if str(_bim).strip().lower() == "queue" else "interrupt"
+        # busy_input_mode: "interrupt" (Enter interrupts current run),
+        # "queue" (Enter queues for next turn), or "steer" (Enter injects
+        # mid-run via /steer, arriving after the next tool call).
+        _bim = str(CLI_CONFIG["display"].get("busy_input_mode", "interrupt")).strip().lower()
+        if _bim == "queue":
+            self.busy_input_mode = "queue"
+        elif _bim == "steer":
+            self.busy_input_mode = "steer"
+        else:
+            self.busy_input_mode = "interrupt"
 
         self.verbose = verbose if verbose is not None else (self.tool_progress_mode == "verbose")
         
@@ -2045,6 +2104,11 @@ class HermesCLI:
         # Never blocks startup on failure.
         _run_state_db_auto_maintenance(self._session_db)
 
+        # Opportunistic shadow-repo cleanup — deletes orphan/stale
+        # checkpoint repos under ~/.hermes/checkpoints/.  Opt-in via
+        # checkpoints.auto_prune, idempotent via .last_prune marker.
+        _run_checkpoint_auto_maintenance()
+
         # Deferred title: stored in memory until the session is created in the DB
         self._pending_title: Optional[str] = None
         
@@ -4915,6 +4979,12 @@ class HermesCLI:
         if self.agent:
             self.agent.session_id = new_session_id
             self.agent.session_start = now
+            # Redirect the JSON session log to the new branch session file so
+            # messages written after branching land in the correct file.
+            if hasattr(self.agent, "session_log_file") and hasattr(self.agent, "logs_dir"):
+                self.agent.session_log_file = (
+                    self.agent.logs_dir / f"session_{new_session_id}.json"
+                )
             self.agent.reset_session_state()
             if hasattr(self.agent, "_last_flushed_db_idx"):
                 self.agent._last_flushed_db_idx = len(self.conversation_history)
@@ -4936,22 +5006,37 @@ class HermesCLI:
         _cprint(f"  Branch session:   {new_session_id}")
 
     def save_conversation(self):
-        """Save the current conversation to a file."""
+        """Save the current conversation to a JSON snapshot under ~/.hermes/sessions/saved/.
+
+        The snapshot is a convenience export for sharing or off-line inspection;
+        every message is already persisted incrementally to the SQLite session
+        DB, so the live session remains resumable via ``hermes --resume <id>``
+        regardless of whether the user ever runs ``/save``.
+        """
         if not self.conversation_history:
             print("(;_;) No conversation to save.")
             return
-        
+
         timestamp = datetime.now().strftime("%Y%m%d_%H%M%S")
-        filename = f"hermes_conversation_{timestamp}.json"
-        
+        saved_dir = get_hermes_home() / "sessions" / "saved"
         try:
-            with open(filename, "w", encoding="utf-8") as f:
+            saved_dir.mkdir(parents=True, exist_ok=True)
+        except Exception as e:
+            print(f"(x_x) Failed to create save directory {saved_dir}: {e}")
+            return
+        path = saved_dir / f"hermes_conversation_{timestamp}.json"
+
+        try:
+            with open(path, "w", encoding="utf-8") as f:
                 json.dump({
                     "model": self.model,
+                    "session_id": self.session_id,
                     "session_start": self.session_start.isoformat(),
                     "messages": self.conversation_history,
                 }, f, indent=2, ensure_ascii=False)
-            print(f"(^_^)v Conversation saved to: {filename}")
+            print(f"(^_^)v Conversation snapshot saved to: {path}")
+            if self.session_id:
+                print(f"       Resume the live session with: hermes --resume {self.session_id}")
         except Exception as e:
             print(f"(x_x) Failed to save: {e}")
     
@@ -6307,6 +6392,12 @@ class HermesCLI:
         turn_route = self._resolve_turn_agent_config(prompt)
 
         def run_background():
+            set_sudo_password_callback(self._sudo_password_callback)
+            set_approval_callback(self._approval_callback)
+            try:
+                set_secret_capture_callback(self._secret_capture_callback)
+            except Exception:
+                pass
             try:
                 bg_agent = AIAgent(
                     model=turn_route["model"],
@@ -6404,6 +6495,12 @@ class HermesCLI:
                 print()
                 _cprint(f"  ❌ Background task #{task_num} failed: {e}")
             finally:
+                try:
+                    set_sudo_password_callback(None)
+                    set_approval_callback(None)
+                    set_secret_capture_callback(None)
+                except Exception:
+                    pass
                 self._background_tasks.pop(task_id, None)
                 # Clear spinner only if no foreground agent owns it
                 if not self._agent_running:
@@ -6798,24 +6895,36 @@ class HermesCLI:
             /busy               Show current busy input mode
             /busy status        Show current busy input mode
             /busy queue         Queue input for the next turn instead of interrupting
+            /busy steer         Inject Enter mid-run via /steer (after next tool call)
             /busy interrupt     Interrupt the current run on Enter (default)
         """
         parts = cmd.strip().split(maxsplit=1)
         if len(parts) < 2 or parts[1].strip().lower() == "status":
             _cprint(f"  {_ACCENT}Busy input mode: {self.busy_input_mode}{_RST}")
-            _cprint(f"  {_DIM}Enter while busy: {'queues for next turn' if self.busy_input_mode == 'queue' else 'interrupts current run'}{_RST}")
-            _cprint(f"  {_DIM}Usage: /busy [queue|interrupt|status]{_RST}")
+            if self.busy_input_mode == "queue":
+                _behavior = "queues for next turn"
+            elif self.busy_input_mode == "steer":
+                _behavior = "steers into current run (after next tool call)"
+            else:
+                _behavior = "interrupts current run"
+            _cprint(f"  {_DIM}Enter while busy: {_behavior}{_RST}")
+            _cprint(f"  {_DIM}Usage: /busy [queue|steer|interrupt|status]{_RST}")
             return
 
         arg = parts[1].strip().lower()
-        if arg not in {"queue", "interrupt"}:
+        if arg not in {"queue", "interrupt", "steer"}:
             _cprint(f"  {_DIM}(._.) Unknown argument: {arg}{_RST}")
-            _cprint(f"  {_DIM}Usage: /busy [queue|interrupt|status]{_RST}")
+            _cprint(f"  {_DIM}Usage: /busy [queue|steer|interrupt|status]{_RST}")
             return
 
         self.busy_input_mode = arg
         if save_config_value("display.busy_input_mode", arg):
-            behavior = "Enter will queue follow-up input while Hermes is busy." if arg == "queue" else "Enter will interrupt the current run while Hermes is busy."
+            if arg == "queue":
+                behavior = "Enter will queue follow-up input while Hermes is busy."
+            elif arg == "steer":
+                behavior = "Enter will steer your message into the current run (after the next tool call)."
+            else:
+                behavior = "Enter will interrupt the current run while Hermes is busy."
             _cprint(f"  {_ACCENT}✓ Busy input mode set to '{arg}' (saved to config){_RST}")
             _cprint(f"  {_DIM}{behavior}{_RST}")
         else:
@@ -7217,7 +7326,7 @@ class HermesCLI:
             change_detail = ". ".join(change_parts) + ". " if change_parts else ""
             self.conversation_history.append({
                 "role": "user",
-                "content": f"[SYSTEM: MCP servers have been reloaded. {change_detail}{tool_summary}. The tool list for this conversation has been updated accordingly.]",
+                "content": f"[IMPORTANT: MCP servers have been reloaded. {change_detail}{tool_summary}. The tool list for this conversation has been updated accordingly.]",
             })
 
             # Persist session immediately so the session log reflects the
@@ -8563,12 +8672,20 @@ class HermesCLI:
             if response and result and not result.get("failed") and not result.get("partial"):
                 try:
                     from agent.title_generator import maybe_auto_title
+                    # Route title-generation failures through the agent's
+                    # user-visible warning channel so a depleted auxiliary
+                    # provider doesn't silently leave sessions untitled
+                    # (issue #15775).
+                    _title_failure_cb = getattr(
+                        self.agent, "_emit_auxiliary_failure", None
+                    ) if self.agent else None
                     maybe_auto_title(
                         self._session_db,
                         self.session_id,
                         message,
                         response,
                         self.conversation_history,
+                        failure_callback=_title_failure_cb,
                     )
                 except Exception:
                     pass
@@ -8991,6 +9108,30 @@ class HermesCLI:
             _welcome_text = "Welcome to Hermes Agent! Type your message or /help for commands."
             _welcome_color = "#FFF8DC"
         self._console_print(f"[{_welcome_color}]{_welcome_text}[/]")
+        # First-time OpenClaw-residue banner — fires once if ~/.openclaw/ exists
+        # after an OpenClaw→Hermes migration (especially migrations done by
+        # OpenClaw's own tool, which doesn't archive the source directory).
+        try:
+            from agent.onboarding import (
+                OPENCLAW_RESIDUE_FLAG,
+                detect_openclaw_residue,
+                is_seen,
+                mark_seen,
+                openclaw_residue_hint_cli,
+            )
+            if not is_seen(self.config, OPENCLAW_RESIDUE_FLAG) and detect_openclaw_residue():
+                try:
+                    _resid_color = _welcome_skin.get_color("banner_dim", "#B8860B")
+                except Exception:
+                    _resid_color = "#B8860B"
+                self._console_print(f"[{_resid_color}]{openclaw_residue_hint_cli()}[/]")
+                try:
+                    from hermes_cli.config import get_config_path as _get_cfg_path_resid
+                    mark_seen(_get_cfg_path_resid(), OPENCLAW_RESIDUE_FLAG)
+                except Exception:
+                    pass  # best-effort — banner will fire again next session
+        except Exception:
+            pass  # banner is non-critical — never break startup
         # Show a random tip to help users discover features
         try:
             from hermes_cli.tips import get_random_tip
@@ -9192,12 +9333,34 @@ class HermesCLI:
                 # Bundle text + images as a tuple when images are present
                 payload = (text, images) if images else text
                 if self._agent_running and not (text and _looks_like_slash_command(text)):
-                    if self.busy_input_mode == "queue":
+                    _effective_mode = self.busy_input_mode
+                    if _effective_mode == "steer":
+                        # Route Enter through /steer — inject mid-run after the
+                        # next tool call.  Images can't ride along (steer only
+                        # appends text), so fall back to queue when images are
+                        # attached.  If the agent lacks steer() or rejects the
+                        # payload, also fall back to queue so nothing is lost.
+                        if images or not text:
+                            _effective_mode = "queue"
+                        else:
+                            accepted = False
+                            try:
+                                if self.agent is not None and hasattr(self.agent, "steer"):
+                                    accepted = bool(self.agent.steer(text))
+                            except Exception as exc:
+                                _cprint(f"  {_DIM}Steer failed ({exc}) — queued for next turn.{_RST}")
+                                accepted = False
+                            if accepted:
+                                preview = text[:80] + ("..." if len(text) > 80 else "")
+                                _cprint(f"  {_ACCENT}⏩ Steered: '{preview}'{_RST}")
+                            else:
+                                _effective_mode = "queue"
+                    if _effective_mode == "queue":
                         # Queue for the next turn instead of interrupting
                         self._pending_input.put(payload)
                         preview = text if text else f"[{len(images)} image{'s' if len(images) != 1 else ''} attached]"
                         _cprint(f"  Queued for the next turn: {preview[:80]}{'...' if len(preview) > 80 else ''}")
-                    else:
+                    elif _effective_mode == "interrupt":
                         self._interrupt_queue.put(payload)
                         # Debug: log to file when message enters interrupt queue
                         try:
@@ -9631,6 +9794,7 @@ class HermesCLI:
             # Normalise line endings — Windows \r\n and old Mac \r both become \n
             # so the 5-line collapse threshold and display are consistent.
             pasted_text = pasted_text.replace('\r\n', '\n').replace('\r', '\n')
+            pasted_text = _strip_leaked_bracketed_paste_wrappers(pasted_text)
             if _should_auto_attach_clipboard_image_on_paste(pasted_text) and self._try_attach_clipboard_image():
                 event.app.invalidate()
             if pasted_text:
@@ -9772,7 +9936,15 @@ class HermesCLI:
                still batch newlines.  Alt+Enter only adds 1 newline per
                event so it never triggers this.
             """
-            text = buf.text
+            text = _strip_leaked_bracketed_paste_wrappers(buf.text)
+            if text != buf.text:
+                cursor = min(buf.cursor_position, len(text))
+                _paste_just_collapsed[0] = True
+                buf.text = text
+                buf.cursor_position = cursor
+                _prev_text_len[0] = len(text)
+                _prev_newline_count[0] = text.count('\n')
+                return
             chars_added = len(text) - _prev_text_len[0]
             _prev_text_len[0] = len(text)
             if _paste_just_collapsed[0] or self._skip_paste_collapse:
@@ -9841,7 +10013,7 @@ class HermesCLI:
                 status = cli_ref._command_status or "Processing command..."
                 return f"{frame} {status}"
             if cli_ref._agent_running:
-                return "type a message + Enter to interrupt, Ctrl+C to cancel"
+                return "msg=interrupt · /queue · /bg · /steer · Ctrl+C cancel"
             if cli_ref._voice_mode:
                 return "type or Ctrl+B to record"
             return ""
@@ -10520,6 +10692,9 @@ class HermesCLI:
                     submit_images = []
                     if isinstance(user_input, tuple):
                         user_input, submit_images = user_input
+
+                    if isinstance(user_input, str):
+                        user_input = _strip_leaked_bracketed_paste_wrappers(user_input)
                     
                     # Check for commands — but detect dragged/pasted file paths first.
                     # See _detect_file_drop() for details.

cron/jobs.py

@@ -311,6 +311,12 @@ def compute_next_run(schedule: Dict[str, Any], last_run_at: Optional[str] = None
 
     elif schedule["kind"] == "cron":
         if not HAS_CRONITER:
+            logger.warning(
+                "Cannot compute next run for cron schedule %r: 'croniter' "
+                "is not installed. Install the 'cron' extra (pip install "
+                "'hermes-agent[cron]') to re-enable recurring cron jobs.",
+                schedule.get("expr"),
+            )
             return None
         cron = croniter(schedule["expr"], now)
         next_run = cron.get_next(datetime)
@@ -698,10 +704,32 @@ def mark_job_run(job_id: str, success: bool, error: Optional[str] = None,
                 # Compute next run
                 job["next_run_at"] = compute_next_run(job["schedule"], now)
 
-                # If no next run (one-shot completed), disable
+                # If no next run, decide whether this is terminal completion
+                # (one-shot) or a transient failure (recurring schedule couldn't
+                # compute — e.g. 'croniter' missing from the runtime env).
+                # Recurring jobs must NEVER be silently disabled: that turns a
+                # missing runtime dep into "job completed" and the user's
+                # schedule quietly goes off. See issue #16265.
                 if job["next_run_at"] is None:
-                    job["enabled"] = False
-                    job["state"] = "completed"
+                    kind = job.get("schedule", {}).get("kind")
+                    if kind in ("cron", "interval"):
+                        job["state"] = "error"
+                        if not job.get("last_error"):
+                            job["last_error"] = (
+                                "Failed to compute next run for recurring "
+                                "schedule (is the 'croniter' package "
+                                "installed in the gateway's Python env?)"
+                            )
+                        logger.error(
+                            "Job '%s' (%s) could not compute next_run_at; "
+                            "leaving enabled and marking state=error so the "
+                            "job is not silently disabled.",
+                            job.get("name", job["id"]),
+                            kind,
+                        )
+                    else:
+                        job["enabled"] = False
+                        job["state"] = "completed"
                 elif job.get("state") != "paused":
                     job["state"] = "scheduled"
 

cron/scheduler.py

@@ -77,7 +77,7 @@ _KNOWN_DELIVERY_PLATFORMS = frozenset({
     "telegram", "discord", "slack", "whatsapp", "signal",
     "matrix", "mattermost", "homeassistant", "dingtalk", "feishu",
     "wecom", "wecom_callback", "weixin", "sms", "email", "webhook", "bluebubbles",
-    "qqbot",
+    "qqbot", "yuanbao",
 })
 
 # Platforms that support a configured cron/notification home target, mapped to
@@ -337,6 +337,7 @@ def _deliver_result(job: dict, content: str, adapters=None, loop=None) -> Option
         "sms": Platform.SMS,
         "bluebubbles": Platform.BLUEBUBBLES,
         "qqbot": Platform.QQBOT,
+        "yuanbao": Platform.YUANBAO,
     }
 
     # Optionally wrap the content with a header/footer so the user knows this
@@ -715,7 +716,7 @@ def _build_job_prompt(job: dict, prerun_script: Optional[tuple] = None) -> str:
     # Always prepend cron execution guidance so the agent knows how
     # delivery works and can suppress delivery when appropriate.
     cron_hint = (
-        "[SYSTEM: You are running as a scheduled cron job. "
+        "[IMPORTANT: You are running as a scheduled cron job. "
         "DELIVERY: Your final response will be automatically delivered "
         "to the user — do NOT use send_message or try to deliver "
         "the output yourself. Just produce your report/output as your "
@@ -751,7 +752,7 @@ def _build_job_prompt(job: dict, prerun_script: Optional[tuple] = None) -> str:
             parts.append("")
         parts.extend(
             [
-                f'[SYSTEM: The user has invoked the "{skill_name}" skill, indicating they want you to follow its instructions. The full skill content is loaded below.]',
+                f'[IMPORTANT: The user has invoked the "{skill_name}" skill, indicating they want you to follow its instructions. The full skill content is loaded below.]',
                 "",
                 content,
             ]
@@ -759,7 +760,7 @@ def _build_job_prompt(job: dict, prerun_script: Optional[tuple] = None) -> str:
 
     if skipped:
         notice = (
-            f"[SYSTEM: The following skill(s) were listed for this job but could not be found "
+            f"[IMPORTANT: The following skill(s) were listed for this job but could not be found "
             f"and were skipped: {', '.join(skipped)}. "
             f"Start your response with a brief notice so the user is aware, e.g.: "
             f"'⚠️ Skill(s) not found and skipped: {', '.join(skipped)}']"
@@ -1308,6 +1309,17 @@ def tick(verbose: bool = True, adapters=None, loop=None) -> int:
                     _futures.append(_tick_pool.submit(_ctx.run, _process_job, job))
                 _results.extend(f.result() for f in _futures)
 
+        # Best-effort sweep of MCP stdio subprocesses that survived their
+        # session teardown during this tick.  Runs AFTER every job has
+        # finished so active sessions (including live user chats) are
+        # never touched — only PIDs explicitly detected as orphans in
+        # tools.mcp_tool._run_stdio's finally block are reaped.
+        try:
+            from tools.mcp_tool import _kill_orphaned_mcp_children
+            _kill_orphaned_mcp_children()
+        except Exception as _e:
+            logger.debug("Post-tick MCP orphan cleanup failed: %s", _e)
+
         return sum(_results)
     finally:
         if fcntl:

gateway/channel_directory.py

@@ -57,7 +57,7 @@ def _session_entry_name(origin: Dict[str, Any]) -> str:
 # Build / refresh
 # ---------------------------------------------------------------------------
 
-def build_channel_directory(adapters: Dict[Any, Any]) -> Dict[str, Any]:
+async def build_channel_directory(adapters: Dict[Any, Any]) -> Dict[str, Any]:
     """
     Build a channel directory from connected platform adapters and session data.
 
@@ -72,7 +72,7 @@ def build_channel_directory(adapters: Dict[Any, Any]) -> Dict[str, Any]:
             if platform == Platform.DISCORD:
                 platforms["discord"] = _build_discord(adapter)
             elif platform == Platform.SLACK:
-                platforms["slack"] = _build_slack(adapter)
+                platforms["slack"] = await _build_slack(adapter)
         except Exception as e:
             logger.warning("Channel directory: failed to build %s: %s", platform.value, e)
 
@@ -136,21 +136,66 @@ def _build_discord(adapter) -> List[Dict[str, str]]:
     return channels
 
 
-def _build_slack(adapter) -> List[Dict[str, str]]:
-    """List Slack channels the bot has joined."""
-    # Slack adapter may expose a web client
-    client = getattr(adapter, "_app", None) or getattr(adapter, "_client", None)
-    if not client:
+async def _build_slack(adapter) -> List[Dict[str, Any]]:
+    """List Slack channels the bot has joined across all workspaces.
+
+    Uses ``users.conversations`` against each workspace's web client. Pulls
+    public + private channels the bot is a member of, then merges in DMs
+    discovered from session history (IMs aren't useful to enumerate
+    proactively).
+    """
+    team_clients = getattr(adapter, "_team_clients", None) or {}
+    if not team_clients:
         return _build_from_sessions("slack")
 
-    try:
-        from tools.send_message_tool import _send_slack  # noqa: F401
-        # Use the Slack Web API directly if available
-    except Exception:
-        pass
+    channels: List[Dict[str, Any]] = []
+    seen_ids: set = set()
 
-    # Fallback to session data
-    return _build_from_sessions("slack")
+    for team_id, client in team_clients.items():
+        try:
+            cursor: Optional[str] = None
+            for _page in range(20):  # safety cap on pagination
+                response = await client.users_conversations(
+                    types="public_channel,private_channel",
+                    exclude_archived=True,
+                    limit=200,
+                    cursor=cursor,
+                )
+                if not response.get("ok"):
+                    logger.warning(
+                        "Channel directory: users.conversations not ok for team %s: %s",
+                        team_id,
+                        response.get("error", "unknown"),
+                    )
+                    break
+                for ch in response.get("channels", []):
+                    cid = ch.get("id")
+                    name = ch.get("name")
+                    if not cid or not name or cid in seen_ids:
+                        continue
+                    seen_ids.add(cid)
+                    channels.append({
+                        "id": cid,
+                        "name": name,
+                        "type": "private" if ch.get("is_private") else "channel",
+                    })
+                cursor = (response.get("response_metadata") or {}).get("next_cursor")
+                if not cursor:
+                    break
+        except Exception as e:
+            logger.warning(
+                "Channel directory: failed to list Slack channels for team %s: %s",
+                team_id, e,
+            )
+            continue
+
+    # Merge in DM/group entries discovered from session history.
+    for entry in _build_from_sessions("slack"):
+        if entry.get("id") not in seen_ids:
+            channels.append(entry)
+            seen_ids.add(entry.get("id"))
+
+    return channels
 
 
 def _build_from_sessions(platform_name: str) -> List[Dict[str, str]]:
@@ -223,6 +268,14 @@ def resolve_channel_name(platform_name: str, name: str) -> Optional[str]:
     if not channels:
         return None
 
+    # 0. Exact ID match — case-sensitive, no normalization. Lets callers pass
+    # raw platform IDs (e.g. Slack "C0B0QV5434G") even when the format guard
+    # in _parse_target_ref hasn't recognized them as explicit.
+    raw = name.strip()
+    for ch in channels:
+        if ch.get("id") == raw:
+            return ch["id"]
+
     query = _normalize_channel_query(name)
 
     # 1. Exact name match, including the display labels shown by send_message(action="list")

gateway/config.py

@@ -67,6 +67,7 @@ class Platform(Enum):
     WEIXIN = "weixin"
     BLUEBUBBLES = "bluebubbles"
     QQBOT = "qqbot"
+    YUANBAO = "yuanbao"
 
 
 @dataclass
@@ -195,6 +196,14 @@ class StreamingConfig:
     edit_interval: float = 1.0    # Seconds between message edits (Telegram rate-limits at ~1/s)
     buffer_threshold: int = 40    # Chars before forcing an edit
     cursor: str = " ▉"           # Cursor shown during streaming
+    # Ported from openclaw/openclaw#72038.  When >0, the final edit for
+    # a long-running streamed response is delivered as a fresh message
+    # if the original preview has been visible for at least this many
+    # seconds, so the platform's visible timestamp reflects completion
+    # time instead of the preview creation time.  Currently applied to
+    # Telegram only (other platforms ignore the setting).  Default 60s
+    # matches the OpenClaw rollout.  Set to 0 to disable.
+    fresh_final_after_seconds: float = 60.0
 
     def to_dict(self) -> Dict[str, Any]:
         return {
@@ -203,6 +212,7 @@ class StreamingConfig:
             "edit_interval": self.edit_interval,
             "buffer_threshold": self.buffer_threshold,
             "cursor": self.cursor,
+            "fresh_final_after_seconds": self.fresh_final_after_seconds,
         }
 
     @classmethod
@@ -215,6 +225,9 @@ class StreamingConfig:
             edit_interval=float(data.get("edit_interval", 1.0)),
             buffer_threshold=int(data.get("buffer_threshold", 40)),
             cursor=data.get("cursor", " ▉"),
+            fresh_final_after_seconds=float(
+                data.get("fresh_final_after_seconds", 60.0)
+            ),
         )
 
 
@@ -314,6 +327,9 @@ class GatewayConfig:
             # QQBot uses extra dict for app credentials
             elif platform == Platform.QQBOT and config.extra.get("app_id") and config.extra.get("client_secret"):
                 connected.append(platform)
+            # Yuanbao uses extra dict for app credentials
+            elif platform == Platform.YUANBAO and config.extra.get("app_id") and config.extra.get("app_secret"):
+                connected.append(platform)
             # DingTalk uses client_id/client_secret from config.extra or env vars
             elif platform == Platform.DINGTALK and (
                 config.extra.get("client_id") or os.getenv("DINGTALK_CLIENT_ID")
@@ -570,6 +586,8 @@ def load_gateway_config() -> GatewayConfig:
                     )
                 if "reply_prefix" in platform_cfg:
                     bridged["reply_prefix"] = platform_cfg["reply_prefix"]
+                if "reply_in_thread" in platform_cfg:
+                    bridged["reply_in_thread"] = platform_cfg["reply_in_thread"]
                 if "require_mention" in platform_cfg:
                     bridged["require_mention"] = platform_cfg["require_mention"]
                 if "free_response_channels" in platform_cfg:
@@ -584,7 +602,7 @@ def load_gateway_config() -> GatewayConfig:
                     bridged["group_policy"] = platform_cfg["group_policy"]
                 if "group_allow_from" in platform_cfg:
                     bridged["group_allow_from"] = platform_cfg["group_allow_from"]
-                if plat == Platform.DISCORD and "channel_skill_bindings" in platform_cfg:
+                if plat in (Platform.DISCORD, Platform.SLACK) and "channel_skill_bindings" in platform_cfg:
                     bridged["channel_skill_bindings"] = platform_cfg["channel_skill_bindings"]
                 if "channel_prompts" in platform_cfg:
                     channel_prompts = platform_cfg["channel_prompts"]
@@ -609,6 +627,8 @@ def load_gateway_config() -> GatewayConfig:
             if isinstance(slack_cfg, dict):
                 if "require_mention" in slack_cfg and not os.getenv("SLACK_REQUIRE_MENTION"):
                     os.environ["SLACK_REQUIRE_MENTION"] = str(slack_cfg["require_mention"]).lower()
+                if "strict_mention" in slack_cfg and not os.getenv("SLACK_STRICT_MENTION"):
+                    os.environ["SLACK_STRICT_MENTION"] = str(slack_cfg["strict_mention"]).lower()
                 if "allow_bots" in slack_cfg and not os.getenv("SLACK_ALLOW_BOTS"):
                     os.environ["SLACK_ALLOW_BOTS"] = str(slack_cfg["allow_bots"]).lower()
                 frc = slack_cfg.get("free_response_channels")
@@ -918,8 +938,12 @@ def _apply_env_overrides(config: GatewayConfig) -> None:
     slack_token = os.getenv("SLACK_BOT_TOKEN")
     if slack_token:
         if Platform.SLACK not in config.platforms:
+            # No yaml config for Slack — env-only setup, enable it
             config.platforms[Platform.SLACK] = PlatformConfig()
-        config.platforms[Platform.SLACK].enabled = True
+            config.platforms[Platform.SLACK].enabled = True
+        # If yaml config exists, respect its enabled flag (don't override
+        # explicit enabled: false). Token is still stored so skills that
+        # send Slack messages can use it without activating the gateway adapter.
         config.platforms[Platform.SLACK].token = slack_token
     slack_home = os.getenv("SLACK_HOME_CHANNEL")
     if slack_home and Platform.SLACK in config.platforms:
@@ -1276,6 +1300,48 @@ def _apply_env_overrides(config: GatewayConfig) -> None:
                 name=os.getenv("QQBOT_HOME_CHANNEL_NAME") or os.getenv(qq_home_name_env, "Home"),
             )
 
+    # Yuanbao — YUANBAO_APP_ID preferred
+    yuanbao_app_id = os.getenv("YUANBAO_APP_ID") or os.getenv("YUANBAO_APP_KEY")
+    yuanbao_app_secret = os.getenv("YUANBAO_APP_SECRET")
+    if yuanbao_app_id and yuanbao_app_secret:
+        if Platform.YUANBAO not in config.platforms:
+            config.platforms[Platform.YUANBAO] = PlatformConfig()
+        config.platforms[Platform.YUANBAO].enabled = True
+        extra = config.platforms[Platform.YUANBAO].extra
+        extra["app_id"] = yuanbao_app_id
+        extra["app_secret"] = yuanbao_app_secret
+        yuanbao_bot_id = os.getenv("YUANBAO_BOT_ID")
+        if yuanbao_bot_id:
+            extra["bot_id"] = yuanbao_bot_id
+        yuanbao_ws_url = os.getenv("YUANBAO_WS_URL")
+        if yuanbao_ws_url:
+            extra["ws_url"] = yuanbao_ws_url
+        yuanbao_api_domain = os.getenv("YUANBAO_API_DOMAIN")
+        if yuanbao_api_domain:
+            extra["api_domain"] = yuanbao_api_domain
+        yuanbao_route_env = os.getenv("YUANBAO_ROUTE_ENV")
+        if yuanbao_route_env:
+            extra["route_env"] = yuanbao_route_env
+        yuanbao_home = os.getenv("YUANBAO_HOME_CHANNEL")
+        if yuanbao_home:
+            config.platforms[Platform.YUANBAO].home_channel = HomeChannel(
+                platform=Platform.YUANBAO,
+                chat_id=yuanbao_home,
+                name=os.getenv("YUANBAO_HOME_CHANNEL_NAME", "Home"),
+            )
+        yuanbao_dm_policy = os.getenv("YUANBAO_DM_POLICY")
+        if yuanbao_dm_policy:
+            extra["dm_policy"] = yuanbao_dm_policy.strip().lower()
+        yuanbao_dm_allow_from = os.getenv("YUANBAO_DM_ALLOW_FROM")
+        if yuanbao_dm_allow_from:
+            extra["dm_allow_from"] = yuanbao_dm_allow_from
+        yuanbao_group_policy = os.getenv("YUANBAO_GROUP_POLICY")
+        if yuanbao_group_policy:
+            extra["group_policy"] = yuanbao_group_policy.strip().lower()
+        yuanbao_group_allow_from = os.getenv("YUANBAO_GROUP_ALLOW_FROM")
+        if yuanbao_group_allow_from:
+            extra["group_allow_from"] = yuanbao_group_allow_from
+
     # Session settings
     idle_minutes = os.getenv("SESSION_IDLE_MINUTES")
     if idle_minutes:

gateway/display_config.py

@@ -79,7 +79,9 @@ _PLATFORM_DEFAULTS: dict[str, dict[str, Any]] = {
     "discord":     _TIER_HIGH,
 
     # Tier 2 — edit support, often customer/workspace channels
-    "slack":           _TIER_MEDIUM,
+    # Slack: tool_progress off by default — Bolt posts cannot be edited like CLI;
+    # "new"/"all" spam permanent lines in channels (hermes-agent#14663).
+    "slack":           {**_TIER_MEDIUM, "tool_progress": "off"},
     "mattermost":      _TIER_MEDIUM,
     "matrix":          _TIER_MEDIUM,
     "feishu":          _TIER_MEDIUM,

gateway/mirror.py

@@ -28,6 +28,7 @@ def mirror_to_session(
     message_text: str,
     source_label: str = "cli",
     thread_id: Optional[str] = None,
+    user_id: Optional[str] = None,
 ) -> bool:
     """
     Append a delivery-mirror message to the target session's transcript.
@@ -39,9 +40,20 @@ def mirror_to_session(
     All errors are caught -- this is never fatal.
     """
     try:
-        session_id = _find_session_id(platform, str(chat_id), thread_id=thread_id)
+        session_id = _find_session_id(
+            platform,
+            str(chat_id),
+            thread_id=thread_id,
+            user_id=user_id,
+        )
         if not session_id:
-            logger.debug("Mirror: no session found for %s:%s:%s", platform, chat_id, thread_id)
+            logger.debug(
+                "Mirror: no session found for %s:%s:%s:%s",
+                platform,
+                chat_id,
+                thread_id,
+                user_id,
+            )
             return False
 
         mirror_msg = {
@@ -59,17 +71,33 @@ def mirror_to_session(
         return True
 
     except Exception as e:
-        logger.debug("Mirror failed for %s:%s:%s: %s", platform, chat_id, thread_id, e)
+        logger.debug(
+            "Mirror failed for %s:%s:%s:%s: %s",
+            platform,
+            chat_id,
+            thread_id,
+            user_id,
+            e,
+        )
         return False
 
 
-def _find_session_id(platform: str, chat_id: str, thread_id: Optional[str] = None) -> Optional[str]:
+def _find_session_id(
+    platform: str,
+    chat_id: str,
+    thread_id: Optional[str] = None,
+    user_id: Optional[str] = None,
+) -> Optional[str]:
     """
     Find the active session_id for a platform + chat_id pair.
 
     Scans sessions.json entries and matches where origin.chat_id == chat_id
     on the right platform.  DM session keys don't embed the chat_id
     (e.g. "agent:main:telegram:dm"), so we check the origin dict.
+
+    When *user_id* is provided, prefer exact sender matches. If multiple
+    same-chat candidates exist and none matches the user, return None instead
+    of guessing and contaminating another participant's session.
     """
     if not _SESSIONS_INDEX.exists():
         return None
@@ -81,8 +109,7 @@ def _find_session_id(platform: str, chat_id: str, thread_id: Optional[str] = Non
         return None
 
     platform_lower = platform.lower()
-    best_match = None
-    best_updated = ""
+    candidates = []
 
     for _key, entry in data.items():
         origin = entry.get("origin") or {}
@@ -96,12 +123,31 @@ def _find_session_id(platform: str, chat_id: str, thread_id: Optional[str] = Non
             origin_thread_id = origin.get("thread_id")
             if thread_id is not None and str(origin_thread_id or "") != str(thread_id):
                 continue
-            updated = entry.get("updated_at", "")
-            if updated > best_updated:
-                best_updated = updated
-                best_match = entry.get("session_id")
+            candidates.append(entry)
 
-    return best_match
+    if not candidates:
+        return None
+
+    if user_id:
+        exact_user_matches = [
+            entry for entry in candidates
+            if str((entry.get("origin") or {}).get("user_id") or "") == str(user_id)
+        ]
+        if exact_user_matches:
+            candidates = exact_user_matches
+        elif len(candidates) > 1:
+            return None
+    elif len(candidates) > 1:
+        distinct_user_ids = {
+            str((entry.get("origin") or {}).get("user_id") or "").strip()
+            for entry in candidates
+            if str((entry.get("origin") or {}).get("user_id") or "").strip()
+        }
+        if len(distinct_user_ids) > 1:
+            return None
+
+    best_entry = max(candidates, key=lambda entry: entry.get("updated_at", ""))
+    return best_entry.get("session_id")
 
 
 def _append_to_jsonl(session_id: str, message: dict) -> None:

gateway/platforms/__init__.py

@@ -10,10 +10,12 @@ Each adapter handles:
 
 from .base import BasePlatformAdapter, MessageEvent, SendResult
 from .qqbot import QQAdapter
+from .yuanbao import YuanbaoAdapter
 
 __all__ = [
     "BasePlatformAdapter",
     "MessageEvent",
     "SendResult",
     "QQAdapter",
+    "YuanbaoAdapter",
 ]

gateway/platforms/base.py

@@ -336,6 +336,39 @@ def proxy_kwargs_for_aiohttp(proxy_url: str | None) -> tuple[dict, dict]:
     return {}, {"proxy": proxy_url}
 
 
+def is_host_excluded_by_no_proxy(hostname: str, no_proxy_value: str | None = None) -> bool:
+    """Return True when ``hostname`` matches a ``NO_PROXY`` entry.
+
+    Supports comma- or whitespace-separated entries with optional leading dots
+    and ``*.`` wildcards, which match both the apex domain and subdomains.
+    """
+    raw = no_proxy_value
+    if raw is None:
+        raw = os.environ.get("NO_PROXY") or os.environ.get("no_proxy") or ""
+
+    raw = raw.strip()
+    if not raw:
+        return False
+
+    lower_hostname = hostname.lower()
+    for entry in re.split(r"[\s,]+", raw):
+        normalized = entry.strip().lower()
+        if not normalized:
+            continue
+        if normalized == "*":
+            return True
+
+        if normalized.startswith("*."):
+            normalized = normalized[2:]
+        elif normalized.startswith("."):
+            normalized = normalized[1:]
+
+        if lower_hostname == normalized or lower_hostname.endswith(f".{normalized}"):
+            return True
+
+    return False
+
+
 from dataclasses import dataclass, field
 from datetime import datetime
 from pathlib import Path
@@ -693,7 +726,15 @@ SUPPORTED_DOCUMENT_TYPES = {
     ".pdf": "application/pdf",
     ".md": "text/markdown",
     ".txt": "text/plain",
+    ".csv": "text/csv",
     ".log": "text/plain",
+    ".json": "application/json",
+    ".xml": "application/xml",
+    ".yaml": "application/yaml",
+    ".yml": "application/yaml",
+    ".toml": "application/toml",
+    ".ini": "text/plain",
+    ".cfg": "text/plain",
     ".zip": "application/zip",
     ".docx": "application/vnd.openxmlformats-officedocument.wordprocessingml.document",
     ".xlsx": "application/vnd.openxmlformats-officedocument.spreadsheetml.sheet",
@@ -982,6 +1023,61 @@ def resolve_channel_prompt(
     return None
 
 
+def resolve_channel_skills(
+    config_extra: dict,
+    channel_id: str,
+    parent_id: str | None = None,
+) -> list[str] | None:
+    """Resolve auto-loaded skill(s) for a channel/thread from platform config.
+
+    Looks up ``channel_skill_bindings`` in the adapter's ``config.extra`` dict.
+
+    Config format::
+
+        channel_skill_bindings:
+          - id: "C0123"          # Slack channel ID or Discord channel/forum ID
+            skills: ["skill-a", "skill-b"]
+          - id: "D0ABCDE"
+            skill: "solo-skill"  # single string also accepted
+
+    Prefers an exact match on *channel_id*; falls back to *parent_id*
+    (useful for forum threads / Slack threads inheriting the parent channel's
+    binding).
+
+    Returns a deduplicated list of skill names (order preserved), or None if
+    no match is found.
+    """
+    bindings = config_extra.get("channel_skill_bindings") or []
+    if not isinstance(bindings, list) or not bindings:
+        return None
+    ids_to_check: set[str] = set()
+    if channel_id:
+        ids_to_check.add(str(channel_id))
+    if parent_id:
+        ids_to_check.add(str(parent_id))
+    if not ids_to_check:
+        return None
+    for entry in bindings:
+        if not isinstance(entry, dict):
+            continue
+        entry_id = str(entry.get("id", ""))
+        if entry_id in ids_to_check:
+            skills = entry.get("skills") or entry.get("skill")
+            if isinstance(skills, str):
+                s = skills.strip()
+                return [s] if s else None
+            if isinstance(skills, list) and skills:
+                seen: list[str] = []
+                for name in skills:
+                    if not isinstance(name, str):
+                        continue
+                    nm = name.strip()
+                    if nm and nm not in seen:
+                        seen.append(nm)
+                return seen or None
+    return None
+
+
 class BasePlatformAdapter(ABC):
     """
     Base class for platform adapters.
@@ -1258,6 +1354,27 @@ class BasePlatformAdapter(ABC):
         """
         return SendResult(success=False, error="Not supported")
 
+    async def delete_message(
+        self,
+        chat_id: str,
+        message_id: str,
+    ) -> bool:
+        """
+        Delete a previously sent message.  Optional — platforms that don't
+        support deletion return ``False`` and callers fall back to leaving
+        the message in place.
+
+        Used by the stream consumer's fresh-final cleanup path (see
+        openclaw/openclaw#72038) to remove long-lived preview messages
+        after sending the completed reply as a fresh message so the
+        platform's visible timestamp reflects completion time.
+
+        Returns ``True`` on successful deletion, ``False`` otherwise.
+        Subclasses should override for platforms with a deletion API
+        (e.g. Telegram ``deleteMessage``).
+        """
+        return False
+
     async def send_typing(self, chat_id: str, metadata=None) -> None:
         """
         Send a typing indicator.

gateway/platforms/discord.py

@@ -2679,21 +2679,8 @@ class DiscordAdapter(BasePlatformAdapter):
                 skills: ["skill-a", "skill-b"]
         Also checks parent_id so forum threads inherit the forum's bindings.
         """
-        bindings = self.config.extra.get("channel_skill_bindings", [])
-        if not bindings:
-            return None
-        ids_to_check = {channel_id}
-        if parent_id:
-            ids_to_check.add(parent_id)
-        for entry in bindings:
-            entry_id = str(entry.get("id", ""))
-            if entry_id in ids_to_check:
-                skills = entry.get("skills") or entry.get("skill")
-                if isinstance(skills, str):
-                    return [skills]
-                if isinstance(skills, list) and skills:
-                    return list(dict.fromkeys(skills))  # dedup, preserve order
-        return None
+        from gateway.platforms.base import resolve_channel_skills
+        return resolve_channel_skills(self.config.extra, channel_id, parent_id)
 
     def _resolve_channel_prompt(self, channel_id: str, parent_id: str | None = None) -> str | None:
         """Resolve a Discord per-channel prompt, preferring the exact channel over its parent."""

gateway/platforms/helpers.py

@@ -57,6 +57,15 @@ class MessageDeduplicator:
         if len(self._seen) > self._max_size:
             cutoff = now - self._ttl
             self._seen = {k: v for k, v in self._seen.items() if v > cutoff}
+            if len(self._seen) > self._max_size:
+                # TTL pruning alone does not cap the cache when every entry is
+                # still fresh. Keep the newest entries so the helper's
+                # max_size bound is enforced under sustained traffic.
+                newest = sorted(
+                    self._seen.items(),
+                    key=lambda item: item[1],
+                )[-self._max_size:]
+                self._seen = dict(newest)
         return False
 
     def clear(self):

gateway/platforms/matrix.py

@@ -1178,13 +1178,83 @@ class MatrixAdapter(BasePlatformAdapter):
     # Event callbacks
     # ------------------------------------------------------------------
 
+    def _is_self_sender(self, sender: str) -> bool:
+        """Return True if the sender refers to the bot's own account.
+
+        Matrix user IDs are byte-compared after trimming whitespace and
+        lowercasing — some homeservers normalize the localpart case
+        differently at different API surfaces, and the reply-loop tail
+        of the "hall of mirrors" bug (#15763) has been observed with the
+        bot's own account bypassing a case-sensitive equality check.
+
+        When ``self._user_id`` is empty (whoami hasn't resolved yet, or
+        login failed), we cannot prove a sender is NOT us, so we return
+        True defensively — an unidentified bot dropping its own events
+        is always preferable to falling into an echo loop.
+        """
+        own = (self._user_id or "").strip().lower()
+        if not own:
+            return True
+        return sender.strip().lower() == own
+
+    @staticmethod
+    def _is_system_or_bridge_sender(sender: str) -> bool:
+        """Return True if the sender looks like a system / bridge / appservice
+        identity rather than a real user.
+
+        Appservice namespaces on Matrix conventionally prefix bot / puppet
+        user IDs with an underscore (e.g. ``@_telegram_12345:server``,
+        ``@_discord_999:server``, ``@_slack_...:server``).  Server-notices
+        bots and bridge-controller bots on many homeservers use the same
+        pattern.
+
+        We treat these as system identities for pairing purposes: they
+        should never be offered a pairing code, because an operator
+        approving the code would hand the bridge itself permanent
+        authorization — and every outbound message relayed by the bridge
+        would then loop back into the agent as an "authorized user
+        message", which is the root of issue #15763.
+
+        Matches:
+            ``@_something:server``   — appservice namespace convention
+            ``@:server``             — malformed / empty localpart
+            ``:server``              — malformed, no leading ``@``
+        """
+        s = (sender or "").strip()
+        if not s:
+            return True
+        # Localpart is everything between leading '@' and ':'
+        if s.startswith("@"):
+            s = s[1:]
+        if ":" in s:
+            localpart, _, _ = s.partition(":")
+        else:
+            localpart = s
+        if not localpart:
+            return True
+        return localpart.startswith("_")
+
     async def _on_room_message(self, event: Any) -> None:
         """Handle incoming room message events (text, media)."""
         room_id = str(getattr(event, "room_id", ""))
         sender = str(getattr(event, "sender", ""))
 
-        # Ignore own messages.
-        if sender == self._user_id:
+        # Ignore own messages (case-insensitive; also drops when our own
+        # user_id hasn't been resolved yet — see _is_self_sender docstring
+        # and issue #15763).
+        if self._is_self_sender(sender):
+            return
+
+        # Ignore appservice / bridge / system identities so they never
+        # trigger the pairing flow.  Once a bridge user is paired, every
+        # outbound message it relays would loop back as an authorized
+        # user message (the "hall of mirrors" in #15763).
+        if self._is_system_or_bridge_sender(sender):
+            logger.debug(
+                "Matrix: ignoring system/bridge sender %s in %s",
+                sender,
+                room_id,
+            )
             return
 
         # Deduplicate by event ID.
@@ -1654,7 +1724,7 @@ class MatrixAdapter(BasePlatformAdapter):
     async def _on_reaction(self, event: Any) -> None:
         """Handle incoming reaction events."""
         sender = str(getattr(event, "sender", ""))
-        if sender == self._user_id:
+        if self._is_self_sender(sender):
             return
         event_id = str(getattr(event, "event_id", ""))
         if self._is_duplicate_event(event_id):

gateway/platforms/telegram.py

@@ -1209,6 +1209,31 @@ class TelegramAdapter(BasePlatformAdapter):
             )
             return SendResult(success=False, error=str(e))
 
+    async def delete_message(self, chat_id: str, message_id: str) -> bool:
+        """Delete a previously sent Telegram message.
+
+        Used by the stream consumer's fresh-final cleanup path (ported
+        from openclaw/openclaw#72038) to remove long-lived preview
+        messages after sending the completed reply as a fresh message.
+        Telegram's Bot API ``deleteMessage`` works for bot-posted
+        messages in the last 48 hours.  Failures are non-fatal — the
+        caller leaves the preview in place and logs at debug level.
+        """
+        if not self._bot:
+            return False
+        try:
+            await self._bot.delete_message(
+                chat_id=int(chat_id),
+                message_id=int(message_id),
+            )
+            return True
+        except Exception as e:
+            logger.debug(
+                "[%s] Failed to delete Telegram message %s: %s",
+                self.name, message_id, e,
+            )
+            return False
+
     async def send_update_prompt(
         self, chat_id: str, prompt: str, default: str = "",
         session_key: str = "",

gateway/platforms/yuanbao_media.py

@@ -0,0 +1,647 @@
+"""
+yuanbao_media.py — 元宝平台媒体处理模块
+
+提供 COS 上传、文件下载、TIM 媒体消息构建等功能。
+移植自 TypeScript 版 media.ts（yuanbao-openclaw-plugin），
+使用 httpx 替代 cos-nodejs-sdk-v5，避免引入额外 SDK 依赖。
+
+COS 上传流程：
+  1. 调用 genUploadInfo 获取临时凭证（tmpSecretId/tmpSecretKey/sessionToken）
+  2. 用临时凭证通过 HMAC-SHA1 签名构建 Authorization 头
+  3. HTTP PUT 上传到 COS
+
+TIM 消息体构建：
+  - buildImageMsgBody() → TIMImageElem
+  - buildFileMsgBody()  → TIMFileElem
+"""
+
+from __future__ import annotations
+
+import hashlib
+import hmac
+import logging
+import os
+import re
+import secrets
+import struct
+import time
+import urllib.parse
+from datetime import datetime, timezone, timedelta
+from typing import Optional, Any
+
+import httpx
+
+logger = logging.getLogger(__name__)
+
+# ============ 常量 ============
+
+UPLOAD_INFO_PATH = "/api/resource/genUploadInfo"
+DEFAULT_API_DOMAIN = "yuanbao.tencent.com"
+DEFAULT_MAX_SIZE_MB = 50
+
+# COS 加速域名后缀（优先使用全球加速）
+COS_USE_ACCELERATE = True
+
+# ============ 类型映射 ============
+
+# MIME → image_format 数字（TIM 协议字段）
+_MIME_TO_IMAGE_FORMAT: dict[str, int] = {
+    "image/jpeg": 1,
+    "image/jpg": 1,
+    "image/gif": 2,
+    "image/png": 3,
+    "image/bmp": 4,
+    "image/webp": 255,
+    "image/heic": 255,
+    "image/tiff": 255,
+}
+
+# 文件扩展名 → MIME
+_EXT_TO_MIME: dict[str, str] = {
+    ".jpg": "image/jpeg",
+    ".jpeg": "image/jpeg",
+    ".png": "image/png",
+    ".gif": "image/gif",
+    ".webp": "image/webp",
+    ".bmp": "image/bmp",
+    ".heic": "image/heic",
+    ".tiff": "image/tiff",
+    ".ico": "image/x-icon",
+    ".pdf": "application/pdf",
+    ".doc": "application/msword",
+    ".docx": "application/vnd.openxmlformats-officedocument.wordprocessingml.document",
+    ".xls": "application/vnd.ms-excel",
+    ".xlsx": "application/vnd.openxmlformats-officedocument.spreadsheetml.sheet",
+    ".ppt": "application/vnd.ms-powerpoint",
+    ".pptx": "application/vnd.openxmlformats-officedocument.presentationml.presentation",
+    ".txt": "text/plain",
+    ".zip": "application/zip",
+    ".tar": "application/x-tar",
+    ".gz": "application/gzip",
+    ".mp3": "audio/mpeg",
+    ".mp4": "video/mp4",
+    ".wav": "audio/wav",
+    ".ogg": "audio/ogg",
+    ".webm": "video/webm",
+}
+
+
+# ============ 工具函数 ============
+
+def guess_mime_type(filename: str) -> str:
+    """根据文件扩展名猜测 MIME 类型。"""
+    ext = os.path.splitext(filename)[-1].lower()
+    return _EXT_TO_MIME.get(ext, "application/octet-stream")
+
+
+def is_image(filename: str, mime_type: str = "") -> bool:
+    """判断是否为图片类型。"""
+    if mime_type.startswith("image/"):
+        return True
+    ext = os.path.splitext(filename)[-1].lower()
+    return ext in {".jpg", ".jpeg", ".png", ".gif", ".webp", ".bmp", ".heic", ".tiff", ".ico"}
+
+
+def get_image_format(mime_type: str) -> int:
+    """获取 TIM 图片格式编号。"""
+    return _MIME_TO_IMAGE_FORMAT.get(mime_type.lower(), 255)
+
+
+def md5_hex(data: bytes) -> str:
+    """计算 MD5 十六进制摘要。"""
+    return hashlib.md5(data).hexdigest()
+
+
+def generate_file_id() -> str:
+    """生成随机文件 ID（32 位 hex）。"""
+    return secrets.token_hex(16)
+
+
+
+# ============ 图片尺寸解析（纯 Python，无需 Pillow） ============
+
+def parse_image_size(data: bytes) -> Optional[dict[str, int]]:
+    """
+    解析图片宽高（支持 JPEG/PNG/GIF/WebP），无需第三方依赖。
+    返回 {"width": w, "height": h} 或 None（无法识别）。
+    """
+    return (
+        _parse_png_size(data)
+        or _parse_jpeg_size(data)
+        or _parse_gif_size(data)
+        or _parse_webp_size(data)
+    )
+
+
+def _parse_png_size(buf: bytes) -> Optional[dict[str, int]]:
+    if len(buf) < 24:
+        return None
+    if buf[:4] != b"\x89PNG":
+        return None
+    w = struct.unpack(">I", buf[16:20])[0]
+    h = struct.unpack(">I", buf[20:24])[0]
+    return {"width": w, "height": h}
+
+
+def _parse_jpeg_size(buf: bytes) -> Optional[dict[str, int]]:
+    if len(buf) < 4 or buf[0] != 0xFF or buf[1] != 0xD8:
+        return None
+    i = 2
+    while i < len(buf) - 9:
+        if buf[i] != 0xFF:
+            i += 1
+            continue
+        marker = buf[i + 1]
+        if marker in (0xC0, 0xC2):
+            h = struct.unpack(">H", buf[i + 5: i + 7])[0]
+            w = struct.unpack(">H", buf[i + 7: i + 9])[0]
+            return {"width": w, "height": h}
+        if i + 3 < len(buf):
+            i += 2 + struct.unpack(">H", buf[i + 2: i + 4])[0]
+        else:
+            break
+    return None
+
+
+def _parse_gif_size(buf: bytes) -> Optional[dict[str, int]]:
+    if len(buf) < 10:
+        return None
+    sig = buf[:6].decode("ascii", errors="replace")
+    if sig not in ("GIF87a", "GIF89a"):
+        return None
+    w = struct.unpack("<H", buf[6:8])[0]
+    h = struct.unpack("<H", buf[8:10])[0]
+    return {"width": w, "height": h}
+
+
+def _parse_webp_size(buf: bytes) -> Optional[dict[str, int]]:
+    if len(buf) < 16:
+        return None
+    if buf[:4] != b"RIFF" or buf[8:12] != b"WEBP":
+        return None
+    chunk = buf[12:16].decode("ascii", errors="replace")
+    if chunk == "VP8 ":
+        if len(buf) >= 30 and buf[23] == 0x9D and buf[24] == 0x01 and buf[25] == 0x2A:
+            w = struct.unpack("<H", buf[26:28])[0] & 0x3FFF
+            h = struct.unpack("<H", buf[28:30])[0] & 0x3FFF
+            return {"width": w, "height": h}
+    elif chunk == "VP8L":
+        if len(buf) >= 25 and buf[20] == 0x2F:
+            bits = struct.unpack("<I", buf[21:25])[0]
+            w = (bits & 0x3FFF) + 1
+            h = ((bits >> 14) & 0x3FFF) + 1
+            return {"width": w, "height": h}
+    elif chunk == "VP8X":
+        if len(buf) >= 30:
+            w = (buf[24] | (buf[25] << 8) | (buf[26] << 16)) + 1
+            h = (buf[27] | (buf[28] << 8) | (buf[29] << 16)) + 1
+            return {"width": w, "height": h}
+    return None
+
+
+# ============ URL 下载 ============
+
+async def download_url(
+    url: str,
+    max_size_mb: int = DEFAULT_MAX_SIZE_MB,
+) -> tuple[bytes, str]:
+    """
+    下载 URL 内容，返回 (bytes, content_type)。
+
+    Args:
+        url:          HTTP(S) URL
+        max_size_mb:  最大允许大小（MB），超过则抛出异常
+
+    Returns:
+        (data_bytes, content_type_string)
+
+    Raises:
+        ValueError:  内容超过大小限制
+        httpx.HTTPError: 网络/HTTP 错误
+    """
+    max_bytes = max_size_mb * 1024 * 1024
+    async with httpx.AsyncClient(timeout=30.0, follow_redirects=True) as client:
+        # 先 HEAD 检查大小
+        try:
+            head = await client.head(url)
+            content_length = int(head.headers.get("content-length", 0) or 0)
+            if content_length > 0 and content_length > max_bytes:
+                raise ValueError(
+                    f"文件过大: {content_length / 1024 / 1024:.1f} MB > {max_size_mb} MB"
+                )
+        except httpx.HTTPStatusError:
+            pass  # 部分服务器不支持 HEAD，忽略
+
+        # GET 下载（流式读取，防止超限）
+        async with client.stream("GET", url) as resp:
+            resp.raise_for_status()
+
+            content_type = resp.headers.get("content-type", "").split(";")[0].strip()
+
+            chunks: list[bytes] = []
+            downloaded = 0
+            async for chunk in resp.aiter_bytes(65536):
+                downloaded += len(chunk)
+                if downloaded > max_bytes:
+                    raise ValueError(
+                        f"文件过大: 已超过 {max_size_mb} MB 限制"
+                    )
+                chunks.append(chunk)
+
+        data = b"".join(chunks)
+        return data, content_type
+
+
+# ============ COS 鉴权（HMAC-SHA1） ============
+
+def _cos_sign(
+    method: str,
+    path: str,
+    params: dict[str, str],
+    headers: dict[str, str],
+    secret_id: str,
+    secret_key: str,
+    start_time: Optional[int] = None,
+    expire_seconds: int = 3600,
+) -> str:
+    """
+    构建 COS 请求签名（q-sign-algorithm=sha1 方案）。
+    参考：https://cloud.tencent.com/document/product/436/7778
+
+    Args:
+        method:         HTTP 方法（小写，如 "put"）
+        path:           URL 路径（URL encode 后的小写）
+        params:         URL 查询参数 dict（用于签名）
+        headers:        参与签名的请求头 dict（key 需小写）
+        secret_id:      临时 SecretId（tmpSecretId）
+        secret_key:     临时 SecretKey（tmpSecretKey）
+        start_time:     签名起始 Unix 时间戳（默认 now）
+        expire_seconds: 签名有效期（秒，默认 3600）
+
+    Returns:
+        Authorization header 值（完整字符串）
+    """
+    now = int(time.time())
+    q_sign_time = f"{start_time or now};{(start_time or now) + expire_seconds}"
+
+    # Step 1: SignKey = HMAC-SHA1(SecretKey, q-sign-time)
+    sign_key = hmac.new(
+        secret_key.encode("utf-8"),
+        q_sign_time.encode("utf-8"),
+        hashlib.sha1,
+    ).hexdigest()
+
+    # Step 2: HttpString
+    # 参数和头部需按字典序排列，key 小写
+    sorted_params = sorted((k.lower(), urllib.parse.quote(str(v), safe="") ) for k, v in params.items())
+    sorted_headers = sorted((k.lower(), urllib.parse.quote(str(v), safe="") ) for k, v in headers.items())
+
+    url_param_list = ";".join(k for k, _ in sorted_params)
+    url_params = "&".join(f"{k}={v}" for k, v in sorted_params)
+    header_list = ";".join(k for k, _ in sorted_headers)
+    header_str = "&".join(f"{k}={v}" for k, v in sorted_headers)
+
+    http_string = "\n".join([
+        method.lower(),
+        path,
+        url_params,
+        header_str,
+        "",
+    ])
+
+    # Step 3: StringToSign = sha1 hash of HttpString
+    sha1_of_http = hashlib.sha1(http_string.encode("utf-8")).hexdigest()
+    string_to_sign = "\n".join([
+        "sha1",
+        q_sign_time,
+        sha1_of_http,
+        "",
+    ])
+
+    # Step 4: Signature = HMAC-SHA1(SignKey, StringToSign)
+    signature = hmac.new(
+        sign_key.encode("utf-8"),
+        string_to_sign.encode("utf-8"),
+        hashlib.sha1,
+    ).hexdigest()
+
+    return (
+        f"q-sign-algorithm=sha1"
+        f"&q-ak={secret_id}"
+        f"&q-sign-time={q_sign_time}"
+        f"&q-key-time={q_sign_time}"
+        f"&q-header-list={header_list}"
+        f"&q-url-param-list={url_param_list}"
+        f"&q-signature={signature}"
+    )
+
+
+# ============ 主要公开 API ============
+
+async def get_cos_credentials(
+    app_key: str,
+    api_domain: str,
+    token: str,
+    filename: str = "file",
+    file_id: Optional[str] = None,
+    bot_id: str = "",
+    route_env: str = "",
+) -> dict:
+    """
+    调用 genUploadInfo 接口获取 COS 临时密钥及上传配置。
+
+    Args:
+        app_key:        应用 Key（用于 X-ID 头）
+        api_domain:     API 域名（如 https://bot.yuanbao.tencent.com）
+        token:          当前有效的签票 token（X-Token 头）
+        filename:       待上传的文件名（含扩展名）
+        file_id:        客户端生成的唯一文件 ID（不传则自动生成）
+        bot_id:         Bot 账号 ID（用于 X-ID 头）
+
+    Returns:
+        COS 上传配置 dict，包含以下字段：
+            bucketName         (str)  — COS Bucket 名称
+            region             (str)  — COS 地域
+            location           (str)  — 上传 Key（对象路径）
+            encryptTmpSecretId (str)  — 临时 SecretId
+            encryptTmpSecretKey(str)  — 临时 SecretKey
+            encryptToken       (str)  — SessionToken
+            startTime          (int)  — 凭证起始时间戳（Unix）
+            expiredTime        (int)  — 凭证过期时间戳（Unix）
+            resourceUrl        (str)  — 上传后的公网访问 URL
+            resourceID         (str)  — 资源 ID（可选）
+
+    Raises:
+        RuntimeError: 接口返回非 0 code 或字段缺失
+    """
+    if file_id is None:
+        file_id = generate_file_id()
+
+    upload_url = f"{api_domain.rstrip('/')}{UPLOAD_INFO_PATH}"
+
+    headers = {
+        "Content-Type": "application/json",
+        "X-Token": token,
+        "X-ID": bot_id or app_key,
+        "X-Source": "web",
+    }
+    if route_env:
+        headers["X-Route-Env"] = route_env
+    body = {
+        "fileName": filename,
+        "fileId": file_id,
+        "docFrom": "localDoc",
+        "docOpenId": "",
+    }
+
+    async with httpx.AsyncClient(timeout=15.0) as client:
+        resp = await client.post(upload_url, json=body, headers=headers)
+        resp.raise_for_status()
+        result: dict[str, Any] = resp.json()
+
+    code = result.get("code")
+    if code != 0 and code is not None:
+        raise RuntimeError(
+            f"genUploadInfo 失败: code={code}, msg={result.get('msg', '')}"
+        )
+
+    data = result.get("data") or result
+    required_fields = ["bucketName", "location"]
+    missing = [f for f in required_fields if not data.get(f)]
+    if missing:
+        raise RuntimeError(
+            f"genUploadInfo 返回字段不完整: 缺少字段 {missing}"
+        )
+
+    return data
+
+
+async def upload_to_cos(
+    file_bytes: bytes,
+    filename: str,
+    content_type: str,
+    credentials: dict,
+    bucket: str,
+    region: str,
+) -> dict:
+    """
+    通过 httpx PUT 请求将文件上传到 COS。
+    使用临时凭证（tmpSecretId/tmpSecretKey/sessionToken）构建 HMAC-SHA1 签名。
+
+    Args:
+        file_bytes:   文件二进制内容
+        filename:     文件名（用于辅助计算 MIME、UUID）
+        content_type: MIME 类型（如 "image/jpeg"）
+        credentials:  get_cos_credentials() 返回的 dict，包含：
+                        encryptTmpSecretId  → tmpSecretId
+                        encryptTmpSecretKey → tmpSecretKey
+                        encryptToken        → sessionToken
+                        location            → COS key（对象路径）
+                        resourceUrl         → 上传后公网 URL
+                        startTime           → 凭证起始时间（Unix）
+                        expiredTime         → 凭证过期时间（Unix）
+        bucket:       COS Bucket 名称（如 chatbot-1234567890）
+        region:       COS 地域（如 ap-guangzhou）
+
+    Returns:
+        上传结果 dict，包含：
+            url       (str)           — COS 公网访问 URL
+            uuid      (str)           — 文件内容 MD5
+            size      (int)           — 文件大小（字节）
+            width     (int, optional) — 图片宽度（仅图片）
+            height    (int, optional) — 图片高度（仅图片）
+
+    Raises:
+        httpx.HTTPStatusError: COS 返回非 2xx 状态
+        RuntimeError:          credentials 字段缺失
+    """
+    secret_id: str = credentials.get("encryptTmpSecretId", "")
+    secret_key: str = credentials.get("encryptTmpSecretKey", "")
+    session_token: str = credentials.get("encryptToken", "")
+    cos_key: str = credentials.get("location", "")
+    resource_url: str = credentials.get("resourceUrl", "")
+    start_time: Optional[int] = credentials.get("startTime")
+    expired_time: Optional[int] = credentials.get("expiredTime")
+
+    if not secret_id or not secret_key or not cos_key:
+        raise RuntimeError(
+            f"COS credentials 不完整: secretId={bool(secret_id)}, "
+            f"secretKey={bool(secret_key)}, location={bool(cos_key)}"
+        )
+
+    # 构建 COS 上传 URL（优先使用全球加速域名）
+    if COS_USE_ACCELERATE:
+        cos_host = f"{bucket}.cos.accelerate.myqcloud.com"
+    else:
+        cos_host = f"{bucket}.cos.{region}.myqcloud.com"
+
+    # URL encode cos_key（保留 /）
+    encoded_key = urllib.parse.quote(cos_key, safe="/")
+    cos_url = f"https://{cos_host}/{encoded_key.lstrip('/')}"
+
+    # 确定 Content-Type
+    if not content_type or content_type == "application/octet-stream":
+        if is_image(filename):
+            content_type = guess_mime_type(filename)
+        else:
+            content_type = "application/octet-stream"
+
+    # 计算文件 MD5 + size
+    file_uuid = md5_hex(file_bytes)
+    file_size = len(file_bytes)
+
+    # 参与签名的请求头
+    sign_headers = {
+        "host": cos_host,
+        "content-type": content_type,
+        "x-cos-security-token": session_token,
+    }
+
+    # 计算签名有效期
+    now = int(time.time())
+    sign_start = start_time if start_time else now
+    sign_expire = (expired_time - now) if expired_time and expired_time > now else 3600
+
+    authorization = _cos_sign(
+        method="put",
+        path=f"/{encoded_key.lstrip('/')}",
+        params={},
+        headers=sign_headers,
+        secret_id=secret_id,
+        secret_key=secret_key,
+        start_time=sign_start,
+        expire_seconds=sign_expire,
+    )
+
+    put_headers = {
+        "Authorization": authorization,
+        "Content-Type": content_type,
+        "x-cos-security-token": session_token,
+    }
+
+    logger.info(
+        "COS PUT: bucket=%s region=%s key=%s size=%d mime=%s",
+        bucket, region, cos_key, file_size, content_type,
+    )
+
+    async with httpx.AsyncClient(timeout=120.0) as client:
+        resp = await client.put(
+            cos_url,
+            content=file_bytes,
+            headers=put_headers,
+        )
+        resp.raise_for_status()
+
+    # 解析图片尺寸（仅图片类型）
+    result: dict[str, Any] = {
+        "url": resource_url or cos_url,
+        "uuid": file_uuid,
+        "size": file_size,
+    }
+
+    if content_type.startswith("image/"):
+        size_info = parse_image_size(file_bytes)
+        if size_info:
+            result["width"] = size_info["width"]
+            result["height"] = size_info["height"]
+
+    logger.info(
+        "COS 上传成功: url=%s size=%d",
+        result["url"], file_size,
+    )
+    return result
+
+
+# ============ TIM 媒体消息构建 ============
+
+def build_image_msg_body(
+    url: str,
+    uuid: Optional[str] = None,
+    filename: Optional[str] = None,
+    size: int = 0,
+    width: int = 0,
+    height: int = 0,
+    mime_type: str = "",
+) -> list[dict]:
+    """
+    构建腾讯 IM TIMImageElem 消息体。
+    参考：https://cloud.tencent.com/document/product/269/2720
+
+    Args:
+        url:       图片公网访问 URL（COS resourceUrl）
+        uuid:      文件 UUID（MD5 或其他唯一标识）
+        filename:  文件名（uuid 为空时作为备用）
+        size:      文件大小（字节）
+        width:     图片宽度（像素）
+        height:    图片高度（像素）
+        mime_type: MIME 类型（用于确定 image_format）
+
+    Returns:
+        TIMImageElem 消息体列表（适合直接放入 msg_body）
+    """
+    _uuid = uuid or filename or _basename_from_url(url) or "image"
+    image_format = get_image_format(mime_type) if mime_type else 255
+
+    return [
+        {
+            "msg_type": "TIMImageElem",
+            "msg_content": {
+                "uuid": _uuid,
+                "image_format": image_format,
+                "image_info_array": [
+                    {
+                        "type": 1,       # 1 = 原图
+                        "size": size,
+                        "width": width,
+                        "height": height,
+                        "url": url,
+                    }
+                ],
+            },
+        }
+    ]
+
+
+def build_file_msg_body(
+    url: str,
+    filename: str,
+    uuid: Optional[str] = None,
+    size: int = 0,
+) -> list[dict]:
+    """
+    构建腾讯 IM TIMFileElem 消息体。
+    参考：https://cloud.tencent.com/document/product/269/2720
+
+    Args:
+        url:      文件公网访问 URL（COS resourceUrl）
+        filename: 文件名（含扩展名）
+        uuid:     文件 UUID（MD5 或其他唯一标识，不传则使用 filename）
+        size:     文件大小（字节）
+
+    Returns:
+        TIMFileElem 消息体列表（适合直接放入 msg_body）
+    """
+    _uuid = uuid or filename
+
+    return [
+        {
+            "msg_type": "TIMFileElem",
+            "msg_content": {
+                "uuid": _uuid,
+                "file_name": filename,
+                "file_size": size,
+                "url": url,
+            },
+        }
+    ]
+
+
+# ============ 内部工具 ============
+
+def _basename_from_url(url: str) -> str:
+    """从 URL 提取文件名。"""
+    try:
+        parsed = urllib.parse.urlparse(url)
+        return os.path.basename(parsed.path)
+    except Exception:
+        return ""

gateway/platforms/yuanbao_sticker.py

@@ -0,0 +1,558 @@
+"""
+Yuanbao sticker (TIMFaceElem) support.
+
+Ported from yuanbao-openclaw-plugin/src/sticker/.
+
+TIMFaceElem wire format:
+    {
+        "msg_type": "TIMFaceElem",
+        "msg_content": {
+            "index": 0,          # always 0 per Yuanbao convention
+            "data": "<json>",    # serialised sticker metadata
+        }
+    }
+
+The `data` field carries a JSON string with the sticker's metadata so the
+receiver can look up the correct asset in the emoji pack.
+"""
+
+from __future__ import annotations
+
+import json
+import random
+import re
+import unicodedata
+from typing import Optional
+
+# ---------------------------------------------------------------------------
+# Sticker catalogue – ported from builtin-stickers.json
+# Key   : canonical name (Chinese)
+# Value : {sticker_id, package_id, name, description, width, height, formats}
+# ---------------------------------------------------------------------------
+STICKER_MAP: dict[str, dict] = {
+    "六六六": {
+        "sticker_id": "278", "package_id": "1003", "name": "六六六",
+        "description": "666 厉害 牛 棒 绝了 好强 awesome",
+        "width": 128, "height": 128, "formats": "png",
+    },
+    "我想开了": {
+        "sticker_id": "262", "package_id": "1003", "name": "我想开了",
+        "description": "想开 佛系 释怀 顿悟 看淡了 无所谓",
+        "width": 128, "height": 128, "formats": "png",
+    },
+    "害羞": {
+        "sticker_id": "130", "package_id": "1003", "name": "害羞",
+        "description": "腼腆 不好意思 脸红 娇羞 羞涩 捂脸",
+        "width": 128, "height": 128, "formats": "png",
+    },
+    "比心": {
+        "sticker_id": "252", "package_id": "1003", "name": "比心",
+        "description": "笔芯 爱你 爱心手势 love heart 喜欢你",
+        "width": 128, "height": 128, "formats": "png",
+    },
+    "委屈": {
+        "sticker_id": "125", "package_id": "1003", "name": "委屈",
+        "description": "难过 想哭 可怜巴巴 瘪嘴 受伤 被欺负",
+        "width": 128, "height": 128, "formats": "png",
+    },
+    "亲亲": {
+        "sticker_id": "146", "package_id": "1003", "name": "亲亲",
+        "description": "么么 mua 亲一下 kiss 飞吻 啵",
+        "width": 128, "height": 128, "formats": "png",
+    },
+    "酷": {
+        "sticker_id": "131", "package_id": "1003", "name": "酷",
+        "description": "帅 墨镜 cool 高冷 有型 swagger",
+        "width": 128, "height": 128, "formats": "png",
+    },
+    "睡": {
+        "sticker_id": "145", "package_id": "1003", "name": "睡",
+        "description": "睡觉 困 zzZ 打盹 躺平 休眠 sleepy",
+        "width": 128, "height": 128, "formats": "png",
+    },
+    "发呆": {
+        "sticker_id": "152", "package_id": "1003", "name": "发呆",
+        "description": "懵 愣住 放空 呆滞 出神 脑子空白",
+        "width": 128, "height": 128, "formats": "png",
+    },
+    "可怜": {
+        "sticker_id": "157", "package_id": "1003", "name": "可怜",
+        "description": "卖萌 求饶 委屈巴巴 弱小 拜托 眼巴巴",
+        "width": 128, "height": 128, "formats": "png",
+    },
+    "摊手": {
+        "sticker_id": "200", "package_id": "1003", "name": "摊手",
+        "description": "无奈 没办法 耸肩 随便 那咋整 whatever",
+        "width": 128, "height": 128, "formats": "png",
+    },
+    "头大": {
+        "sticker_id": "213", "package_id": "1003", "name": "头大",
+        "description": "头疼 烦恼 郁闷 难搞 崩溃 一团乱",
+        "width": 128, "height": 128, "formats": "png",
+    },
+    "吓": {
+        "sticker_id": "256", "package_id": "1003", "name": "吓",
+        "description": "害怕 惊恐 震惊 吓一跳 恐怖 怂",
+        "width": 128, "height": 128, "formats": "png",
+    },
+    "吐血": {
+        "sticker_id": "203", "package_id": "1003", "name": "吐血",
+        "description": "无语 崩溃 被雷 内伤 一口老血 屮",
+        "width": 128, "height": 128, "formats": "png",
+    },
+    "哼": {
+        "sticker_id": "185", "package_id": "1003", "name": "哼",
+        "description": "傲娇 生气 不满 撇嘴 不理 赌气",
+        "width": 128, "height": 128, "formats": "png",
+    },
+    "嘿嘿": {
+        "sticker_id": "220", "package_id": "1003", "name": "嘿嘿",
+        "description": "坏笑 猥琐笑 偷笑 憨笑 得意 你懂的",
+        "width": 128, "height": 128, "formats": "png",
+    },
+    "头秃": {
+        "sticker_id": "218", "package_id": "1003", "name": "头秃",
+        "description": "程序员 加班 焦虑 没头发 秃了 肝爆",
+        "width": 128, "height": 128, "formats": "png",
+    },
+    "暗中观察": {
+        "sticker_id": "221", "package_id": "1003", "name": "暗中观察",
+        "description": "窥屏 潜水 偷偷看 角落 围观 屏住呼吸",
+        "width": 128, "height": 128, "formats": "png",
+    },
+    "我酸了": {
+        "sticker_id": "224", "package_id": "1003", "name": "我酸了",
+        "description": "嫉妒 柠檬精 羡慕 吃柠檬 眼红 恰柠檬",
+        "width": 128, "height": 128, "formats": "png",
+    },
+    "打call": {
+        "sticker_id": "246", "package_id": "1003", "name": "打call",
+        "description": "应援 加油 支持 喝彩 助威 call",
+        "width": 128, "height": 128, "formats": "png",
+    },
+    "庆祝": {
+        "sticker_id": "251", "package_id": "1003", "name": "庆祝",
+        "description": "祝贺 开心 耶 party 胜利 干杯",
+        "width": 128, "height": 128, "formats": "png",
+    },
+    "奋斗": {
+        "sticker_id": "151", "package_id": "1003", "name": "奋斗",
+        "description": "努力 加油 拼搏 冲 干劲 卷起来",
+        "width": 128, "height": 128, "formats": "png",
+    },
+    "惊讶": {
+        "sticker_id": "143", "package_id": "1003", "name": "惊讶",
+        "description": "震惊 哇 不敢相信 OMG 居然 这么离谱",
+        "width": 128, "height": 128, "formats": "png",
+    },
+    "疑问": {
+        "sticker_id": "144", "package_id": "1003", "name": "疑问",
+        "description": "问号 不懂 啥 为什么 啥情况 懵逼问",
+        "width": 128, "height": 128, "formats": "png",
+    },
+    "仔细分析": {
+        "sticker_id": "248", "package_id": "1003", "name": "仔细分析",
+        "description": "思考 推敲 认真 研究 琢磨 让我想想",
+        "width": 128, "height": 128, "formats": "png",
+    },
+    "撅嘴": {
+        "sticker_id": "184", "package_id": "1003", "name": "撅嘴",
+        "description": "嘟嘴 卖萌 不高兴 撒娇 嘴翘",
+        "width": 128, "height": 128, "formats": "png",
+    },
+    "泪奔": {
+        "sticker_id": "199", "package_id": "1003", "name": "泪奔",
+        "description": "大哭 伤心 破防 感动哭 泪流满面 呜呜",
+        "width": 128, "height": 128, "formats": "png",
+    },
+    "尊嘟假嘟": {
+        "sticker_id": "276", "package_id": "1003", "name": "尊嘟假嘟",
+        "description": "真的假的 真假 可爱问 你骗我 是不是",
+        "width": 128, "height": 128, "formats": "png",
+    },
+    "略略略": {
+        "sticker_id": "113", "package_id": "1003", "name": "略略略",
+        "description": "调皮 吐舌 不服 略 气死你 鬼脸",
+        "width": 128, "height": 128, "formats": "png",
+    },
+    "困": {
+        "sticker_id": "180", "package_id": "1003", "name": "困",
+        "description": "想睡 倦 打哈欠 睁不开眼 好困啊 sleepy",
+        "width": 128, "height": 128, "formats": "png",
+    },
+    "折磨": {
+        "sticker_id": "181", "package_id": "1003", "name": "折磨",
+        "description": "难受 痛苦 煎熬 蚌埠住了 受不了 要命",
+        "width": 128, "height": 128, "formats": "png",
+    },
+    "抠鼻": {
+        "sticker_id": "182", "package_id": "1003", "name": "抠鼻",
+        "description": "不屑 无聊 淡定 无所谓 鄙视 挖鼻",
+        "width": 128, "height": 128, "formats": "png",
+    },
+    "鼓掌": {
+        "sticker_id": "183", "package_id": "1003", "name": "鼓掌",
+        "description": "拍手 叫好 赞同 666 喝彩 掌声",
+        "width": 128, "height": 128, "formats": "png",
+    },
+    "斜眼笑": {
+        "sticker_id": "204", "package_id": "1003", "name": "斜眼笑",
+        "description": "滑稽 坏笑 doge 意味深长 阴阳怪气 嘿嘿嘿",
+        "width": 128, "height": 128, "formats": "png",
+    },
+    "辣眼睛": {
+        "sticker_id": "216", "package_id": "1003", "name": "辣眼睛",
+        "description": "看不下去 cringe 毁三观 太丑了 瞎了",
+        "width": 128, "height": 128, "formats": "png",
+    },
+    "哦哟": {
+        "sticker_id": "217", "package_id": "1003", "name": "哦哟",
+        "description": "惊讶 起哄 哇哦 有戏 不简单 哟",
+        "width": 128, "height": 128, "formats": "png",
+    },
+    "吃瓜": {
+        "sticker_id": "222", "package_id": "1003", "name": "吃瓜",
+        "description": "围观 看戏 八卦 路人 看热闹 板凳",
+        "width": 128, "height": 128, "formats": "png",
+    },
+    "狗头": {
+        "sticker_id": "225", "package_id": "1003", "name": "狗头",
+        "description": "doge 保命 开玩笑 滑稽 反讽 懂的都懂",
+        "width": 128, "height": 128, "formats": "png",
+    },
+    "敬礼": {
+        "sticker_id": "227", "package_id": "1003", "name": "敬礼",
+        "description": "salute 尊重 收到 遵命 致敬 报告",
+        "width": 128, "height": 128, "formats": "png",
+    },
+    "哦": {
+        "sticker_id": "231", "package_id": "1003", "name": "哦",
+        "description": "知道了 明白 敷衍 嗯 这样啊 收到",
+        "width": 128, "height": 128, "formats": "png",
+    },
+    "拿到红包": {
+        "sticker_id": "236", "package_id": "1003", "name": "拿到红包",
+        "description": "红包 谢谢老板 发财 开心 抢到了 欧气",
+        "width": 128, "height": 128, "formats": "png",
+    },
+    "牛吖": {
+        "sticker_id": "239", "package_id": "1003", "name": "牛吖",
+        "description": "牛 厉害 强 666 佩服 大佬",
+        "width": 128, "height": 128, "formats": "png",
+    },
+    "贴贴": {
+        "sticker_id": "272", "package_id": "1003", "name": "贴贴",
+        "description": "抱抱 亲昵 蹭蹭 亲密 靠靠 撒娇贴",
+        "width": 128, "height": 128, "formats": "png",
+    },
+    "爱心": {
+        "sticker_id": "138", "package_id": "1003", "name": "爱心",
+        "description": "心 love 喜欢你 红心 示爱 么么哒",
+        "width": 128, "height": 128, "formats": "png",
+    },
+    "晚安": {
+        "sticker_id": "170", "package_id": "1003", "name": "晚安",
+        "description": "好梦 睡了 night 早点休息 安啦 moon",
+        "width": 128, "height": 128, "formats": "png",
+    },
+    "太阳": {
+        "sticker_id": "176", "package_id": "1003", "name": "太阳",
+        "description": "晴天 早上好 阳光 morning 好天气 日",
+        "width": 128, "height": 128, "formats": "png",
+    },
+    "柠檬": {
+        "sticker_id": "266", "package_id": "1003", "name": "柠檬",
+        "description": "酸 嫉妒 柠檬精 羡慕 我酸 恰柠檬",
+        "width": 128, "height": 128, "formats": "png",
+    },
+    "大冤种": {
+        "sticker_id": "267", "package_id": "1003", "name": "大冤种",
+        "description": "倒霉 吃亏 自嘲 好心没好报 背锅 工具人",
+        "width": 128, "height": 128, "formats": "png",
+    },
+    "吐了": {
+        "sticker_id": "132", "package_id": "1003", "name": "吐了",
+        "description": "恶心 yue 受不了 嫌弃 想吐 生理不适",
+        "width": 128, "height": 128, "formats": "png",
+    },
+    "怒": {
+        "sticker_id": "134", "package_id": "1003", "name": "怒",
+        "description": "生气 愤怒 火大 暴躁 气炸 怼",
+        "width": 128, "height": 128, "formats": "png",
+    },
+    "玫瑰": {
+        "sticker_id": "165", "package_id": "1003", "name": "玫瑰",
+        "description": "花 示爱 表白 浪漫 送你花 情人节",
+        "width": 128, "height": 128, "formats": "png",
+    },
+    "凋谢": {
+        "sticker_id": "119", "package_id": "1003", "name": "凋谢",
+        "description": "花谢 失恋 难过 枯萎 心碎 凉了",
+        "width": 128, "height": 128, "formats": "png",
+    },
+    "点赞": {
+        "sticker_id": "159", "package_id": "1003", "name": "点赞",
+        "description": "赞 认同 好棒 good like 大拇指 顶",
+        "width": 128, "height": 128, "formats": "png",
+    },
+    "握手": {
+        "sticker_id": "164", "package_id": "1003", "name": "握手",
+        "description": "合作 你好 商务 hello deal 成交 友好",
+        "width": 128, "height": 128, "formats": "png",
+    },
+    "抱拳": {
+        "sticker_id": "163", "package_id": "1003", "name": "抱拳",
+        "description": "谢谢 失敬 江湖 承让 拜托 有礼",
+        "width": 128, "height": 128, "formats": "png",
+    },
+    "ok": {
+        "sticker_id": "169", "package_id": "1003", "name": "ok",
+        "description": "好的 收到 没问题 okay 行 可以 懂了",
+        "width": 128, "height": 128, "formats": "png",
+    },
+    "拳头": {
+        "sticker_id": "174", "package_id": "1003", "name": "拳头",
+        "description": "加油 干 冲 fight 力量 击拳 硬气",
+        "width": 128, "height": 128, "formats": "png",
+    },
+    "鞭炮": {
+        "sticker_id": "191", "package_id": "1003", "name": "鞭炮",
+        "description": "过年 喜庆 爆竹 春节 噼里啪啦 红",
+        "width": 128, "height": 128, "formats": "png",
+    },
+    "烟花": {
+        "sticker_id": "258", "package_id": "1003", "name": "烟花",
+        "description": "庆典 漂亮 新年 嘭 绽放 节日快乐",
+        "width": 128, "height": 128, "formats": "png",
+    },
+}
+
+
+def get_sticker_by_name(name: str) -> Optional[dict]:
+    """
+    按名称查找贴纸，支持模糊匹配。
+
+    匹配优先级：
+      1. 完全相等（name）
+      2. name 包含查询词（前缀/子串）
+      3. description 包含查询词（同义词搜索）
+      4. 通用模糊评分（与 sticker-search 同算法），命中即返回得分最高的一条
+
+    返回 sticker dict，找不到返回 None。
+    """
+    if not name:
+        return None
+
+    query = name.strip()
+
+    if query in STICKER_MAP:
+        return STICKER_MAP[query]
+
+    for key, sticker in STICKER_MAP.items():
+        if query in key or key in query:
+            return sticker
+
+    for sticker in STICKER_MAP.values():
+        desc = sticker.get("description", "")
+        if query in desc:
+            return sticker
+
+    matches = search_stickers(query, limit=1)
+    return matches[0] if matches else None
+
+
+def get_random_sticker(category: str = None) -> dict:
+    """
+    随机返回一个贴纸。
+
+    若指定 category，则在 description 中含有该关键词的贴纸里随机选取；
+    category 为 None 时从全表随机。
+    """
+    if category:
+        candidates = [
+            s for s in STICKER_MAP.values()
+            if category in s.get("description", "") or category in s.get("name", "")
+        ]
+        if candidates:
+            return random.choice(candidates)
+    return random.choice(list(STICKER_MAP.values()))
+
+
+def get_sticker_by_id(sticker_id: str) -> Optional[dict]:
+    """按 sticker_id 精确查找贴纸。"""
+    if not sticker_id:
+        return None
+    sid = str(sticker_id).strip()
+    for sticker in STICKER_MAP.values():
+        if sticker.get("sticker_id") == sid:
+            return sticker
+    return None
+
+
+# ---------------------------------------------------------------------------
+# 模糊搜索（对齐 chatbot-web yuanbao-openclaw-plugin/sticker-cache.ts.searchStickers）
+# ---------------------------------------------------------------------------
+
+_PUNCT_RE = re.compile(r"[\s\u3000\-_·.,，。!！?？\"“”'‘’、/\\]+")
+
+
+def _normalize_text(raw: str) -> str:
+    return unicodedata.normalize("NFKC", str(raw or "")).strip().lower()
+
+
+def _compact_text(raw: str) -> str:
+    return _PUNCT_RE.sub("", _normalize_text(raw))
+
+
+def _multiset_char_hit_ratio(needle: str, haystack: str) -> float:
+    if not needle:
+        return 0.0
+    bag: dict[str, int] = {}
+    for ch in haystack:
+        bag[ch] = bag.get(ch, 0) + 1
+    hits = 0
+    for ch in needle:
+        n = bag.get(ch, 0)
+        if n > 0:
+            hits += 1
+            bag[ch] = n - 1
+    return hits / len(needle)
+
+
+def _bigram_jaccard(a: str, b: str) -> float:
+    if len(a) < 2 or len(b) < 2:
+        return 0.0
+    A = {a[i:i + 2] for i in range(len(a) - 1)}
+    B = {b[i:i + 2] for i in range(len(b) - 1)}
+    inter = len(A & B)
+    union = len(A) + len(B) - inter
+    return inter / union if union else 0.0
+
+
+def _longest_subsequence_ratio(needle: str, haystack: str) -> float:
+    if not needle:
+        return 0.0
+    j = 0
+    for ch in haystack:
+        if j >= len(needle):
+            break
+        if ch == needle[j]:
+            j += 1
+    return j / len(needle)
+
+
+def _score_field(haystack: str, query: str) -> float:
+    hay = _normalize_text(haystack)
+    q = _normalize_text(query)
+    if not hay or not q:
+        return 0.0
+    hay_c = _compact_text(haystack)
+    q_c = _compact_text(query)
+    best = 0.0
+    if hay == q:
+        best = max(best, 100.0)
+    if q in hay:
+        best = max(best, 92 + min(6, len(q)))
+    if len(q) >= 2 and hay.startswith(q):
+        best = max(best, 88.0)
+    if q_c and q_c in hay_c:
+        best = max(best, 86.0)
+    best = max(best, _multiset_char_hit_ratio(q_c, hay_c) * 62)
+    best = max(best, _bigram_jaccard(q_c, hay_c) * 58)
+    best = max(best, _longest_subsequence_ratio(q_c, hay_c) * 52)
+    if len(q) == 1 and q in hay:
+        best = max(best, 68.0)
+    return best
+
+
+def search_stickers(query: str, limit: int = 10) -> list[dict]:
+    """
+    在内置贴纸表中按模糊匹配排序返回前 N 条结果。
+
+    评分综合 name/description 字段的子串、字符多重集覆盖、bigram Jaccard、子序列比例。
+    name 权重略高于 description（×0.88）。空 query 时按字典顺序返回前 N 条。
+    """
+    safe_limit = max(1, min(500, int(limit) if limit else 10))
+    if not query or not _normalize_text(query):
+        return list(STICKER_MAP.values())[:safe_limit]
+
+    scored: list[tuple[float, dict]] = []
+    for sticker in STICKER_MAP.values():
+        name_s = _score_field(sticker.get("name", ""), query)
+        desc_s = _score_field(sticker.get("description", ""), query) * 0.88
+        sid = str(sticker.get("sticker_id", "")).strip()
+        q_norm = _normalize_text(query)
+        id_s = 0.0
+        if sid and q_norm:
+            sid_norm = _normalize_text(sid)
+            if sid_norm == q_norm:
+                id_s = 100.0
+            elif q_norm in sid_norm:
+                id_s = 84.0
+        scored.append((max(name_s, desc_s, id_s), sticker))
+
+    scored.sort(key=lambda x: x[0], reverse=True)
+    top = scored[0][0] if scored else 0
+    if top <= 0:
+        return [s for _, s in scored[:safe_limit]]
+
+    if top >= 22:
+        floor = 18.0
+    elif top >= 12:
+        floor = max(10.0, top * 0.5)
+    else:
+        floor = max(6.0, top * 0.35)
+
+    filtered = [pair for pair in scored if pair[0] >= floor]
+    out = filtered if filtered else scored
+    return [s for _, s in out[:safe_limit]]
+
+
+def build_face_msg_body(
+    face_index: int,
+    face_type: int = 1,
+    data: Optional[str] = None,
+) -> list:
+    """
+    构造 TIMFaceElem 消息体。
+
+    Yuanbao 约定：
+      - index 固定传 0（服务端通过 data 字段识别具体表情）
+      - data 为 JSON 字符串，包含 sticker_id / package_id 等字段
+
+    Args:
+        face_index: 保留字段，暂时不影响 wire format（Yuanbao 固定 index=0）。
+                    当 face_index > 0 时视为旧版 QQ 表情 ID，直接放入 index。
+        face_type:  保留字段（兼容旧接口，当前未使用）。
+        data:       已序列化的 JSON 字符串；为 None 时仅传 index。
+
+    Returns:
+        符合 Yuanbao TIM 协议的 msg_body list，如::
+
+            [{"msg_type": "TIMFaceElem", "msg_content": {"index": 0, "data": "..."}}]
+    """
+    msg_content: dict = {"index": face_index}
+    if data is not None:
+        msg_content["data"] = data
+    return [{"msg_type": "TIMFaceElem", "msg_content": msg_content}]
+
+
+def build_sticker_msg_body(sticker: dict) -> list:
+    """
+    从 STICKER_MAP 中的 sticker dict 直接构造 TIMFaceElem 消息体。
+
+    这是 send_sticker() 的内部辅助，确保 data 字段与原始 JS 插件一致。
+    """
+    data_payload = json.dumps(
+        {
+            "sticker_id": sticker["sticker_id"],
+            "package_id": sticker["package_id"],
+            "width": sticker.get("width", 128),
+            "height": sticker.get("height", 128),
+            "formats": sticker.get("formats", "png"),
+            "name": sticker["name"],
+        },
+        ensure_ascii=False,
+        separators=(",", ":"),
+    )
+    return build_face_msg_body(face_index=0, data=data_payload)

gateway/run.py

@@ -591,20 +591,20 @@ def _parse_session_key(session_key: str) -> "dict | None":
 
 
 def _format_gateway_process_notification(evt: dict) -> "str | None":
-    """Format a watch pattern event from completion_queue into a [SYSTEM:] message."""
+    """Format a watch pattern event from completion_queue into a [IMPORTANT:] message."""
     evt_type = evt.get("type", "completion")
     _sid = evt.get("session_id", "unknown")
     _cmd = evt.get("command", "unknown")
 
     if evt_type == "watch_disabled":
-        return f"[SYSTEM: {evt.get('message', '')}]"
+        return f"[IMPORTANT: {evt.get('message', '')}]"
 
     if evt_type == "watch_match":
         _pat = evt.get("pattern", "?")
         _out = evt.get("output", "")
         _sup = evt.get("suppressed", 0)
         text = (
-            f"[SYSTEM: Background process {_sid} matched "
+            f"[IMPORTANT: Background process {_sid} matched "
             f"watch pattern \"{_pat}\".\n"
             f"Command: {_cmd}\n"
             f"Matched output:\n{_out}"
@@ -682,6 +682,16 @@ class GatewayRunner:
         self._running_agents: Dict[str, Any] = {}
         self._running_agents_ts: Dict[str, float] = {}  # start timestamp per session
         self._pending_messages: Dict[str, str] = {}  # Queued messages during interrupt
+        # Overflow buffer for explicit /queue commands.  The adapter-level
+        # _pending_messages dict is a single slot per session (designed for
+        # "next-turn" follow-ups where repeated sends collapse into one
+        # event).  /queue has different semantics: each invocation must
+        # produce its own full agent turn, in FIFO order, with no merging.
+        # When the slot is occupied, additional /queue items land here and
+        # are promoted one-at-a-time after each run's drain.  Cleared on
+        # /new and /reset.  /model and other mid-session operations
+        # preserve the queue.
+        self._queued_events: Dict[str, List[MessageEvent]] = {}
         self._busy_ack_ts: Dict[str, float] = {}  # last busy-ack timestamp per session (debounce)
         self._session_run_generation: Dict[str, int] = {}
 
@@ -753,10 +763,27 @@ class GatewayRunner:
                         retention_days=int(_sess_cfg.get("retention_days", 90)),
                         min_interval_hours=int(_sess_cfg.get("min_interval_hours", 24)),
                         vacuum=bool(_sess_cfg.get("vacuum_after_prune", True)),
+                        sessions_dir=self.config.sessions_dir,
                     )
             except Exception as exc:
                 logger.debug("state.db auto-maintenance skipped: %s", exc)
 
+        # Opportunistic shadow-repo cleanup — deletes orphan/stale
+        # checkpoint repos under ~/.hermes/checkpoints/.  Opt-in via
+        # checkpoints.auto_prune, idempotent via .last_prune marker.
+        try:
+            from hermes_cli.config import load_config as _load_full_config
+            _ckpt_cfg = (_load_full_config().get("checkpoints") or {})
+            if _ckpt_cfg.get("auto_prune", False):
+                from tools.checkpoint_manager import maybe_auto_prune_checkpoints
+                maybe_auto_prune_checkpoints(
+                    retention_days=int(_ckpt_cfg.get("retention_days", 7)),
+                    min_interval_hours=int(_ckpt_cfg.get("min_interval_hours", 24)),
+                    delete_orphans=bool(_ckpt_cfg.get("delete_orphans", True)),
+                )
+        except Exception as exc:
+            logger.debug("checkpoint auto-maintenance skipped: %s", exc)
+
         # DM pairing store for code-based user authorization
         from gateway.pairing import PairingStore
         self.pairing_store = PairingStore()
@@ -1202,7 +1229,80 @@ class GatewayRunner:
         return "restarting" if self._restart_requested else "shutting down"
 
     def _queue_during_drain_enabled(self) -> bool:
-        return self._restart_requested and self._busy_input_mode == "queue"
+        # Both "queue" and "steer" modes imply the user doesn't want messages
+        # to be lost during restart — queue them for the newly-spawned gateway
+        # process to pick up.  "interrupt" mode drops them (current behaviour).
+        return self._restart_requested and self._busy_input_mode in ("queue", "steer")
+
+    # -------- /queue FIFO helpers --------------------------------------
+    # /queue must produce one full agent turn per invocation, in FIFO
+    # order, with no merging.  The adapter's _pending_messages dict is a
+    # single "next-up" slot (shared with photo-burst follow-ups), so we
+    # use it for the head of the queue and an overflow list for the
+    # tail.  Enqueue puts new items in the slot when free, otherwise in
+    # the overflow.  Promotion (called after each run's drain) moves the
+    # next overflow item into the slot so the following recursion picks
+    # it up.  Clearing happens on /new and /reset via
+    # _handle_reset_command.
+
+    def _enqueue_fifo(self, session_key: str, queued_event: "MessageEvent", adapter: Any) -> None:
+        """Append a /queue event to the FIFO chain for a session."""
+        if adapter is None:
+            return
+        pending_slot = getattr(adapter, "_pending_messages", None)
+        if pending_slot is None:
+            return
+        queued_events = getattr(self, "_queued_events", None)
+        if queued_events is None:
+            queued_events = {}
+            self._queued_events = queued_events
+        if session_key in pending_slot:
+            queued_events.setdefault(session_key, []).append(queued_event)
+        else:
+            pending_slot[session_key] = queued_event
+
+    def _promote_queued_event(
+        self,
+        session_key: str,
+        adapter: Any,
+        pending_event: Optional["MessageEvent"],
+    ) -> Optional["MessageEvent"]:
+        """Promote the next overflow item after the slot was drained.
+
+        Called at the drain site after _dequeue_pending_event consumed
+        (or failed to consume) the slot.  If there's an overflow item:
+          - When pending_event is None (slot was empty), return the
+            overflow head as the new pending_event.
+          - When pending_event already exists (slot was populated by an
+            interrupt follow-up or similar), stage the overflow head in
+            the slot so the NEXT recursion picks it up.
+        Returns the (possibly updated) pending_event for drain to use.
+        """
+        queued_events = getattr(self, "_queued_events", None)
+        if not queued_events:
+            return pending_event
+        overflow = queued_events.get(session_key)
+        if not overflow:
+            return pending_event
+        next_queued = overflow.pop(0)
+        if not overflow:
+            queued_events.pop(session_key, None)
+        if pending_event is None:
+            return next_queued
+        if adapter is not None and hasattr(adapter, "_pending_messages"):
+            adapter._pending_messages[session_key] = next_queued
+        else:
+            # No adapter — push back so we don't silently drop the item.
+            queued_events.setdefault(session_key, []).insert(0, next_queued)
+        return pending_event
+
+    def _queue_depth(self, session_key: str, *, adapter: Any = None) -> int:
+        """Total pending /queue items for a session — slot + overflow."""
+        queued_events = getattr(self, "_queued_events", None) or {}
+        depth = len(queued_events.get(session_key, []))
+        if adapter is not None and session_key in getattr(adapter, "_pending_messages", {}):
+            depth += 1
+        return depth
 
     def _update_runtime_status(self, gateway_state: Optional[str] = None, exit_reason: Optional[str] = None) -> None:
         try:
@@ -1433,7 +1533,11 @@ class GatewayRunner:
                     mode = str(cfg.get("display", {}).get("busy_input_mode", "") or "").strip().lower()
             except Exception:
                 pass
-        return "queue" if mode == "queue" else "interrupt"
+        if mode == "queue":
+            return "queue"
+        if mode == "steer":
+            return "steer"
+        return "interrupt"
 
     @staticmethod
     def _load_restart_drain_timeout() -> float:
@@ -1571,18 +1675,46 @@ class GatewayRunner:
         if not adapter:
             return False  # let default path handle it
 
+        running_agent = self._running_agents.get(session_key)
+
+        # Steer mode: inject mid-run via running_agent.steer() instead of
+        # queueing + interrupting.  If the agent isn't running yet
+        # (sentinel) or lacks steer(), or the payload is empty, fall back
+        # to queue semantics so nothing is lost.
+        effective_mode = self._busy_input_mode
+        steered = False
+        if effective_mode == "steer":
+            steer_text = (event.text or "").strip()
+            can_steer = (
+                steer_text
+                and running_agent is not None
+                and running_agent is not _AGENT_PENDING_SENTINEL
+                and hasattr(running_agent, "steer")
+            )
+            if can_steer:
+                try:
+                    steered = bool(running_agent.steer(steer_text))
+                except Exception as exc:
+                    logger.warning("Gateway steer failed for session %s: %s", session_key, exc)
+                    steered = False
+            if not steered:
+                # Fall back to queue (merge into pending messages, no interrupt)
+                effective_mode = "queue"
+
         # Store the message so it's processed as the next turn after the
-        # current run finishes (or is interrupted).
-        from gateway.platforms.base import merge_pending_message_event
-        merge_pending_message_event(adapter._pending_messages, session_key, event)
+        # current run finishes (or is interrupted).  Skip this for a
+        # successful steer — the text already landed inside the run and
+        # must NOT also be replayed as a next-turn user message.
+        if not steered:
+            merge_pending_message_event(adapter._pending_messages, session_key, event)
 
-        is_queue_mode = self._busy_input_mode == "queue"
+        is_queue_mode = effective_mode == "queue"
+        is_steer_mode = effective_mode == "steer"
 
-        # If not in queue mode, interrupt the running agent immediately.
+        # If not in queue/steer mode, interrupt the running agent immediately.
         # This aborts in-flight tool calls and causes the agent loop to exit
         # at the next check point.
-        running_agent = self._running_agents.get(session_key)
-        if not is_queue_mode and running_agent and running_agent is not _AGENT_PENDING_SENTINEL:
+        if effective_mode == "interrupt" and running_agent and running_agent is not _AGENT_PENDING_SENTINEL:
             try:
                 running_agent.interrupt(event.text)
             except Exception:
@@ -1619,7 +1751,12 @@ class GatewayRunner:
                 pass
 
         status_detail = f" ({', '.join(status_parts)})" if status_parts else ""
-        if is_queue_mode:
+        if is_steer_mode:
+            message = (
+                f"⏩ Steered into current run{status_detail}. "
+                f"Your message arrives after the next tool call."
+            )
+        elif is_queue_mode:
             message = (
                 f"⏳ Queued for the next turn{status_detail}. "
                 f"I'll respond once the current task finishes."
@@ -1643,9 +1780,15 @@ class GatewayRunner:
             )
             _user_cfg = _load_gateway_config()
             if not is_seen(_user_cfg, BUSY_INPUT_FLAG):
+                if is_steer_mode:
+                    _hint_mode = "steer"
+                elif is_queue_mode:
+                    _hint_mode = "queue"
+                else:
+                    _hint_mode = "interrupt"
                 message = (
                     f"{message}\n\n"
-                    f"{busy_input_hint_gateway('queue' if is_queue_mode else 'interrupt')}"
+                    f"{busy_input_hint_gateway(_hint_mode)}"
                 )
                 mark_seen(_hermes_home / "config.yaml", BUSY_INPUT_FLAG)
         except Exception as _onb_err:
@@ -1996,6 +2139,7 @@ class GatewayRunner:
                        "WEIXIN_ALLOWED_USERS",
                        "BLUEBUBBLES_ALLOWED_USERS",
                        "QQ_ALLOWED_USERS",
+                       "YUANBAO_ALLOWED_USERS",
                        "GATEWAY_ALLOWED_USERS")
         )
         _allow_all = os.getenv("GATEWAY_ALLOW_ALL_USERS", "").lower() in ("true", "1", "yes") or any(
@@ -2010,7 +2154,8 @@ class GatewayRunner:
                        "WECOM_CALLBACK_ALLOW_ALL_USERS",
                        "WEIXIN_ALLOW_ALL_USERS",
                        "BLUEBUBBLES_ALLOW_ALL_USERS",
-                       "QQ_ALLOW_ALL_USERS")
+                       "QQ_ALLOW_ALL_USERS",
+                       "YUANBAO_ALLOW_ALL_USERS")
         )
         if not _any_allowlist and not _allow_all:
             logger.warning(
@@ -2254,7 +2399,7 @@ class GatewayRunner:
         # Build initial channel directory for send_message name resolution
         try:
             from gateway.channel_directory import build_channel_directory
-            directory = build_channel_directory(self.adapters)
+            directory = await build_channel_directory(self.adapters)
             ch_count = sum(len(chs) for chs in directory.get("platforms", {}).values())
             logger.info("Channel directory built: %d target(s)", ch_count)
         except Exception as e:
@@ -2538,7 +2683,7 @@ class GatewayRunner:
                         # Rebuild channel directory with the new adapter
                         try:
                             from gateway.channel_directory import build_channel_directory
-                            build_channel_directory(self.adapters)
+                            await build_channel_directory(self.adapters)
                         except Exception:
                             pass
                     else:
@@ -2720,6 +2865,23 @@ class GatewayRunner:
 
             self._finalize_shutdown_agents(active_agents)
 
+            # Also shut down memory providers on idle cached agents.
+            # _finalize_shutdown_agents only handles agents that were
+            # mid-turn at drain time; the _agent_cache may still hold
+            # idle agents whose MemoryProviders never received
+            # on_session_end().
+            _cache_lock = getattr(self, "_agent_cache_lock", None)
+            _cache = getattr(self, "_agent_cache", None)
+            if _cache_lock is not None and _cache is not None:
+                with _cache_lock:
+                    _idle_agents = list(_cache.values())
+                    _cache.clear()
+                for _entry in _idle_agents:
+                    _agent = (
+                        _entry[0] if isinstance(_entry, tuple) else _entry
+                    )
+                    self._cleanup_agent_resources(_agent)
+
             for platform, adapter in list(self.adapters.items()):
                 try:
                     await adapter.cancel_background_tasks()
@@ -2970,8 +3132,14 @@ class GatewayRunner:
                 return None
             return QQAdapter(config)
 
-        return None
+        elif platform == Platform.YUANBAO:
+            from gateway.platforms.yuanbao import YuanbaoAdapter, WEBSOCKETS_AVAILABLE
+            if not WEBSOCKETS_AVAILABLE:
+                logger.warning("Yuanbao: websockets not installed. Run: pip install websockets")
+                return None
+            return YuanbaoAdapter(config)
 
+        return None
     def _is_user_authorized(self, source: SessionSource) -> bool:
         """
         Check if a user is authorized to use the bot.
@@ -3012,6 +3180,7 @@ class GatewayRunner:
             Platform.WEIXIN: "WEIXIN_ALLOWED_USERS",
             Platform.BLUEBUBBLES: "BLUEBUBBLES_ALLOWED_USERS",
             Platform.QQBOT: "QQ_ALLOWED_USERS",
+            Platform.YUANBAO: "YUANBAO_ALLOWED_USERS",
         }
         platform_group_env_map = {
             Platform.TELEGRAM: "TELEGRAM_GROUP_ALLOWED_USERS",
@@ -3034,6 +3203,7 @@ class GatewayRunner:
             Platform.WEIXIN: "WEIXIN_ALLOW_ALL_USERS",
             Platform.BLUEBUBBLES: "BLUEBUBBLES_ALLOW_ALL_USERS",
             Platform.QQBOT: "QQ_ALLOW_ALL_USERS",
+            Platform.YUANBAO: "YUANBAO_ALLOW_ALL_USERS",
         }
 
         # Per-platform allow-all flag (e.g., DISCORD_ALLOW_ALL_USERS=true)
@@ -3282,6 +3452,10 @@ class GatewayRunner:
         # The update process (detached) wrote .update_prompt.json; the watcher
         # forwarded it to the user; now the user's reply goes back via
         # .update_response so the update process can continue.
+        #
+        # IMPORTANT: recognized slash commands must bypass this interception.
+        # Otherwise control/session commands like /new or /help get silently
+        # consumed as update answers instead of being dispatched normally.
         _quick_key = self._session_key_for_source(source)
         _update_prompts = getattr(self, "_update_prompt_pending", {})
         if _update_prompts.get(_quick_key):
@@ -3293,7 +3467,22 @@ class GatewayRunner:
             elif cmd in ("deny", "no"):
                 response_text = "n"
             else:
-                response_text = raw
+                _recognized_cmd = None
+                if cmd:
+                    try:
+                        from hermes_cli.commands import resolve_command as _resolve_update_cmd
+                    except Exception:
+                        _resolve_update_cmd = None
+                    if _resolve_update_cmd is not None:
+                        try:
+                            _cmd_def = _resolve_update_cmd(cmd)
+                            _recognized_cmd = _cmd_def.name if _cmd_def else None
+                        except Exception:
+                            _recognized_cmd = None
+                if _recognized_cmd:
+                    response_text = ""
+                else:
+                    response_text = raw
             if response_text:
                 response_path = _hermes_home / ".update_response"
                 try:
@@ -3306,6 +3495,30 @@ class GatewayRunner:
                 _update_prompts.pop(_quick_key, None)
                 label = response_text if len(response_text) <= 20 else response_text[:20] + "…"
                 return f"✓ Sent `{label}` to the update process."
+            # Recognized slash command during a pending update prompt:
+            # unblock the detached update subprocess by writing a blank
+            # response so ``_gateway_prompt`` returns the prompt's default
+            # (typically a safe "n" / skip) and exits cleanly instead of
+            # blocking on stdin until the 30-minute watcher timeout.
+            # The slash command then falls through to normal dispatch.
+            if _recognized_cmd:
+                response_path = _hermes_home / ".update_response"
+                try:
+                    tmp = response_path.with_suffix(".tmp")
+                    tmp.write_text("")
+                    tmp.replace(response_path)
+                    logger.info(
+                        "Recognized /%s during pending update prompt for %s; "
+                        "cancelled prompt with default and dispatching command",
+                        _recognized_cmd,
+                        _quick_key,
+                    )
+                except OSError as e:
+                    logger.warning(
+                        "Failed to write cancel response for pending update prompt: %s",
+                        e,
+                    )
+                _update_prompts.pop(_quick_key, None)
 
         # PRIORITY handling when an agent is already running for this session.
         # Default behavior is to interrupt immediately so user text/stop messages
@@ -3416,7 +3629,10 @@ class GatewayRunner:
                 # doesn't think an agent is still active.
                 return await self._handle_reset_command(event)
 
-            # /queue <prompt> — queue without interrupting
+            # /queue <prompt> — queue without interrupting.
+            # Semantics: each /queue invocation produces its own full agent
+            # turn, processed in FIFO order after the current run (and any
+            # earlier /queue items) finishes.  Messages are NOT merged.
             if event.get_command() in ("queue", "q"):
                 queued_text = event.get_command_args().strip()
                 if not queued_text:
@@ -3430,8 +3646,11 @@ class GatewayRunner:
                         message_id=event.message_id,
                         channel_prompt=event.channel_prompt,
                     )
-                    adapter._pending_messages[_quick_key] = queued_event
-                return "Queued for the next turn."
+                    self._enqueue_fifo(_quick_key, queued_event, adapter)
+                depth = self._queue_depth(_quick_key, adapter=self.adapters.get(source.platform))
+                if depth <= 1:
+                    return "Queued for the next turn."
+                return f"Queued for the next turn. ({depth} queued)"
 
             # /steer <prompt> — inject mid-run after the next tool call.
             # Unlike /queue (turn boundary), /steer lands BETWEEN tool-call
@@ -3608,6 +3827,24 @@ class GatewayRunner:
                 logger.debug("PRIORITY queue follow-up for session %s", _quick_key)
                 self._queue_or_replace_pending_event(_quick_key, event)
                 return None
+            if self._busy_input_mode == "steer":
+                # Steer mode: inject text into the running agent mid-run via
+                # agent.steer().  Falls back to queue semantics if the payload
+                # is empty, the agent lacks steer(), or steer() rejects.
+                steer_text = (event.text or "").strip()
+                steered = False
+                if steer_text and hasattr(running_agent, "steer"):
+                    try:
+                        steered = bool(running_agent.steer(steer_text))
+                    except Exception as exc:
+                        logger.warning("PRIORITY steer failed for session %s: %s", _quick_key, exc)
+                        steered = False
+                if steered:
+                    logger.debug("PRIORITY steer for session %s", _quick_key)
+                    return None
+                logger.debug("PRIORITY steer-fallback-to-queue for session %s", _quick_key)
+                self._queue_or_replace_pending_event(_quick_key, event)
+                return None
             logger.debug("PRIORITY interrupt for session %s", _quick_key)
             running_agent.interrupt(event.text)
             if _quick_key in self._pending_messages:
@@ -4118,7 +4355,14 @@ class GatewayRunner:
         session_entry = self.session_store.get_or_create_session(source)
         session_key = session_entry.session_key
         if getattr(session_entry, "was_auto_reset", False):
+            # Treat auto-reset as a full conversation boundary — drop every
+            # session-scoped transient state so the fresh session does not
+            # inherit the previous conversation's model/reasoning overrides
+            # or a queued "/model switched" note.
+            self._session_model_overrides.pop(session_key, None)
             self._set_session_reasoning_override(session_key, None)
+            if hasattr(self, "_pending_model_notes"):
+                self._pending_model_notes.pop(session_key, None)
         
         # Emit session:start for new or auto-reset sessions
         _is_new_session = (
@@ -4232,7 +4476,7 @@ class GatewayRunner:
                     if _loaded:
                         _loaded_skill, _skill_dir, _display_name = _loaded
                         _note = (
-                            f'[SYSTEM: The "{_display_name}" skill is auto-loaded. '
+                            f'[IMPORTANT: The "{_display_name}" skill is auto-loaded. '
                             f"Follow its instructions for this session.]"
                         )
                         _part = _build_skill_message(_loaded_skill, _skill_dir, _note)
@@ -4520,12 +4764,20 @@ class GatewayRunner:
             if not os.getenv(env_key):
                 adapter = self.adapters.get(source.platform)
                 if adapter:
+                    # Slack dispatches all Hermes commands through a single
+                    # parent slash command `/hermes`; bare `/sethome` is not
+                    # registered and would fail with "app did not respond".
+                    sethome_cmd = (
+                        "/hermes sethome"
+                        if source.platform == Platform.SLACK
+                        else "/sethome"
+                    )
                     await adapter.send(
                         source.chat_id,
                         f"📬 No home channel is set for {platform_name.title()}. "
                         f"A home channel is where Hermes delivers cron job results "
                         f"and cross-platform messages.\n\n"
-                        f"Type /sethome to make this chat your home channel, "
+                        f"Type {sethome_cmd} to make this chat your home channel, "
                         f"or ignore to skip."
                     )
         
@@ -4790,6 +5042,8 @@ class GatewayRunner:
                 self._evict_cached_agent(session_key)
                 self._session_model_overrides.pop(session_key, None)
                 self._set_session_reasoning_override(session_key, None)
+                if hasattr(self, "_pending_model_notes"):
+                    self._pending_model_notes.pop(session_key, None)
                 response = (response or "") + (
                     "\n\n🔄 Session auto-reset — the conversation exceeded the "
                     "maximum context size and could not be compressed further. "
@@ -5058,6 +5312,13 @@ class GatewayRunner:
                 self._cleanup_agent_resources(_old_agent)
         self._evict_cached_agent(session_key)
 
+        # Discard any /queue overflow for this session — /new is a
+        # conversation-boundary operation, queued follow-ups from the
+        # previous conversation must not bleed into the new one.
+        _qe = getattr(self, "_queued_events", None)
+        if _qe is not None:
+            _qe.pop(session_key, None)
+
         try:
             from tools.env_passthrough import clear_env_passthrough
             clear_env_passthrough()
@@ -5077,6 +5338,8 @@ class GatewayRunner:
         # picks up configured defaults instead of previous session switches.
         self._session_model_overrides.pop(session_key, None)
         self._set_session_reasoning_override(session_key, None)
+        if hasattr(self, "_pending_model_notes"):
+            self._pending_model_notes.pop(session_key, None)
 
         # Clear session-scoped dangerous-command approvals and /yolo state.
         # /new is a conversation-boundary operation — approval state from the
@@ -5165,6 +5428,10 @@ class GatewayRunner:
         session_key = session_entry.session_key
         is_running = session_key in self._running_agents
 
+        # Count pending /queue follow-ups (slot + overflow).
+        adapter = self.adapters.get(source.platform) if source else None
+        queue_depth = self._queue_depth(session_key, adapter=adapter)
+
         title = None
         if self._session_db:
             try:
@@ -5184,6 +5451,10 @@ class GatewayRunner:
             f"**Last Activity:** {session_entry.updated_at.strftime('%Y-%m-%d %H:%M')}",
             f"**Tokens:** {session_entry.total_tokens:,}",
             f"**Agent Running:** {'Yes ⚡' if is_running else 'No'}",
+        ])
+        if queue_depth:
+            lines.append(f"**Queued follow-ups:** {queue_depth}")
+        lines.extend([
             "",
             f"**Connected Platforms:** {', '.join(connected_platforms)}",
         ])
@@ -6640,6 +6911,7 @@ class GatewayRunner:
                             chat_id=source.chat_id,
                             image_url=image_url,
                             caption=alt_text,
+                            metadata=_thread_metadata,
                         )
                     except Exception:
                         pass
@@ -6650,6 +6922,7 @@ class GatewayRunner:
                         await adapter.send_document(
                             chat_id=source.chat_id,
                             file_path=media_path,
+                            metadata=_thread_metadata,
                         )
                     except Exception:
                         pass
@@ -7473,7 +7746,7 @@ class GatewayRunner:
             change_detail = ". ".join(change_parts) + ". " if change_parts else ""
             reload_msg = {
                 "role": "user",
-                "content": f"[SYSTEM: MCP servers have been reloaded. {change_detail}{tool_summary}. The tool list for this conversation has been updated accordingly.]",
+                "content": f"[IMPORTANT: MCP servers have been reloaded. {change_detail}{tool_summary}. The tool list for this conversation has been updated accordingly.]",
             }
             try:
                 session_entry = self.session_store.get_or_create_session(event.source)
@@ -8412,7 +8685,7 @@ class GatewayRunner:
                     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"[IMPORTANT: Background process {session_id} completed "
                         f"(exit code {session.exit_code}).\n"
                         f"Command: {session.command}\n"
                         f"Output:\n{_out}]"
@@ -8615,7 +8888,7 @@ class GatewayRunner:
         return True
 
     def _clear_session_boundary_security_state(self, session_key: str) -> None:
-        """Clear approval state that must not survive a real conversation switch."""
+        """Clear per-session control state that must not survive a boundary switch."""
         if not session_key:
             return
 
@@ -8623,6 +8896,10 @@ class GatewayRunner:
         if isinstance(pending_approvals, dict):
             pending_approvals.pop(session_key, None)
 
+        update_prompt_pending = getattr(self, "_update_prompt_pending", None)
+        if isinstance(update_prompt_pending, dict):
+            update_prompt_pending.pop(session_key, None)
+
         try:
             from tools.approval import clear_session as _clear_approval_session
         except Exception:
@@ -8722,6 +8999,25 @@ class GatewayRunner:
             with _lock:
                 self._agent_cache.pop(session_key, None)
 
+    @staticmethod
+    def _init_cached_agent_for_turn(agent: Any, interrupt_depth: int) -> None:
+        """Reset per-turn state on a cached agent before a new turn starts.
+
+        Both _last_activity_ts and _last_activity_desc are only reset for
+        fresh external turns (depth 0); they are semantically paired —
+        desc describes the activity *at* ts, so updating one without the
+        other would make get_activity_summary() misleading.
+        For interrupt-recursive turns both are preserved so the inactivity
+        watchdog can accumulate stuck-turn idle time and fire the 30-min
+        timeout (#15654).  The depth-0 reset is still needed: a session
+        idle for 29 min would otherwise trip the watchdog before the new
+        turn makes its first API call (#9051).
+        """
+        if interrupt_depth == 0:
+            agent._last_activity_ts = time.time()
+            agent._last_activity_desc = "starting new turn (cached)"
+        agent._api_call_count = 0
+
     def _release_evicted_agent_soft(self, agent: Any) -> None:
         """Soft cleanup for cache-evicted agents — preserves session tool state.
 
@@ -9009,11 +9305,21 @@ class GatewayRunner:
                     if source.platform == Platform.MATRIX:
                         _effective_cursor = ""
                         _buffer_only = True
+                    # Fresh-final applies to Telegram only — other
+                    # platforms either edit in place cheaply (Discord,
+                    # Slack) or don't have the timestamp-on-edit
+                    # problem.  (Ported from openclaw/openclaw#72038.)
+                    _fresh_final_secs = (
+                        float(getattr(_scfg, "fresh_final_after_seconds", 0.0) or 0.0)
+                        if source.platform == Platform.TELEGRAM
+                        else 0.0
+                    )
                     _consumer_cfg = StreamConsumerConfig(
                         edit_interval=_scfg.edit_interval,
                         buffer_threshold=_scfg.buffer_threshold,
                         cursor=_effective_cursor,
                         buffer_only=_buffer_only,
+                        fresh_final_after_seconds=_fresh_final_secs,
                     )
                     _stream_consumer = GatewayStreamConsumer(
                         adapter=_adapter,
@@ -9697,11 +10003,21 @@ class GatewayRunner:
                         if source.platform == Platform.MATRIX:
                             _effective_cursor = ""
                             _buffer_only = True
+                        # Fresh-final applies to Telegram only — other
+                        # platforms either edit in place cheaply or don't
+                        # have the edit-timestamp-stays-stale problem.
+                        # (Ported from openclaw/openclaw#72038.)
+                        _fresh_final_secs = (
+                            float(getattr(_scfg, "fresh_final_after_seconds", 0.0) or 0.0)
+                            if source.platform == Platform.TELEGRAM
+                            else 0.0
+                        )
                         _consumer_cfg = StreamConsumerConfig(
                             edit_interval=_scfg.edit_interval,
                             buffer_threshold=_scfg.buffer_threshold,
                             cursor=_effective_cursor,
                             buffer_only=_buffer_only,
+                            fresh_final_after_seconds=_fresh_final_secs,
                         )
                         _stream_consumer = GatewayStreamConsumer(
                             adapter=_adapter,
@@ -9766,12 +10082,7 @@ class GatewayRunner:
                                 _cache.move_to_end(session_key)
                             except KeyError:
                                 pass
-                        # Reset activity timestamp so the inactivity timeout
-                        # handler doesn't see stale idle time from the previous
-                        # turn and immediately kill this agent.  (#9051)
-                        agent._last_activity_ts = time.time()
-                        agent._last_activity_desc = "starting new turn (cached)"
-                        agent._api_call_count = 0
+                        self._init_cached_agent_for_turn(agent, _interrupt_depth)
                         logger.debug("Reusing cached agent for session %s", session_key)
 
             if agent is None:
@@ -10189,12 +10500,20 @@ class GatewayRunner:
                 try:
                     from agent.title_generator import maybe_auto_title
                     all_msgs = result_holder[0].get("messages", []) if result_holder[0] else []
+                    # Route title-generation failures through the agent's
+                    # user-visible warning channel so a depleted auxiliary
+                    # provider doesn't silently leave sessions untitled
+                    # (issue #15775).
+                    _title_failure_cb = getattr(
+                        agent, "_emit_auxiliary_failure", None
+                    )
                     maybe_auto_title(
                         self._session_db,
                         effective_session_id,
                         message,
                         final_response,
                         all_msgs,
+                        failure_callback=_title_failure_cb,
                     )
                 except Exception:
                     pass
@@ -10554,6 +10873,13 @@ class GatewayRunner:
             pending = None
             if result and adapter and session_key:
                 pending_event = _dequeue_pending_event(adapter, session_key)
+                # /queue overflow: after consuming the adapter's "next-up"
+                # slot, promote the next queued event into it so the
+                # recursive run's drain will see it.  This keeps the slot
+                # occupied for the full FIFO chain, which (a) preserves
+                # order, and (b) causes any mid-chain /queue to correctly
+                # route to overflow rather than jumping the queue.
+                pending_event = self._promote_queued_event(session_key, adapter, pending_event)
                 if result.get("interrupted") and not pending_event and result.get("interrupt_message"):
                     interrupt_message = result.get("interrupt_message")
                     if _is_control_interrupt_message(interrupt_message):
@@ -10848,7 +11174,15 @@ def _start_cron_ticker(stop_event: threading.Event, adapters=None, loop=None, in
         if tick_count % CHANNEL_DIR_EVERY == 0 and adapters:
             try:
                 from gateway.channel_directory import build_channel_directory
-                build_channel_directory(adapters)
+                if loop is not None:
+                    # build_channel_directory is async (Slack web calls), and
+                    # this ticker runs in a background thread. Schedule onto
+                    # the gateway event loop and wait briefly for completion
+                    # so refresh failures are still logged via the except.
+                    fut = asyncio.run_coroutine_threadsafe(
+                        build_channel_directory(adapters), loop
+                    )
+                    fut.result(timeout=30)
             except Exception as e:
                 logger.debug("Channel directory refresh error: %s", e)
 

gateway/session.py

@@ -310,8 +310,9 @@ def build_session_context_prompt(
             "**Platform notes:** You are running inside Slack. "
             "You do NOT have access to Slack-specific APIs — you cannot search "
             "channel history, pin/unpin messages, manage channels, or list users. "
-            "Do not promise to perform these actions. If the user asks, explain "
-            "that you can only read messages sent directly to you and respond."
+            "Do not promise to perform these actions. The gateway may inline the "
+            "current message's Slack block/attachment payload when available, but "
+            "you still cannot call Slack APIs yourself."
         )
     elif context.source.platform == Platform.DISCORD:
         # Inject the Discord IDs block only when the agent actually has
@@ -353,6 +354,14 @@ def build_session_context_prompt(
             "If the user needs a detailed answer, give the short version first "
             "and offer to elaborate."
         )
+    elif context.source.platform == Platform.YUANBAO:
+        lines.append("")
+        lines.append(
+            "**Platform notes:** You are running inside Yuanbao. "
+            "You CAN send private (DM) messages via the send_message tool. "
+            "Use target='yuanbao:direct:<account_id>' for DM "
+            "and target='yuanbao:group:<group_code>' for group chat."
+        )
 
     # Connected platforms
     platforms_list = ["local (files on this machine)"]

gateway/stream_consumer.py

@@ -44,6 +44,14 @@ class StreamConsumerConfig:
     buffer_threshold: int = 40
     cursor: str = " ▉"
     buffer_only: bool = False
+    # When >0, the final edit for a streamed response is delivered as a
+    # fresh message if the original preview has been visible for at least
+    # this many seconds.  This makes the platform's visible timestamp
+    # reflect completion time instead of first-token time for long-running
+    # responses (e.g. reasoning models that stream slowly).  Ported from
+    # openclaw/openclaw#72038.  Default 0 = always edit in place (legacy
+    # behavior).  The gateway enables this selectively per-platform.
+    fresh_final_after_seconds: float = 0.0
 
 
 class GatewayStreamConsumer:
@@ -91,6 +99,12 @@ class GatewayStreamConsumer:
         self._queue: queue.Queue = queue.Queue()
         self._accumulated = ""
         self._message_id: Optional[str] = None
+        # Wall-clock timestamp (time.monotonic) when ``_message_id`` was
+        # first assigned from a successful first-send.  Used by the
+        # fresh-final logic to detect long-lived previews whose edit
+        # timestamps would be stale by completion time.  Ported from
+        # openclaw/openclaw#72038.
+        self._message_created_ts: Optional[float] = None
         self._already_sent = False
         self._edit_supported = True  # Disabled when progressive edits are no longer usable
         self._last_edit_time = 0.0
@@ -136,6 +150,7 @@ class GatewayStreamConsumer:
         if preserve_no_edit and self._message_id == "__no_edit__":
             return
         self._message_id = None
+        self._message_created_ts = None
         self._accumulated = ""
         self._last_sent_text = ""
         self._fallback_final_send = False
@@ -734,6 +749,81 @@ class GatewayStreamConsumer:
             logger.error("Commentary send error: %s", e)
             return False
 
+    def _should_send_fresh_final(self) -> bool:
+        """Return True when a long-lived preview should be replaced with a
+        fresh final message instead of an edit.
+
+        Conditions:
+        - Fresh-final is enabled (``fresh_final_after_seconds > 0``).
+        - We have a real preview message id (not the ``__no_edit__`` sentinel
+          and not ``None``).
+        - The preview has been visible for at least the configured threshold.
+
+        Ported from openclaw/openclaw#72038.
+        """
+        threshold = getattr(self.cfg, "fresh_final_after_seconds", 0.0) or 0.0
+        if threshold <= 0:
+            return False
+        if not self._message_id or self._message_id == "__no_edit__":
+            return False
+        if self._message_created_ts is None:
+            return False
+        age = time.monotonic() - self._message_created_ts
+        return age >= threshold
+
+    async def _try_fresh_final(self, text: str) -> bool:
+        """Send ``text`` as a brand-new message (best-effort delete the old
+        preview) so the platform's visible timestamp reflects completion
+        time.  Returns True on successful delivery, False on any failure so
+        the caller falls back to the normal edit path.
+
+        Ported from openclaw/openclaw#72038.
+        """
+        old_message_id = self._message_id
+        try:
+            result = await self.adapter.send(
+                chat_id=self.chat_id,
+                content=text,
+                metadata=self.metadata,
+            )
+        except Exception as e:
+            logger.debug("Fresh-final send failed, falling back to edit: %s", e)
+            return False
+        if not getattr(result, "success", False):
+            return False
+        # Successful fresh send — try to delete the stale preview so the
+        # user doesn't see the old edit-stuck message underneath.  Cleanup
+        # is best-effort; platforms that don't implement ``delete_message``
+        # just leave the preview behind (still an acceptable outcome —
+        # the visible final timestamp is the important part).
+        if old_message_id and old_message_id != "__no_edit__":
+            delete_fn = getattr(self.adapter, "delete_message", None)
+            if delete_fn is not None:
+                try:
+                    await delete_fn(self.chat_id, old_message_id)
+                except Exception as e:
+                    logger.debug(
+                        "Fresh-final preview cleanup failed (%s): %s",
+                        old_message_id, e,
+                    )
+        # Adopt the new message id as the current message so subsequent
+        # callers (e.g. overflow split loops, finalize retries) see a
+        # consistent state.
+        new_message_id = getattr(result, "message_id", None)
+        if new_message_id:
+            self._message_id = new_message_id
+            self._message_created_ts = time.monotonic()
+        else:
+            # Send succeeded but platform didn't return an id — treat the
+            # delivery as final-only and fall back to "__no_edit__" so we
+            # don't try to edit something we can't address.
+            self._message_id = "__no_edit__"
+            self._message_created_ts = None
+        self._already_sent = True
+        self._last_sent_text = text
+        self._final_response_sent = True
+        return True
+
     async def _send_or_edit(self, text: str, *, finalize: bool = False) -> bool:
         """Send or edit the streaming message.
 
@@ -786,6 +876,22 @@ class GatewayStreamConsumer:
                         finalize and self._adapter_requires_finalize
                     ):
                         return True
+                    # Fresh-final for long-lived previews: when finalizing
+                    # the last edit in a streaming sequence, if the
+                    # original preview has been visible for at least
+                    # ``fresh_final_after_seconds``, send the completed
+                    # reply as a fresh message so the platform's visible
+                    # timestamp reflects completion time instead of the
+                    # preview creation time.  Best-effort cleanup of the
+                    # old preview follows.  Ported from
+                    # openclaw/openclaw#72038.  Gated by config so the
+                    # legacy edit-in-place path stays the default.
+                    if (
+                        finalize
+                        and self._should_send_fresh_final()
+                        and await self._try_fresh_final(text)
+                    ):
+                        return True
                     # Edit existing message
                     result = await self.adapter.edit_message(
                         chat_id=self.chat_id,
@@ -852,6 +958,10 @@ class GatewayStreamConsumer:
                 if result.success:
                     if result.message_id:
                         self._message_id = result.message_id
+                        # Track when the preview first became visible to
+                        # the user so fresh-final logic can detect stale
+                        # preview timestamps on long-running responses.
+                        self._message_created_ts = time.monotonic()
                     else:
                         self._edit_supported = False
                     self._already_sent = True

gateway/whatsapp_identity.py

@@ -31,8 +31,17 @@ Hermes' own session keys.
 from __future__ import annotations
 
 import json
+import logging
+import re
 from typing import Set
 
+logger = logging.getLogger(__name__)
+
+# WhatsApp JIDs are numeric (or plus-prefixed numeric) with optional
+# ``@``, ``.`` and ``:`` separators. ``\w`` is pinned to ASCII so
+# full-width digits / Unicode word chars can't sneak through.
+_SAFE_IDENTIFIER_RE = re.compile(r"^[A-Za-z0-9@.+\-]+$")
+
 from hermes_constants import get_hermes_home
 
 
@@ -81,6 +90,16 @@ def expand_whatsapp_aliases(identifier: str) -> Set[str]:
         current = queue.pop(0)
         if not current or current in resolved:
             continue
+        # Defense-in-depth: reject identifiers that could sneak path
+        # separators / traversal segments into the ``lid-mapping-{current}``
+        # filename below. The hardcoded ``lid-mapping-`` prefix already
+        # prevents escape via pathlib's component split (an attacker can't
+        # create ``lid-mapping-..`` as a real directory in session_dir), but
+        # this keeps the identifier space to the characters WhatsApp JIDs
+        # actually use and avoids depending on that filesystem-layout
+        # invariant.
+        if not _SAFE_IDENTIFIER_RE.match(current):
+            continue
 
         resolved.add(current)
         for suffix in ("", "_reverse"):
@@ -91,7 +110,8 @@ def expand_whatsapp_aliases(identifier: str) -> Set[str]:
                 mapped = normalize_whatsapp_identifier(
                     json.loads(mapping_path.read_text(encoding="utf-8"))
                 )
-            except Exception:
+            except (OSError, json.JSONDecodeError) as exc:
+                logger.debug("whatsapp_identity: failed to read %s: %s", mapping_path, exc)
                 continue
             if mapped and mapped not in resolved:
                 queue.append(mapped)

hermes_cli/auth.py

@@ -467,11 +467,27 @@ def _resolve_api_key_provider_secret(
             pass
         return "", ""
 
+    from hermes_cli.config import get_env_value
     for env_var in pconfig.api_key_env_vars:
-        val = os.getenv(env_var, "").strip()
+        # Check both os.environ and ~/.hermes/.env file
+        val = (get_env_value(env_var) or "").strip()
         if has_usable_secret(val):
             return val, env_var
 
+    # Fallback: try credential pool (e.g. zai key stored via auth.json)
+    try:
+        from agent.credential_pool import load_pool
+        pool = load_pool(provider_id)
+        if pool and pool.has_credentials():
+            entry = pool.peek()
+            if entry:
+                key = getattr(entry, "access_token", "") or getattr(entry, "runtime_api_key", "")
+                key = str(key).strip()
+                if has_usable_secret(key):
+                    return key, f"credential_pool:{provider_id}"
+    except Exception:
+        pass
+
     return "", ""
 
 

hermes_cli/commands.py

@@ -126,8 +126,8 @@ COMMAND_REGISTRY: list[CommandDef] = [
     CommandDef("voice", "Toggle voice mode", "Configuration",
                args_hint="[on|off|tts|status]", subcommands=("on", "off", "tts", "status")),
     CommandDef("busy", "Control what Enter does while Hermes is working", "Configuration",
-               cli_only=True, args_hint="[queue|interrupt|status]",
-               subcommands=("queue", "interrupt", "status")),
+               cli_only=True, args_hint="[queue|steer|interrupt|status]",
+               subcommands=("queue", "steer", "interrupt", "status")),
 
     # Tools & Skills
     CommandDef("tools", "Manage tools: /tools [list|disable|enable] [name...]", "Tools & Skills",
@@ -806,6 +806,114 @@ def discord_skill_commands_by_category(
     return trimmed_categories, uncategorized, hidden
 
 
+# ---------------------------------------------------------------------------
+# Slack native slash commands
+# ---------------------------------------------------------------------------
+
+# Slack slash command name constraints: lowercase a-z, 0-9, hyphens,
+# underscores. Max 32 chars. Slack app manifest accepts up to 50 slash
+# commands per app.
+_SLACK_MAX_SLASH_COMMANDS = 50
+_SLACK_NAME_LIMIT = 32
+_SLACK_INVALID_CHARS = re.compile(r"[^a-z0-9_\-]")
+
+
+def _sanitize_slack_name(raw: str) -> str:
+    """Convert a command name to a valid Slack slash command name.
+
+    Slack allows lowercase a-z, digits, hyphens, and underscores. Max 32
+    chars. Uppercase is lowercased; invalid chars are stripped.
+    """
+    name = raw.lower()
+    name = _SLACK_INVALID_CHARS.sub("", name)
+    name = name.strip("-_")
+    return name[:_SLACK_NAME_LIMIT]
+
+
+def slack_native_slashes() -> list[tuple[str, str, str]]:
+    """Return (slash_name, description, usage_hint) triples for Slack.
+
+    Every gateway-available command in ``COMMAND_REGISTRY`` is surfaced as
+    a standalone Slack slash command (e.g. ``/btw``, ``/stop``, ``/model``),
+    matching Discord's and Telegram's model where every command is a
+    first-class slash and not a ``/hermes <verb>`` subcommand.
+
+    Both canonical names and aliases are included so users can type any
+    documented form (e.g. ``/background``, ``/bg``, and ``/btw`` all work).
+    Plugin-registered slash commands are included too.
+
+    Results are clamped to Slack's 50-command limit with duplicate-name
+    avoidance. ``/hermes`` is always reserved as the first entry so the
+    legacy ``/hermes <subcommand>`` form keeps working for anything that
+    gets dropped by the clamp or for free-form questions.
+    """
+    overrides = _resolve_config_gates()
+    entries: list[tuple[str, str, str]] = []
+    seen: set[str] = set()
+
+    # Reserve /hermes as the catch-all top-level command.
+    entries.append(("hermes", "Talk to Hermes or run a subcommand", "[subcommand] [args]"))
+    seen.add("hermes")
+
+    def _add(name: str, desc: str, hint: str) -> None:
+        slack_name = _sanitize_slack_name(name)
+        if not slack_name or slack_name in seen:
+            return
+        if len(entries) >= _SLACK_MAX_SLASH_COMMANDS:
+            return
+        # Slack description cap is 2000 chars; keep it short.
+        entries.append((slack_name, desc[:140], hint[:100]))
+        seen.add(slack_name)
+
+    # First pass: canonical names (so they win slots if we hit the cap).
+    for cmd in COMMAND_REGISTRY:
+        if not _is_gateway_available(cmd, overrides):
+            continue
+        _add(cmd.name, cmd.description, cmd.args_hint or "")
+
+    # Second pass: aliases.
+    for cmd in COMMAND_REGISTRY:
+        if not _is_gateway_available(cmd, overrides):
+            continue
+        for alias in cmd.aliases:
+            # Skip aliases that only differ from canonical by case/punctuation
+            # normalization (already covered by _add dedup).
+            _add(alias, f"Alias for /{cmd.name} — {cmd.description}", cmd.args_hint or "")
+
+    # Third pass: plugin commands.
+    for name, description, args_hint in _iter_plugin_command_entries():
+        _add(name, description, args_hint or "")
+
+    return entries
+
+
+def slack_app_manifest(request_url: str = "https://hermes-agent.local/slack/commands") -> dict[str, Any]:
+    """Generate a Slack app manifest with all gateway commands as slashes.
+
+    ``request_url`` is required by Slack's manifest schema for every slash
+    command, but in Socket Mode (which we use) Slack ignores it and routes
+    the command event through the WebSocket. A placeholder URL is fine.
+
+    The returned dict is the ``features.slash_commands`` portion only —
+    callers compose it into a full manifest (or merge into an existing
+    one). Keeping it narrow avoids coupling us to the rest of the manifest
+    schema (display_information, oauth_config, settings, etc.) which users
+    set up once in the Slack UI and rarely change.
+    """
+    slashes = []
+    for name, desc, usage in slack_native_slashes():
+        entry = {
+            "command": f"/{name}",
+            "description": desc or f"Run /{name}",
+            "should_escape": False,
+            "url": request_url,
+        }
+        if usage:
+            entry["usage_hint"] = usage
+        slashes.append(entry)
+    return {"features": {"slash_commands": slashes}}
+
+
 def slack_subcommand_map() -> dict[str, str]:
     """Return subcommand -> /command mapping for Slack /hermes handler.
 

hermes_cli/config.py

@@ -465,6 +465,7 @@ DEFAULT_CONFIG = {
         "command_timeout": 30,  # Timeout for browser commands in seconds (screenshot, navigate, etc.)
         "record_sessions": False,  # Auto-record browser sessions as WebM videos
         "allow_private_urls": False,  # Allow navigating to private/internal IPs (localhost, 192.168.x.x, etc.)
+        "auto_local_for_private_urls": True,  # When a cloud provider is set, auto-spawn local Chromium for LAN/localhost URLs instead of sending them to the cloud
         "cdp_url": "",  # Optional persistent CDP endpoint for attaching to an existing Chromium/Chrome
         # CDP supervisor — dialog + frame detection via a persistent WebSocket.
         # Active only when a CDP-capable backend is attached (Browserbase or
@@ -486,6 +487,19 @@ DEFAULT_CONFIG = {
     "checkpoints": {
         "enabled": True,
         "max_snapshots": 50,  # Max checkpoints to keep per directory
+        # Auto-maintenance: shadow repos accumulate forever under
+        # ~/.hermes/checkpoints/ (one per cd'd working directory). Field
+        # reports put the typical offender at 1000+ repos / ~12 GB. When
+        # auto_prune is on, hermes sweeps at startup (at most once per
+        # min_interval_hours) and deletes:
+        #   * orphan repos: HERMES_WORKDIR no longer exists on disk
+        #   * stale repos:  newest mtime older than retention_days
+        # Opt-in so users who rely on /rollback against long-ago sessions
+        # never lose data silently.
+        "auto_prune": False,
+        "retention_days": 7,
+        "delete_orphans": True,
+        "min_interval_hours": 24,
     },
 
     # Maximum characters returned by a single read_file call.  Reads that
@@ -626,7 +640,7 @@ DEFAULT_CONFIG = {
         "compact": False,
         "personality": "kawaii",
         "resume_display": "full",
-        "busy_input_mode": "interrupt",
+        "busy_input_mode": "interrupt",  # interrupt | queue | steer
         "bell_on_complete": False,
         "show_reasoning": False,
         "streaming": False,
@@ -1581,6 +1595,44 @@ OPTIONAL_ENV_VARS = {
         "category": "tool",
     },
 
+    # ── Bundled skills (opt-in: only needed if the user uses that skill) ──
+    # These use category="skill" (distinct from "tool") so the sandbox
+    # env blocklist in tools/environments/local.py does NOT rewrite them —
+    # skills legitimately need these passed through to curl via
+    # tools/env_passthrough.py when the user's skill calls out.
+    "NOTION_API_KEY": {
+        "description": "Notion integration token (used by the `notion` skill)",
+        "prompt": "Notion API key",
+        "url": "https://www.notion.so/my-integrations",
+        "password": True,
+        "category": "skill",
+        "advanced": True,
+    },
+    "LINEAR_API_KEY": {
+        "description": "Linear personal API key (used by the `linear` skill)",
+        "prompt": "Linear API key",
+        "url": "https://linear.app/settings/api",
+        "password": True,
+        "category": "skill",
+        "advanced": True,
+    },
+    "AIRTABLE_API_KEY": {
+        "description": "Airtable personal access token (used by the `airtable` skill)",
+        "prompt": "Airtable API key",
+        "url": "https://airtable.com/create/tokens",
+        "password": True,
+        "category": "skill",
+        "advanced": True,
+    },
+    "TENOR_API_KEY": {
+        "description": "Tenor API key for GIF search (used by the `gif-search` skill)",
+        "prompt": "Tenor API key",
+        "url": "https://developers.google.com/tenor/guides/quickstart",
+        "password": True,
+        "category": "skill",
+        "advanced": True,
+    },
+
     # ── Honcho ──
     "HONCHO_API_KEY": {
         "description": "Honcho API key for AI-native persistent memory",

hermes_cli/gateway.py

@@ -2724,6 +2724,24 @@ _PLATFORMS = [
              "help": "OpenID to deliver cron results and notifications to."},
         ],
     },
+    {
+        "key": "yuanbao",
+        "label": "Yuanbao",
+        "emoji": "💎",
+        "token_var": "YUANBAO_APP_ID",
+        "setup_instructions": [
+            "1. Download the Yuanbao app from https://yuanbao.tencent.com/",
+            "2. In the app, go to PAI → My Bot and create a new bot",
+            "3. After the bot is created, copy the App ID and App Secret",
+            "4. Enter them below and Hermes will connect automatically over WebSocket",
+        ],
+        "vars": [
+            {"name": "YUANBAO_APP_ID", "prompt": "App ID", "password": False,
+             "help": "The App ID from your Yuanbao IM Bot credentials."},
+            {"name": "YUANBAO_APP_SECRET", "prompt": "App Secret", "password": True,
+             "help": "The App Secret (used for HMAC signing) from your Yuanbao IM Bot."},
+        ],
+    },
 ]
 
 
@@ -3108,6 +3126,12 @@ def _setup_wecom():
     print_success("💬 WeCom configured!")
 
 
+def _setup_yuanbao():
+    """Configure Yuanbao via the standard platform setup."""
+    yuanbao_platform = next(p for p in _PLATFORMS if p["key"] == "yuanbao")
+    _setup_standard_platform(yuanbao_platform)
+
+
 def _is_service_installed() -> bool:
     """Check if the gateway is installed as a system service."""
     if supports_systemd_services():

hermes_cli/main.py

@@ -1043,6 +1043,7 @@ def _launch_tui(
     )
     env.setdefault("HERMES_PYTHON", sys.executable)
     env.setdefault("HERMES_CWD", os.getcwd())
+    env.setdefault("NODE_ENV", "development" if tui_dev else "production")
     if model:
         env["HERMES_MODEL"] = model
         env["HERMES_INFERENCE_MODEL"] = model
@@ -3331,7 +3332,26 @@ def _model_flow_named_custom(config, provider_info):
             provider_entry = providers_cfg.get(provider_key)
             if isinstance(provider_entry, dict):
                 provider_entry["default_model"] = model_name
-                if config_api_key and not str(provider_entry.get("api_key", "") or "").strip():
+                # Only persist an inline api_key when the user originally had
+                # one (either a literal secret or a ``${VAR}`` template). When
+                # the entry relies on ``key_env``, do not synthesize a
+                # ``${key_env}`` api_key — the runtime already resolves the
+                # key from ``key_env`` directly, and writing the resolved
+                # secret (or even a synthesized template) would silently
+                # downgrade credential hygiene on entries that intentionally
+                # keep plaintext out of ``config.yaml``. See issue #15803.
+                original_api_key_ref = str(
+                    provider_info.get("api_key_ref", "") or ""
+                ).strip()
+                original_api_key = str(
+                    provider_info.get("api_key", "") or ""
+                ).strip()
+                had_inline_api_key = bool(original_api_key_ref or original_api_key)
+                if (
+                    had_inline_api_key
+                    and config_api_key
+                    and not str(provider_entry.get("api_key", "") or "").strip()
+                ):
                     provider_entry["api_key"] = config_api_key
                 if key_env and not str(provider_entry.get("key_env", "") or "").strip():
                     provider_entry["key_env"] = key_env
@@ -4412,8 +4432,14 @@ def _model_flow_api_key_provider(config, provider_id, current_model=""):
         from hermes_cli.models import fetch_ollama_cloud_models
 
         api_key_for_probe = existing_key or (get_env_value(key_env) if key_env else "")
+        # During setup, force a live refresh so the picker reflects newly
+        # released models (e.g. deepseek v4 flash, kimi k2.6) the moment
+        # the user enters their key — not an hour later when the disk
+        # cache TTL expires.
         model_list = fetch_ollama_cloud_models(
-            api_key=api_key_for_probe, base_url=effective_base
+            api_key=api_key_for_probe,
+            base_url=effective_base,
+            force_refresh=True,
         )
         if model_list:
             print(f"  Found {len(model_list)} model(s) from Ollama Cloud")
@@ -4780,6 +4806,37 @@ def cmd_webhook(args):
     webhook_command(args)
 
 
+def cmd_slack(args):
+    """Slack integration helpers.
+
+    Dispatches ``hermes slack <subcommand>``. Currently supports:
+      manifest — print or write a Slack app manifest with every gateway
+                 command registered as a first-class slash.
+    """
+    sub = getattr(args, "slack_command", None)
+    if sub in (None, ""):
+        # No subcommand — print usage hint.
+        print(
+            "usage: hermes slack <subcommand>\n"
+            "\n"
+            "subcommands:\n"
+            "  manifest   Generate a Slack app manifest with every gateway\n"
+            "             command registered as a native slash\n"
+            "\n"
+            "Run `hermes slack manifest -h` for details.",
+            file=sys.stderr,
+        )
+        return 1
+
+    if sub == "manifest":
+        from hermes_cli.slack_cli import slack_manifest_command
+
+        return slack_manifest_command(args)
+
+    print(f"Unknown slack subcommand: {sub}", file=sys.stderr)
+    return 1
+
+
 def cmd_hooks(args):
     """Shell-hook inspection and management."""
     from hermes_cli.hooks import hooks_command
@@ -4953,6 +5010,83 @@ def _gateway_prompt(prompt_text: str, default: str = "", timeout: float = 300.0)
     return default
 
 
+def _web_ui_build_needed(web_dir: Path) -> bool:
+    """Return True if the web UI dist is missing or stale.
+
+    Mirrors the staleness logic used by ``_tui_build_needed()`` for the TUI.
+    The Vite build outputs to ``hermes_cli/web_dist/`` (per vite.config.ts
+    outDir: "../hermes_cli/web_dist"), NOT to ``web/dist/``.  Uses the Vite
+    manifest as the sentinel because it is written last and therefore has the
+    newest mtime of any build output.
+    """
+    dist_dir = web_dir.parent / "hermes_cli" / "web_dist"
+    sentinel = dist_dir / ".vite" / "manifest.json"
+    if not sentinel.exists():
+        sentinel = dist_dir / "index.html"
+    if not sentinel.exists():
+        return True
+    dist_mtime = sentinel.stat().st_mtime
+    skip = frozenset({"node_modules", "dist"})
+    for dirpath, dirnames, filenames in os.walk(web_dir, topdown=True):
+        dirnames[:] = [d for d in dirnames if d not in skip]
+        for fn in filenames:
+            if fn.endswith((".ts", ".tsx", ".js", ".jsx", ".css", ".html", ".vue")):
+                if os.path.getmtime(os.path.join(dirpath, fn)) > dist_mtime:
+                    return True
+    for meta in (
+        "package.json",
+        "package-lock.json",
+        "yarn.lock",
+        "pnpm-lock.yaml",
+        "vite.config.ts",
+        "vite.config.js",
+    ):
+        mp = web_dir / meta
+        if mp.exists() and mp.stat().st_mtime > dist_mtime:
+            return True
+    return False
+
+
+def _run_npm_install_deterministic(
+    npm: str,
+    cwd: Path,
+    *,
+    extra_args: tuple[str, ...] = (),
+    capture_output: bool = True,
+) -> subprocess.CompletedProcess:
+    """Run a deterministic npm install that does not mutate ``package-lock.json``.
+
+    Prefers ``npm ci`` (strict, lockfile-preserving) when a lockfile is present;
+    falls back to ``npm install`` only if ``npm ci`` fails (e.g. lockfile out of
+    sync on a WIP checkout).  Without this, ``npm install`` on npm ≥ 10 silently
+    rewrites committed lockfiles (stripping ``"peer": true`` etc.), which leaves
+    the working tree dirty and causes the next ``hermes update`` to stash the
+    lockfile — repeatedly.
+    """
+    lockfile = cwd / "package-lock.json"
+    if lockfile.exists():
+        ci_cmd = [npm, "ci", *extra_args]
+        ci_result = subprocess.run(
+            ci_cmd,
+            cwd=cwd,
+            capture_output=capture_output,
+            text=True,
+            check=False,
+        )
+        if ci_result.returncode == 0:
+            return ci_result
+        # Fall through to `npm install` — lockfile may be out of sync on a
+        # WIP fork/branch, or `npm ci` may not be available on very old npm.
+    install_cmd = [npm, "install", *extra_args]
+    return subprocess.run(
+        install_cmd,
+        cwd=cwd,
+        capture_output=capture_output,
+        text=True,
+        check=False,
+    )
+
+
 def _build_web_ui(web_dir: Path, *, fatal: bool = False) -> bool:
     """Build the web UI frontend if npm is available.
 
@@ -4966,6 +5100,9 @@ def _build_web_ui(web_dir: Path, *, fatal: bool = False) -> bool:
     if not (web_dir / "package.json").exists():
         return True
 
+    if not _web_ui_build_needed(web_dir):
+        return True
+
     npm = shutil.which("npm")
     if not npm:
         if fatal:
@@ -4973,7 +5110,7 @@ def _build_web_ui(web_dir: Path, *, fatal: bool = False) -> bool:
             print("Install Node.js, then run:  cd web && npm install && npm run build")
         return not fatal
     print("→ Building web UI...")
-    r1 = subprocess.run([npm, "install", "--silent"], cwd=web_dir, capture_output=True)
+    r1 = _run_npm_install_deterministic(npm, web_dir, extra_args=("--silent",))
     if r1.returncode != 0:
         print(
             f"  {'✗' if fatal else '⚠'} Web UI npm install failed"
@@ -5684,12 +5821,10 @@ def _update_node_dependencies() -> None:
         if not (path / "package.json").exists():
             continue
 
-        result = subprocess.run(
-            [npm, "install", "--silent", "--no-fund", "--no-audit", "--progress=false"],
-            cwd=path,
-            capture_output=True,
-            text=True,
-            check=False,
+        result = _run_npm_install_deterministic(
+            npm,
+            path,
+            extra_args=("--silent", "--no-fund", "--no-audit", "--progress=false"),
         )
         if result.returncode == 0:
             print(f"  ✓ {label}")
@@ -5925,6 +6060,88 @@ def _cmd_update_check():
         print(f"  Run '{recommended_update_command()}' to install.")
 
 
+def _ensure_fhs_path_guard() -> None:
+    """Ensure /usr/local/bin is on PATH for RHEL-family root non-login shells.
+
+    Mirrors the post-symlink probe added to ``scripts/install.sh`` so that
+    existing FHS-layout root installs on RHEL/CentOS/Rocky/Alma 8+ get
+    repaired on ``hermes update`` without requiring a reinstall.  The
+    installer's assumption that ``/usr/local/bin`` is on PATH for every
+    standard shell breaks on those distros in non-login interactive shells
+    (su, sudo -s, tmux panes, some web terminals): /etc/bashrc doesn't
+    add /usr/local/bin and /root/.bash_profile doesn't either.  Symptom:
+    ``hermes`` prints ``command not found`` even though the symlink lives
+    at /usr/local/bin/hermes.
+
+    Silent no-op on: non-Linux, non-root, non-FHS installs, and any system
+    where ``bash -i -c 'command -v hermes'`` already resolves.  Idempotent.
+    """
+    if sys.platform != "linux":
+        return
+    try:
+        if os.geteuid() != 0:
+            return
+    except AttributeError:
+        return
+    # Only act when this is actually an FHS-layout install (command link at
+    # /usr/local/bin/hermes, code at /usr/local/lib/hermes-agent).
+    fhs_link = Path("/usr/local/bin/hermes")
+    if not fhs_link.is_symlink() and not fhs_link.exists():
+        return
+
+    # Probe a fresh non-login interactive bash the way the user will use it.
+    # ``bash -i -c`` sources ~/.bashrc but NOT ~/.bash_profile or /etc/profile,
+    # which is the exact scenario where RHEL root loses /usr/local/bin.
+    home = os.environ.get("HOME") or "/root"
+    try:
+        probe = subprocess.run(
+            ["env", "-i",
+             f"HOME={home}",
+             f"TERM={os.environ.get('TERM', 'dumb')}",
+             "bash", "-i", "-c", "command -v hermes"],
+            capture_output=True, text=True, timeout=10,
+        )
+    except (FileNotFoundError, subprocess.TimeoutExpired):
+        return  # no bash or probe hung — don't block update on this
+    if probe.returncode == 0:
+        return  # already on PATH, nothing to do
+
+    path_line = 'export PATH="/usr/local/bin:$PATH"'
+    path_comment = (
+        "# Hermes Agent — ensure /usr/local/bin is on PATH "
+        "(RHEL non-login shells)"
+    )
+    wrote_any = False
+    for candidate in (".bashrc", ".bash_profile"):
+        cfg = Path(home) / candidate
+        if not cfg.is_file():
+            continue
+        try:
+            existing = cfg.read_text(errors="replace")
+        except OSError:
+            continue
+        # Idempotency: skip if any uncommented PATH= line already references
+        # /usr/local/bin.  Mirrors the grep pattern used by install.sh.
+        already_guarded = any(
+            "/usr/local/bin" in line
+            and "PATH" in line
+            and not line.lstrip().startswith("#")
+            for line in existing.splitlines()
+        )
+        if already_guarded:
+            continue
+        try:
+            with cfg.open("a", encoding="utf-8") as f:
+                f.write("\n" + path_comment + "\n" + path_line + "\n")
+        except OSError as e:
+            print(f"  ⚠ Could not update {cfg}: {e}")
+            continue
+        print(f"  ✓ Added /usr/local/bin to PATH in {cfg}")
+        wrote_any = True
+    if wrote_any:
+        print("    (reload your shell or run 'source ~/.bashrc' to pick it up)")
+
+
 def cmd_update(args):
     """Update Hermes Agent to the latest version.
 
@@ -6368,6 +6585,13 @@ def _cmd_update_impl(args, gateway_mode: bool):
         print()
         print("✓ Update complete!")
 
+        # Repair RHEL-family root installs where /usr/local/bin isn't on PATH
+        # for non-login interactive shells.  No-op on every other platform.
+        try:
+            _ensure_fhs_path_guard()
+        except Exception as e:
+            logger.debug("FHS PATH guard check failed: %s", e)
+
         # Write exit code *before* the gateway restart attempt.
         # When running as ``hermes update --gateway`` (spawned by the gateway's
         # /update command), this process lives inside the gateway's systemd
@@ -7798,6 +8022,54 @@ For more help on a command:
     )
     whatsapp_parser.set_defaults(func=cmd_whatsapp)
 
+    # =========================================================================
+    # slack command
+    # =========================================================================
+    slack_parser = subparsers.add_parser(
+        "slack",
+        help="Slack integration helpers (manifest generation, etc.)",
+        description="Slack integration helpers for Hermes.",
+    )
+    slack_sub = slack_parser.add_subparsers(dest="slack_command")
+    slack_manifest = slack_sub.add_parser(
+        "manifest",
+        help="Print or write a Slack app manifest with every gateway command "
+             "registered as a native slash (/btw, /stop, /model, ...)",
+        description=(
+            "Generate a Slack app manifest that registers every gateway "
+            "command in COMMAND_REGISTRY as a first-class Slack slash "
+            "command (matching Discord and Telegram parity). Paste the "
+            "output into Slack app config → Features → App Manifest → "
+            "Edit, then Save. Reinstall the app if Slack prompts for it."
+        ),
+    )
+    slack_manifest.add_argument(
+        "--write",
+        nargs="?",
+        const=True,
+        default=None,
+        metavar="PATH",
+        help="Write manifest to a file instead of stdout. With no PATH "
+             "writes to $HERMES_HOME/slack-manifest.json.",
+    )
+    slack_manifest.add_argument(
+        "--name",
+        default=None,
+        help='Bot display name (default: "Hermes")',
+    )
+    slack_manifest.add_argument(
+        "--description",
+        default=None,
+        help="Bot description shown in Slack's app directory.",
+    )
+    slack_manifest.add_argument(
+        "--slashes-only",
+        action="store_true",
+        help="Emit only the features.slash_commands array (for merging "
+             "into an existing manifest manually).",
+    )
+    slack_parser.set_defaults(func=cmd_slack)
+
     # =========================================================================
     # login command
     # =========================================================================
@@ -8429,11 +8701,17 @@ Examples:
 
     skills_install = skills_subparsers.add_parser("install", help="Install a skill")
     skills_install.add_argument(
-        "identifier", help="Skill identifier (e.g. openai/skills/skill-creator)"
+        "identifier",
+        help="Skill identifier (e.g. openai/skills/skill-creator) or a direct HTTP(S) URL to a SKILL.md file",
     )
     skills_install.add_argument(
         "--category", default="", help="Category folder to install into"
     )
+    skills_install.add_argument(
+        "--name",
+        default="",
+        help="Override the skill name (useful when installing from a URL whose SKILL.md has no `name:` frontmatter)",
+    )
     skills_install.add_argument(
         "--force", action="store_true", help="Install despite blocked scan verdict"
     )
@@ -8453,6 +8731,12 @@ Examples:
     skills_list.add_argument(
         "--source", default="all", choices=["all", "hub", "builtin", "local"]
     )
+    skills_list.add_argument(
+        "--enabled-only",
+        action="store_true",
+        help="Hide disabled skills. Use with -p <profile> to see exactly "
+             "which skills will load for that profile.",
+    )
 
     skills_check = skills_subparsers.add_parser(
         "check", help="Check installed hub skills for updates"
@@ -8959,7 +9243,7 @@ Examples:
         "--source", help="Filter by source (cli, telegram, discord, etc.)"
     )
     sessions_browse.add_argument(
-        "--limit", type=int, default=50, help="Max sessions to load (default: 50)"
+        "--limit", type=int, default=500, help="Max sessions to load (default: 500)"
     )
 
     def _confirm_prompt(prompt: str) -> bool:
@@ -9056,7 +9340,8 @@ Examples:
                 ):
                     print("Cancelled.")
                     return
-            if db.delete_session(resolved_session_id):
+            sessions_dir = get_hermes_home() / "sessions"
+            if db.delete_session(resolved_session_id, sessions_dir=sessions_dir):
                 print(f"Deleted session '{resolved_session_id}'.")
             else:
                 print(f"Session '{args.session_id}' not found.")
@@ -9070,7 +9355,9 @@ Examples:
                 ):
                     print("Cancelled.")
                     return
-            count = db.prune_sessions(older_than_days=days, source=args.source)
+            sessions_dir = get_hermes_home() / "sessions"
+            count = db.prune_sessions(older_than_days=days, source=args.source,
+                                      sessions_dir=sessions_dir)
             print(f"Pruned {count} session(s).")
 
         elif action == "rename":
@@ -9088,7 +9375,7 @@ Examples:
                 print(f"Error: {e}")
 
         elif action == "browse":
-            limit = getattr(args, "limit", 50) or 50
+            limit = getattr(args, "limit", 500) or 500
             source = getattr(args, "source", None)
             _browse_exclude = None if source else ["tool"]
             sessions = db.list_sessions_rich(

hermes_cli/models.py

@@ -33,8 +33,6 @@ COPILOT_REASONING_EFFORTS_O_SERIES = ["low", "medium", "high"]
 # (model_id, display description shown in menus)
 OPENROUTER_MODELS: list[tuple[str, str]] = [
     ("moonshotai/kimi-k2.6",            "recommended"),
-    ("deepseek/deepseek-v4-pro",        ""),
-    ("deepseek/deepseek-v4-flash",      ""),
     ("anthropic/claude-opus-4.7",       ""),
     ("anthropic/claude-opus-4.6",       ""),
     ("anthropic/claude-sonnet-4.6",     ""),
@@ -111,8 +109,6 @@ def _codex_curated_models() -> list[str]:
 _PROVIDER_MODELS: dict[str, list[str]] = {
     "nous": [
         "moonshotai/kimi-k2.6",
-        "deepseek/deepseek-v4-pro",
-        "deepseek/deepseek-v4-flash",
         "xiaomi/mimo-v2.5-pro",
         "xiaomi/mimo-v2.5",
         "anthropic/claude-opus-4.7",
@@ -2230,6 +2226,52 @@ def copilot_model_api_mode(
     return "chat_completions"
 
 
+# Azure Foundry model families that require the Responses API.  Azure
+# rejects /chat/completions against these deployments with
+# ``400 "The requested operation is unsupported."`` — the same payload Bob
+# Dobolina hit in April 2026 on ``gpt-5.3-codex`` while ``gpt-4o-pure`` on
+# the same endpoint worked fine.  Keep the patterns broad enough to cover
+# vendor-renamed deployments (e.g. ``gpt-5.3-codex``, ``gpt-5-codex``,
+# ``gpt-5.4``, ``o1-preview``) but tight enough to leave GPT-4 / 3.5 / Llama /
+# Mistral / Grok deployments on chat completions.
+_AZURE_FOUNDRY_RESPONSES_PREFIXES = (
+    "codex",       # codex-*, codex-mini
+    "gpt-5",       # gpt-5, gpt-5.x, gpt-5-codex, gpt-5.x-codex
+    "o1",          # o1, o1-preview, o1-mini
+    "o3",          # o3, o3-mini
+    "o4",          # o4, o4-mini
+)
+
+
+def azure_foundry_model_api_mode(model_name: Optional[str]) -> Optional[str]:
+    """Infer Azure Foundry api_mode from a deployment/model name.
+
+    Returns ``"codex_responses"`` when the model name matches a family that
+    only accepts the Responses API on Azure Foundry (GPT-5.x, codex, o1/o3/o4
+    reasoning models).  Returns ``None`` otherwise — the caller should fall
+    back to the configured/default api_mode (typically ``chat_completions``)
+    so GPT-4o, GPT-4 Turbo, Llama, Mistral, etc. keep working.
+
+    Intentionally does NOT return ``anthropic_messages``; Anthropic-style
+    Azure endpoints are disambiguated by URL (``/anthropic`` suffix) in
+    ``runtime_provider._detect_api_mode_for_url`` and by the user setting
+    ``model.api_mode: anthropic_messages`` explicitly.
+    """
+    raw = str(model_name or "").strip().lower()
+    if not raw:
+        return None
+    # Strip any vendor/ prefix a user may have copied from OpenRouter / Copilot.
+    if "/" in raw:
+        raw = raw.rsplit("/", 1)[-1]
+    # gpt-5-mini speaks chat completions on Copilot but Azure Foundry deploys
+    # the full gpt-5 family uniformly on Responses API — don't carve an
+    # exception here.
+    for prefix in _AZURE_FOUNDRY_RESPONSES_PREFIXES:
+        if raw.startswith(prefix):
+            return "codex_responses"
+    return None
+
+
 def normalize_opencode_model_id(provider_id: Optional[str], model_id: Optional[str]) -> str:
     """Normalize OpenCode config IDs to the bare model slug used in API requests."""
     provider = normalize_provider(provider_id)

hermes_cli/nous_subscription.py

@@ -9,6 +9,7 @@ from typing import Dict, Iterable, Optional, Set
 from hermes_cli.auth import get_nous_auth_status
 from hermes_cli.config import get_env_value, load_config
 from tools.managed_tool_gateway import is_managed_tool_gateway_ready
+from utils import is_truthy_value
 from tools.tool_backend_helpers import (
     fal_key_is_configured,
     has_direct_modal_credentials,
@@ -25,6 +26,13 @@ _DEFAULT_PLATFORM_TOOLSETS = {
 }
 
 
+def _uses_gateway(section: object) -> bool:
+    """Return True when a config section explicitly opts into the gateway."""
+    if not isinstance(section, dict):
+        return False
+    return is_truthy_value(section.get("use_gateway"), default=False)
+
+
 @dataclass(frozen=True)
 class NousFeatureState:
     key: str
@@ -262,11 +270,11 @@ def get_nous_subscription_features(
     # use_gateway flags — when True, the user explicitly opted into the
     # Tool Gateway via `hermes model`, so direct credentials should NOT
     # prevent gateway routing.
-    web_use_gateway = bool(web_cfg.get("use_gateway"))
-    tts_use_gateway = bool(tts_cfg.get("use_gateway"))
-    browser_use_gateway = bool(browser_cfg.get("use_gateway"))
+    web_use_gateway = _uses_gateway(web_cfg)
+    tts_use_gateway = _uses_gateway(tts_cfg)
+    browser_use_gateway = _uses_gateway(browser_cfg)
     image_gen_cfg = config.get("image_gen") if isinstance(config.get("image_gen"), dict) else {}
-    image_use_gateway = bool(image_gen_cfg.get("use_gateway"))
+    image_use_gateway = _uses_gateway(image_gen_cfg)
 
     direct_exa = bool(get_env_value("EXA_API_KEY"))
     direct_firecrawl = bool(get_env_value("FIRECRAWL_API_KEY") or get_env_value("FIRECRAWL_API_URL"))
@@ -601,10 +609,10 @@ def get_gateway_eligible_tools(
     # no direct keys exist — we only skip the prompt for tools where
     # use_gateway was explicitly set.
     opted_in = {
-        "web": bool((config.get("web") if isinstance(config.get("web"), dict) else {}).get("use_gateway")),
-        "image_gen": bool((config.get("image_gen") if isinstance(config.get("image_gen"), dict) else {}).get("use_gateway")),
-        "tts": bool((config.get("tts") if isinstance(config.get("tts"), dict) else {}).get("use_gateway")),
-        "browser": bool((config.get("browser") if isinstance(config.get("browser"), dict) else {}).get("use_gateway")),
+        "web": _uses_gateway(config.get("web")),
+        "image_gen": _uses_gateway(config.get("image_gen")),
+        "tts": _uses_gateway(config.get("tts")),
+        "browser": _uses_gateway(config.get("browser")),
     }
 
     unconfigured: list[str] = []

hermes_cli/platforms.py

@@ -36,6 +36,7 @@ PLATFORMS: OrderedDict[str, PlatformInfo] = OrderedDict([
     ("wecom_callback", PlatformInfo(label="💬 WeCom Callback",  default_toolset="hermes-wecom-callback")),
     ("weixin",         PlatformInfo(label="💬 Weixin",          default_toolset="hermes-weixin")),
     ("qqbot",          PlatformInfo(label="💬 QQBot",           default_toolset="hermes-qqbot")),
+    ("yuanbao",        PlatformInfo(label="🤖 Yuanbao",         default_toolset="hermes-yuanbao")),
     ("webhook",        PlatformInfo(label="🔗 Webhook",         default_toolset="hermes-webhook")),
     ("api_server",     PlatformInfo(label="🌐 API Server",      default_toolset="hermes-api-server")),
     ("cron",           PlatformInfo(label="⏰ Cron",            default_toolset="hermes-cron")),

hermes_cli/runtime_provider.py

@@ -231,6 +231,19 @@ def _resolve_runtime_from_pool_entry(
             configured_mode = _parse_api_mode(model_cfg.get("api_mode"))
             if configured_mode:
                 api_mode = configured_mode
+        # Model-family inference for GPT-5.x / codex / o1-o4: Azure rejects
+        # /chat/completions on these with 400 "operation unsupported" — see
+        # azure_foundry_model_api_mode() for rationale.  Skip when the user
+        # explicitly picked anthropic_messages (Anthropic-style endpoint).
+        if effective_model and api_mode != "anthropic_messages":
+            try:
+                from hermes_cli.models import azure_foundry_model_api_mode
+
+                inferred = azure_foundry_model_api_mode(effective_model)
+            except Exception:
+                inferred = None
+            if inferred:
+                api_mode = inferred
         # For Anthropic-style endpoints, strip /v1 suffix
         if api_mode == "anthropic_messages":
             base_url = re.sub(r"/v1/?$", "", base_url)
@@ -608,6 +621,7 @@ def _resolve_azure_foundry_runtime(
     model_cfg: Dict[str, Any],
     explicit_api_key: Optional[str] = None,
     explicit_base_url: Optional[str] = None,
+    target_model: Optional[str] = None,
 ) -> Dict[str, Any]:
     """Resolve an Azure Foundry runtime entry.
 
@@ -628,6 +642,22 @@ def _resolve_azure_foundry_runtime(
         cfg_base_url = str(model_cfg.get("base_url") or "").strip().rstrip("/")
         cfg_api_mode = _parse_api_mode(model_cfg.get("api_mode")) or "chat_completions"
 
+    # Model-family inference: Azure Foundry deploys GPT-5.x / codex / o1-o4
+    # reasoning models as Responses-API-only.  Calling /chat/completions
+    # against them returns 400 "The requested operation is unsupported."
+    # Upgrade api_mode when the model name matches, unless the user has
+    # explicitly chosen anthropic_messages (Anthropic-style endpoint).
+    effective_model = str(target_model or model_cfg.get("default") or "").strip()
+    if effective_model and cfg_api_mode != "anthropic_messages":
+        try:
+            from hermes_cli.models import azure_foundry_model_api_mode
+
+            inferred = azure_foundry_model_api_mode(effective_model)
+        except Exception:
+            inferred = None
+        if inferred:
+            cfg_api_mode = inferred
+
     env_base_url = os.getenv("AZURE_FOUNDRY_BASE_URL", "").strip().rstrip("/")
     base_url = explicit_base_url_clean or cfg_base_url or env_base_url
     if not base_url:
@@ -864,6 +894,7 @@ def resolve_runtime_provider(
             model_cfg=_get_model_config(),
             explicit_api_key=explicit_api_key,
             explicit_base_url=explicit_base_url,
+            target_model=target_model,
         )
         return azure_runtime
 

hermes_cli/setup.py

@@ -1856,27 +1856,32 @@ def _setup_slack():
     if existing:
         print_info("Slack: already configured")
         if not prompt_yes_no("Reconfigure Slack?", False):
+            # Even without reconfiguring, offer to refresh the manifest so
+            # new commands (e.g. /btw, /stop, ...) get registered in Slack.
+            if prompt_yes_no(
+                "Regenerate the Slack app manifest with the latest command "
+                "list? (recommended after `hermes update`)",
+                True,
+            ):
+                _write_slack_manifest_and_instruct()
             return
 
     print_info("Steps to create a Slack app:")
-    print_info("   1. Go to https://api.slack.com/apps → Create New App (from scratch)")
+    print_info("   1. Go to https://api.slack.com/apps → Create New App")
+    print_info("      Pick 'From an app manifest' — we'll generate one for you below.")
     print_info("   2. Enable Socket Mode: Settings → Socket Mode → Enable")
     print_info("      • Create an App-Level Token with 'connections:write' scope")
-    print_info("   3. Add Bot Token Scopes: Features → OAuth & Permissions")
-    print_info("      Required scopes: chat:write, app_mentions:read,")
-    print_info("      channels:history, channels:read, im:history,")
-    print_info("      im:read, im:write, users:read, files:read, files:write")
-    print_info("      Optional for private channels: groups:history")
-    print_info("   4. Subscribe to Events: Features → Event Subscriptions → Enable")
-    print_info("      Required events: message.im, message.channels, app_mention")
-    print_info("      Optional for private channels: message.groups")
-    print_warning("   ⚠ Without message.channels the bot will ONLY work in DMs,")
-    print_warning("     not public channels.")
-    print_info("   5. Install to Workspace: Settings → Install App")
-    print_info("   6. Reinstall the app after any scope or event changes")
-    print_info("   7. After installing, invite the bot to channels: /invite @YourBot")
+    print_info("   3. Install to Workspace: Settings → Install App")
+    print_info("   4. After installing, invite the bot to channels: /invite @YourBot")
     print()
     print_info("   Full guide: https://hermes-agent.nousresearch.com/docs/user-guide/messaging/slack/")
+    print()
+
+    # Generate and write manifest up-front so the user can paste it into
+    # the "Create from manifest" flow instead of clicking through scopes /
+    # events / slash commands one at a time.
+    _write_slack_manifest_and_instruct()
+
     print()
     bot_token = prompt("Slack Bot Token (xoxb-...)", password=True)
     if not bot_token:
@@ -1902,6 +1907,49 @@ def _setup_slack():
         print_info("   Set SLACK_ALLOW_ALL_USERS=true or GATEWAY_ALLOW_ALL_USERS=true only if you intentionally want open workspace access.")
 
 
+def _write_slack_manifest_and_instruct():
+    """Generate the Slack manifest, write it under HERMES_HOME, and print
+    paste-into-Slack instructions.
+
+    Exposed as its own helper so both the initial setup flow and the
+    "reconfigure? → no" branch can refresh the manifest without the user
+    re-entering tokens. Failures are non-fatal — if the manifest write
+    fails for any reason, we print a warning and skip rather than abort
+    the whole Slack setup.
+    """
+    try:
+        from hermes_cli.slack_cli import _build_full_manifest
+        from hermes_constants import get_hermes_home
+
+        manifest = _build_full_manifest(
+            bot_name="Hermes",
+            bot_description="Your Hermes agent on Slack",
+        )
+        target = Path(get_hermes_home()) / "slack-manifest.json"
+        target.parent.mkdir(parents=True, exist_ok=True)
+        import json as _json
+        target.write_text(
+            _json.dumps(manifest, indent=2, ensure_ascii=False) + "\n",
+            encoding="utf-8",
+        )
+        print_success(f"Slack app manifest written to: {target}")
+        print_info(
+            "   Paste it into https://api.slack.com/apps → your app → Features "
+            "→ App Manifest → Edit, then Save.  Slack will prompt to "
+            "reinstall if scopes or slash commands changed."
+        )
+        print_info(
+            "   Re-run `hermes slack manifest --write` anytime to refresh after "
+            "Hermes adds new commands."
+        )
+    except Exception as exc:  # pragma: no cover - best-effort UX helper
+        print_warning(f"Couldn't write Slack manifest: {exc}")
+        print_info(
+            "   You can generate it manually later with: "
+            "hermes slack manifest --write"
+        )
+
+
 def _setup_matrix():
     """Configure Matrix credentials."""
     print_header("Matrix")
@@ -2085,6 +2133,12 @@ def _setup_feishu():
     _gateway_setup_feishu()
 
 
+def _setup_yuanbao():
+    """Configure Yuanbao via gateway setup."""
+    from hermes_cli.gateway import _setup_yuanbao as _gateway_setup_yuanbao
+    _gateway_setup_yuanbao()
+
+
 def _setup_wecom():
     """Configure WeCom (Enterprise WeChat) via gateway setup."""
     from hermes_cli.gateway import _setup_wecom as _gateway_setup_wecom
@@ -2229,6 +2283,7 @@ _GATEWAY_PLATFORMS = [
     ("WhatsApp", "WHATSAPP_ENABLED", _setup_whatsapp),
     ("DingTalk", "DINGTALK_CLIENT_ID", _setup_dingtalk),
     ("Feishu / Lark", "FEISHU_APP_ID", _setup_feishu),
+    ("Yuanbao", "YUANBAO_APP_ID", _setup_yuanbao),
     ("WeCom (Enterprise WeChat)", "WECOM_BOT_ID", _setup_wecom),
     ("WeCom Callback (Self-Built App)", "WECOM_CALLBACK_CORP_ID", _setup_wecom_callback),
     ("Weixin (WeChat)", "WEIXIN_ACCOUNT_ID", _setup_weixin),

hermes_cli/skills_hub.py

@@ -11,9 +11,10 @@ handler are thin wrappers that parse args and delegate.
 """
 
 import json
+import re
 import shutil
 from pathlib import Path
-from typing import Any, Dict, Optional
+from typing import Any, Dict, List, Optional
 
 from rich.console import Console
 from rich.panel import Panel
@@ -141,6 +142,103 @@ def _derive_category_from_install_path(install_path: str) -> str:
     return "" if parent == "." else parent
 
 
+# ---------------------------------------------------------------------------
+# Interactive name/category resolution for URL-installed skills
+# ---------------------------------------------------------------------------
+
+_VALID_NAME_RE = re.compile(r"^[a-z][a-z0-9_-]*$")
+_VALID_CATEGORY_RE = re.compile(r"^[a-z][a-z0-9_/-]*$")
+
+
+def _is_valid_installed_skill_name(name: str) -> bool:
+    """Accept identifier-shaped names, reject empty / sentinel-y values."""
+    if not isinstance(name, str):
+        return False
+    candidate = name.strip().lower()
+    if not candidate or candidate in {"skill", "readme", "index", "unnamed-skill"}:
+        return False
+    return bool(_VALID_NAME_RE.match(candidate))
+
+
+def _existing_categories() -> List[str]:
+    """Return sorted subdirectory names under ``~/.hermes/skills/`` that look
+    like category buckets (contain at least one ``SKILL.md`` somewhere below).
+
+    Used to suggest reusable categories when interactively installing from a
+    URL. Hidden dirs (``.hub``, ``.trash``) are skipped.
+    """
+    from tools.skills_hub import SKILLS_DIR
+    out: List[str] = []
+    try:
+        for entry in SKILLS_DIR.iterdir():
+            if not entry.is_dir() or entry.name.startswith("."):
+                continue
+            # Only count as a category if it contains skills, not if it IS a skill.
+            # Heuristic: if ``<entry>/SKILL.md`` exists, it's a skill at the
+            # top level (no category); otherwise treat as a category bucket.
+            if (entry / "SKILL.md").exists():
+                continue
+            # Has at least one nested SKILL.md?
+            try:
+                if any(entry.rglob("SKILL.md")):
+                    out.append(entry.name)
+            except OSError:
+                continue
+    except (FileNotFoundError, OSError):
+        return []
+    return sorted(set(out))
+
+
+def _prompt_for_skill_name(c: Console, url: str, default: str = "") -> Optional[str]:
+    """Prompt interactively for a skill name. Returns None on cancel/EOF."""
+    c.print()
+    c.print(
+        f"[yellow]The SKILL.md at {url} doesn't declare a `name:` in its "
+        f"frontmatter,[/]\n[yellow]and the URL path doesn't produce a valid "
+        f"identifier either.[/]"
+    )
+    default_hint = f" [{default}]" if default else ""
+    c.print(
+        f"[bold]Enter a skill name{default_hint}:[/] "
+        f"[dim](lowercase letters, digits, hyphens, underscores; starts with a letter)[/]"
+    )
+    try:
+        answer = input("Name: ").strip()
+    except (EOFError, KeyboardInterrupt):
+        return None
+    if not answer and default:
+        answer = default
+    if not _is_valid_installed_skill_name(answer):
+        c.print(f"[bold red]Invalid name:[/] {answer!r}. Aborting install.\n")
+        return None
+    return answer
+
+
+def _prompt_for_category(c: Console, existing: List[str]) -> str:
+    """Prompt interactively for a category. Empty/None input means flat install."""
+    c.print()
+    if existing:
+        c.print(
+            "[bold]Pick a category[/] "
+            "[dim](reuse an existing bucket, type a new one, or press Enter to install flat)[/]"
+        )
+        c.print(f"[dim]Existing: {', '.join(existing)}[/]")
+    else:
+        c.print(
+            "[bold]Category[/] [dim](optional — press Enter to install flat at ~/.hermes/skills/<name>/)[/]"
+        )
+    try:
+        answer = input("Category: ").strip()
+    except (EOFError, KeyboardInterrupt):
+        return ""
+    if not answer:
+        return ""
+    if not _VALID_CATEGORY_RE.match(answer):
+        c.print(f"[dim]Invalid category {answer!r} — installing flat.[/]")
+        return ""
+    return answer
+
+
 def do_search(query: str, source: str = "all", limit: int = 10,
               console: Optional[Console] = None) -> None:
     """Search registries and display results as a Rich table."""
@@ -309,8 +407,17 @@ def do_browse(page: int = 1, page_size: int = 20, source: str = "all",
 
 def do_install(identifier: str, category: str = "", force: bool = False,
                console: Optional[Console] = None, skip_confirm: bool = False,
-               invalidate_cache: bool = True) -> None:
-    """Fetch, quarantine, scan, confirm, and install a skill."""
+               invalidate_cache: bool = True,
+               name_override: str = "") -> None:
+    """Fetch, quarantine, scan, confirm, and install a skill.
+
+    ``name_override`` lets non-interactive callers (slash commands, gateway,
+    scripts) supply a skill name when the upstream SKILL.md lacks a valid
+    ``name:`` frontmatter field. On interactive TTY surfaces, a missing name
+    triggers a prompt instead; ``skip_confirm=True`` means "non-interactive"
+    (so pair it with ``name_override`` when installing from a URL that has
+    no frontmatter).
+    """
     from tools.skills_hub import (
         GitHubAuth, create_source_router, ensure_hub_dirs,
         quarantine_bundle, install_from_quarantine, HubLockFile,
@@ -354,6 +461,58 @@ def do_install(identifier: str, category: str = "", force: bool = False,
             c.print()
         return
 
+    # URL-sourced skills may arrive with an empty name when SKILL.md has no
+    # ``name:`` in frontmatter AND the URL path doesn't yield a valid
+    # identifier. Resolve by (1) --name override, (2) interactive prompt on
+    # a TTY, (3) refuse with an actionable error on non-interactive surfaces.
+    bundle_meta = getattr(bundle, "metadata", {}) or {}
+    if bundle.source == "url" and (not bundle.name or bundle_meta.get("awaiting_name")):
+        if name_override and _is_valid_installed_skill_name(name_override):
+            bundle.name = name_override.strip()
+            bundle_meta["awaiting_name"] = False
+        elif name_override:
+            c.print(
+                f"[bold red]Invalid --name:[/] {name_override!r}. "
+                "Must be a lowercase identifier (letters, digits, hyphens, "
+                "underscores; starts with a letter).\n"
+            )
+            return
+        elif skip_confirm:
+            # Non-interactive surface (slash command / TUI / gateway). Can't
+            # prompt — emit an actionable error.
+            url = bundle_meta.get("url") or identifier
+            c.print(
+                f"[bold red]Cannot install from URL:[/] {url}\n"
+                "[yellow]The SKILL.md has no `name:` in its frontmatter, "
+                "and the URL path doesn't produce a valid identifier.[/]\n\n"
+                "Retry with an explicit name:\n"
+                f"  [bold]/skills install {url} --name <your-name>[/]\n"
+                f"  [bold]hermes skills install {url} --name <your-name>[/]\n\n"
+                "[dim]Or ask the SKILL.md's author to add a `name:` field to "
+                "its YAML frontmatter.[/]\n"
+            )
+            return
+        else:
+            # Interactive TTY — prompt.
+            url = bundle_meta.get("url") or identifier
+            chosen = _prompt_for_skill_name(c, url)
+            if not chosen:
+                c.print("[dim]Installation cancelled.[/]\n")
+                return
+            bundle.name = chosen
+            bundle_meta["awaiting_name"] = False
+        # Keep SkillMeta in sync so downstream "already installed" checks,
+        # audit logs, and display all see the final name.
+        if meta is not None:
+            meta.name = bundle.name
+            meta.path = bundle.name
+
+    # URL-sourced skills: offer to pick a category interactively when the
+    # caller didn't specify one (TTY only — non-interactive installs fall
+    # through to flat install, matching all other sources).
+    if bundle.source == "url" and not category and not skip_confirm:
+        category = _prompt_for_category(c, _existing_categories())
+
     # Auto-detect category for official skills (e.g. "official/autonomous-ai-agents/blackbox")
     if bundle.source == "official" and not category:
         id_parts = bundle.identifier.split("/")  # ["official", "category", "skill"]
@@ -599,11 +758,24 @@ def inspect_skill(identifier: str) -> Optional[dict]:
     return out
 
 
-def do_list(source_filter: str = "all", console: Optional[Console] = None) -> None:
-    """List installed skills, distinguishing hub, builtin, and local skills."""
+def do_list(source_filter: str = "all",
+            enabled_only: bool = False,
+            console: Optional[Console] = None) -> None:
+    """List installed skills, distinguishing hub, builtin, and local skills.
+
+    Args:
+        source_filter: ``all`` | ``hub`` | ``builtin`` | ``local``.
+        enabled_only: If True, hide disabled skills from the output.
+
+    Enabled/disabled state is resolved against the currently active profile's
+    config — ``hermes -p <profile> skills list`` reads that profile's
+    ``skills.disabled`` list because ``-p`` swaps ``HERMES_HOME`` at process
+    start.  No explicit profile flag needed here.
+    """
     from tools.skills_hub import HubLockFile, ensure_hub_dirs
     from tools.skills_sync import _read_manifest
     from tools.skills_tool import _find_all_skills
+    from agent.skill_utils import get_disabled_skill_names
 
     c = console or _console
     ensure_hub_dirs()
@@ -611,17 +783,26 @@ def do_list(source_filter: str = "all", console: Optional[Console] = None) -> No
     hub_installed = {e["name"]: e for e in lock.list_installed()}
     builtin_names = set(_read_manifest())
 
-    all_skills = _find_all_skills()
+    # Pull ALL skills (including disabled ones) so we can annotate status.
+    all_skills = _find_all_skills(skip_disabled=True)
+    disabled_names = get_disabled_skill_names()
 
-    table = Table(title="Installed Skills")
+    title = "Installed Skills"
+    if enabled_only:
+        title += " (enabled only)"
+
+    table = Table(title=title)
     table.add_column("Name", style="bold cyan")
     table.add_column("Category", style="dim")
     table.add_column("Source", style="dim")
     table.add_column("Trust", style="dim")
+    table.add_column("Status", style="dim")
 
     hub_count = 0
     builtin_count = 0
     local_count = 0
+    enabled_count = 0
+    disabled_count = 0
 
     for skill in sorted(all_skills, key=lambda s: (s.get("category") or "", s["name"])):
         name = skill["name"]
@@ -632,29 +813,48 @@ def do_list(source_filter: str = "all", console: Optional[Console] = None) -> No
             source_type = "hub"
             source_display = hub_entry.get("source", "hub")
             trust = hub_entry.get("trust_level", "community")
-            hub_count += 1
         elif name in builtin_names:
             source_type = "builtin"
             source_display = "builtin"
             trust = "builtin"
-            builtin_count += 1
         else:
             source_type = "local"
             source_display = "local"
             trust = "local"
-            local_count += 1
 
         if source_filter != "all" and source_filter != source_type:
             continue
 
+        is_enabled = name not in disabled_names
+        if enabled_only and not is_enabled:
+            continue
+
+        if source_type == "hub":
+            hub_count += 1
+        elif source_type == "builtin":
+            builtin_count += 1
+        else:
+            local_count += 1
+
+        if is_enabled:
+            enabled_count += 1
+            status_cell = "[bold green]enabled[/]"
+        else:
+            disabled_count += 1
+            status_cell = "[dim red]disabled[/]"
+
         trust_style = {"builtin": "bright_cyan", "trusted": "green", "community": "yellow", "local": "dim"}.get(trust, "dim")
         trust_label = "official" if source_display == "official" else trust
-        table.add_row(name, category, source_display, f"[{trust_style}]{trust_label}[/]")
+        table.add_row(name, category, source_display, f"[{trust_style}]{trust_label}[/]", status_cell)
 
     c.print(table)
-    c.print(
-        f"[dim]{hub_count} hub-installed, {builtin_count} builtin, {local_count} local[/]\n"
-    )
+    summary = f"[dim]{hub_count} hub-installed, {builtin_count} builtin, {local_count} local"
+    if enabled_only:
+        summary += f" — {enabled_count} enabled shown"
+    else:
+        summary += f" — {enabled_count} enabled, {disabled_count} disabled"
+    summary += "[/]\n"
+    c.print(summary)
 
 
 def do_check(name: Optional[str] = None, console: Optional[Console] = None) -> None:
@@ -1123,11 +1323,15 @@ def skills_command(args) -> None:
         do_search(args.query, source=args.source, limit=args.limit)
     elif action == "install":
         do_install(args.identifier, category=args.category, force=args.force,
-                   skip_confirm=getattr(args, "yes", False))
+                   skip_confirm=getattr(args, "yes", False),
+                   name_override=getattr(args, "name", "") or "")
     elif action == "inspect":
         do_inspect(args.identifier)
     elif action == "list":
-        do_list(source_filter=args.source)
+        do_list(
+            source_filter=args.source,
+            enabled_only=getattr(args, "enabled_only", False),
+        )
     elif action == "check":
         do_check(name=getattr(args, "name", None))
     elif action == "update":
@@ -1177,6 +1381,7 @@ def handle_skills_slash(cmd: str, console: Optional[Console] = None) -> None:
         /skills search kubernetes
         /skills install openai/skills/skill-creator
         /skills install openai/skills/skill-creator --force
+        /skills install https://example.com/path/SKILL.md
         /skills inspect openai/skills/skill-creator
         /skills list
         /skills list --source hub
@@ -1253,10 +1458,11 @@ def handle_skills_slash(cmd: str, console: Optional[Console] = None) -> None:
 
     elif action == "install":
         if not args:
-            c.print("[bold red]Usage:[/] /skills install <identifier> [--category <cat>] [--force] [--now]\n")
+            c.print("[bold red]Usage:[/] /skills install <identifier-or-url> [--name <name>] [--category <cat>] [--force] [--now]\n")
             return
         identifier = args[0]
         category = ""
+        name_override = ""
         # Slash commands run inside prompt_toolkit where input() hangs.
         # Always skip confirmation — the user typing the command is implicit consent.
         skip_confirm = True
@@ -1267,9 +1473,11 @@ def handle_skills_slash(cmd: str, console: Optional[Console] = None) -> None:
         for i, a in enumerate(args):
             if a == "--category" and i + 1 < len(args):
                 category = args[i + 1]
+            elif a == "--name" and i + 1 < len(args):
+                name_override = args[i + 1]
         do_install(identifier, category=category, force=force,
                    skip_confirm=skip_confirm, invalidate_cache=invalidate_cache,
-                   console=c)
+                   name_override=name_override, console=c)
 
     elif action == "inspect":
         if not args:
@@ -1279,11 +1487,12 @@ def handle_skills_slash(cmd: str, console: Optional[Console] = None) -> None:
 
     elif action == "list":
         source_filter = "all"
+        enabled_only = "--enabled-only" in args or "--enabled" in args
         if "--source" in args:
             idx = args.index("--source")
             if idx + 1 < len(args):
                 source_filter = args[idx + 1]
-        do_list(source_filter=source_filter, console=c)
+        do_list(source_filter=source_filter, enabled_only=enabled_only, console=c)
 
     elif action == "check":
         name = args[0] if args else None
@@ -1371,7 +1580,8 @@ def _print_skills_help(console: Console) -> None:
         "  [cyan]search[/] <query>              Search registries for skills\n"
         "  [cyan]install[/] <identifier>        Install a skill (with security scan)\n"
         "  [cyan]inspect[/] <identifier>        Preview a skill without installing\n"
-        "  [cyan]list[/] [--source hub|builtin|local] List installed skills\n"
+        "  [cyan]list[/] [--source hub|builtin|local] [--enabled-only]\n"
+        "       List installed skills; --enabled-only filters to the active profile's live set\n"
         "  [cyan]check[/] [name]                Check hub skills for upstream updates\n"
         "  [cyan]update[/] [name]               Update hub skills with upstream changes\n"
         "  [cyan]audit[/] [name]                Re-scan hub skills for security\n"

hermes_cli/slack_cli.py

@@ -0,0 +1,152 @@
+"""``hermes slack ...`` CLI subcommands.
+
+Today only ``hermes slack manifest`` is implemented — it generates the
+Slack app manifest JSON for registering every gateway command as a native
+Slack slash (``/btw``, ``/stop``, ``/model``, …) so users get the same
+first-class slash UX Discord and Telegram already have.
+
+Typical workflow::
+
+    $ hermes slack manifest > slack-manifest.json
+    # or:
+    $ hermes slack manifest --write
+
+Then paste the printed JSON into the Slack app config (Features → App
+Manifest → Edit) and click Save. Slack diffs the manifest and prompts
+for reinstall when scopes/commands change.
+"""
+from __future__ import annotations
+
+import json
+import sys
+from pathlib import Path
+
+
+def _build_full_manifest(bot_name: str, bot_description: str) -> dict:
+    """Build a full Slack manifest merging display info + our slash list.
+
+    The slash-command list is always generated from ``COMMAND_REGISTRY`` so
+    it stays in sync with the rest of Hermes. Other manifest sections
+    (display info, OAuth scopes, socket mode) are set to sensible defaults
+    for a Hermes deployment — users can tweak them in the Slack UI after
+    pasting.
+    """
+    from hermes_cli.commands import slack_app_manifest
+
+    partial = slack_app_manifest()
+    slashes = partial["features"]["slash_commands"]
+
+    return {
+        "_metadata": {
+            "major_version": 1,
+            "minor_version": 1,
+        },
+        "display_information": {
+            "name": bot_name[:35],
+            "description": (bot_description or "Your Hermes agent on Slack")[:140],
+            "background_color": "#1a1a2e",
+        },
+        "features": {
+            "bot_user": {
+                "display_name": bot_name[:80],
+                "always_online": True,
+            },
+            "slash_commands": slashes,
+            "assistant_view": {
+                "assistant_description": "Chat with Hermes in threads and DMs.",
+            },
+        },
+        "oauth_config": {
+            "scopes": {
+                "bot": [
+                    "app_mentions:read",
+                    "assistant:write",
+                    "channels:history",
+                    "channels:read",
+                    "chat:write",
+                    "commands",
+                    "files:read",
+                    "files:write",
+                    "groups:history",
+                    "im:history",
+                    "im:read",
+                    "im:write",
+                    "users:read",
+                ],
+            },
+        },
+        "settings": {
+            "event_subscriptions": {
+                "bot_events": [
+                    "app_mention",
+                    "assistant_thread_context_changed",
+                    "assistant_thread_started",
+                    "message.channels",
+                    "message.groups",
+                    "message.im",
+                ],
+            },
+            "interactivity": {
+                "is_enabled": True,
+            },
+            "org_deploy_enabled": False,
+            "socket_mode_enabled": True,
+            "token_rotation_enabled": False,
+        },
+    }
+
+
+def slack_manifest_command(args) -> int:
+    """Print or write a Slack app manifest JSON.
+
+    Flags (all parsed in ``hermes_cli/main.py``):
+      --write [PATH]  Write to file instead of stdout (default path:
+                      ``$HERMES_HOME/slack-manifest.json``)
+      --name NAME     Override the bot display name (default: "Hermes")
+      --description DESC  Override the bot description
+      --slashes-only  Emit only the ``features.slash_commands`` array (for
+                      merging into an existing manifest manually)
+    """
+    name = getattr(args, "name", None) or "Hermes"
+    description = getattr(args, "description", None) or "Your Hermes agent on Slack"
+
+    if getattr(args, "slashes_only", False):
+        from hermes_cli.commands import slack_app_manifest
+
+        manifest = slack_app_manifest()["features"]["slash_commands"]
+    else:
+        manifest = _build_full_manifest(name, description)
+
+    payload = json.dumps(manifest, indent=2, ensure_ascii=False) + "\n"
+
+    write_target = getattr(args, "write", None)
+    if write_target is not None:
+        if isinstance(write_target, bool) and write_target:
+            # --write with no value → default location
+            try:
+                from hermes_constants import get_hermes_home
+
+                target = Path(get_hermes_home()) / "slack-manifest.json"
+            except Exception:
+                target = Path.home() / ".hermes" / "slack-manifest.json"
+        else:
+            target = Path(write_target).expanduser()
+        target.parent.mkdir(parents=True, exist_ok=True)
+        target.write_text(payload, encoding="utf-8")
+        print(f"Slack manifest written to: {target}", file=sys.stderr)
+        print(
+            "\nNext steps:\n"
+            "  1. Open https://api.slack.com/apps and pick your Hermes app\n"
+            "     (or create a new one: Create New App → From an app manifest).\n"
+            f"  2. Features → App Manifest → paste the contents of\n"
+            f"     {target}\n"
+            "  3. Save; Slack will prompt to reinstall the app if scopes or\n"
+            "     slash commands changed.\n"
+            "  4. Make sure Socket Mode is enabled and you have a bot token\n"
+            "     (xoxb-...) and app token (xapp-...) configured via\n"
+            "     `hermes setup`.\n",
+            file=sys.stderr,
+        )
+    else:
+        sys.stdout.write(payload)
+    return 0

hermes_cli/status.py

@@ -326,7 +326,8 @@ def show_status(args):
         "WeCom Callback": ("WECOM_CALLBACK_CORP_ID", None),
         "Weixin": ("WEIXIN_ACCOUNT_ID", "WEIXIN_HOME_CHANNEL"),
         "BlueBubbles": ("BLUEBUBBLES_SERVER_URL", "BLUEBUBBLES_HOME_CHANNEL"),
-        "QQBot": ("QQ_APP_ID", "QQBOT_HOME_CHANNEL"),
+        "QQBot": ("QQ_APP_ID", "QQ_HOME_CHANNEL"),
+        "Yuanbao": ("YUANBAO_APP_ID", "YUANBAO_HOME_CHANNEL"),
     }
     
     for name, (token_var, home_var) in platforms.items():

hermes_cli/timeouts.py

@@ -20,10 +20,10 @@ def get_provider_request_timeout(
 
     try:
         from hermes_cli.config import load_config
-    except ImportError:
+        config = load_config()
+    except Exception:
         return None
 
-    config = load_config()
     providers = config.get("providers", {}) if isinstance(config, dict) else {}
     provider_config = (
         providers.get(provider_id, {}) if isinstance(providers, dict) else {}
@@ -49,10 +49,10 @@ def get_provider_stale_timeout(
 
     try:
         from hermes_cli.config import load_config
-    except ImportError:
+        config = load_config()
+    except Exception:
         return None
 
-    config = load_config()
     providers = config.get("providers", {}) if isinstance(config, dict) else {}
     provider_config = (
         providers.get(provider_id, {}) if isinstance(providers, dict) else {}

hermes_cli/tips.py

@@ -106,7 +106,7 @@ TIPS = [
     "Set display.streaming: true to see tokens appear in real time as the model generates.",
     "Set display.show_reasoning: true to watch the model's chain-of-thought reasoning.",
     "Set display.compact: true to reduce whitespace in output for denser information.",
-    "Set display.busy_input_mode: queue to queue messages instead of interrupting the agent.",
+    "Set display.busy_input_mode: queue to queue messages instead of interrupting the agent, or steer to inject them mid-run via /steer.",
     "Set display.resume_display: minimal to skip the full conversation recap on session resume.",
     "Set compression.threshold: 0.50 to control when auto-compression fires (default: 50% of context).",
     "Set agent.max_turns: 200 to let the agent take more tool-calling steps per turn.",

hermes_cli/tools_config.py

@@ -11,6 +11,7 @@ the `platform_toolsets` key.
 
 import json as _json
 import logging
+import os
 import sys
 from pathlib import Path
 from typing import Dict, List, Optional, Set
@@ -25,7 +26,7 @@ from hermes_cli.nous_subscription import (
     get_nous_subscription_features,
 )
 from tools.tool_backend_helpers import fal_key_is_configured, managed_nous_tools_enabled
-from utils import base_url_hostname
+from utils import base_url_hostname, is_truthy_value
 
 logger = logging.getLogger(__name__)
 
@@ -70,6 +71,7 @@ CONFIGURABLE_TOOLSETS = [
     ("spotify",          "🎵 Spotify",                  "playback, search, playlists, library"),
     ("discord",         "💬 Discord (read/participate)", "fetch messages, search members, create thread"),
     ("discord_admin",   "🛡️  Discord Server Admin",    "list channels/roles, pin, assign roles"),
+    ("yuanbao",          "🤖 Yuanbao",                  "group info, member queries, DM"),
 ]
 
 # Toolsets that are OFF by default for new installs.
@@ -676,6 +678,15 @@ def _get_platform_tools(
         # their own platform (e.g. `discord` + `discord` should stay OFF).
         if platform in default_off and platform not in _TOOLSET_PLATFORM_RESTRICTIONS:
             default_off.remove(platform)
+        # Home Assistant is already runtime-gated by its check_fn (requires
+        # HASS_TOKEN to register any tools). When a user has configured
+        # HASS_TOKEN, they've explicitly opted in — don't also strip it via
+        # _DEFAULT_OFF_TOOLSETS, which would silently drop HA from platforms
+        # (e.g. cron) that run through _get_platform_tools without an
+        # explicit saved toolset list. Without this, Norbert's HA cron jobs
+        # regressed after #14798 made cron honor per-platform tool config.
+        if "homeassistant" in default_off and os.getenv("HASS_TOKEN"):
+            default_off.remove("homeassistant")
         enabled_toolsets -= default_off
 
     # Recover non-configurable platform toolsets (e.g. discord, feishu_doc,
@@ -1177,7 +1188,7 @@ def _is_provider_active(provider: dict, config: dict) -> bool:
                 configured_provider = image_cfg.get("provider")
                 if configured_provider not in (None, "", "fal"):
                     return False
-                if image_cfg.get("use_gateway") is False:
+                if image_cfg.get("use_gateway") is not None and not is_truthy_value(image_cfg.get("use_gateway"), default=False):
                     return False
             return feature.managed_by_nous
         if provider.get("tts_provider"):
@@ -1209,7 +1220,7 @@ def _is_provider_active(provider: dict, config: dict) -> bool:
         return (
             provider["imagegen_backend"] == "fal"
             and configured_provider in (None, "", "fal")
-            and not image_cfg.get("use_gateway")
+            and not is_truthy_value(image_cfg.get("use_gateway"), default=False)
         )
     return False
 

hermes_cli/web_server.py

@@ -287,7 +287,7 @@ _SCHEMA_OVERRIDES: Dict[str, Dict[str, Any]] = {
     "display.busy_input_mode": {
         "type": "select",
         "description": "Input behavior while agent is running",
-        "options": ["interrupt", "queue"],
+        "options": ["interrupt", "queue", "steer"],
     },
     "memory.provider": {
         "type": "select",
@@ -2327,16 +2327,14 @@ def _resolve_chat_argv(
     from hermes_cli.main import PROJECT_ROOT, _make_tui_argv
 
     argv, cwd = _make_tui_argv(PROJECT_ROOT / "ui-tui", tui_dev=False)
-    env: Optional[dict] = None
+    env = os.environ.copy()
+    env.setdefault("NODE_ENV", "production")
 
-    if resume or sidecar_url:
-        env = os.environ.copy()
+    if resume:
+        env["HERMES_TUI_RESUME"] = resume
 
-        if resume:
-            env["HERMES_TUI_RESUME"] = resume
-
-        if sidecar_url:
-            env["HERMES_TUI_SIDECAR_URL"] = sidecar_url
+    if sidecar_url:
+        env["HERMES_TUI_SIDECAR_URL"] = sidecar_url
 
     return list(argv), str(cwd) if cwd else None, env
 

hermes_logging.py

@@ -195,10 +195,6 @@ def setup_logging(
         The ``logs/`` directory where files are written.
     """
     global _logging_initialized
-    if _logging_initialized and not force:
-        home = hermes_home or get_hermes_home()
-        return home / "logs"
-
     home = hermes_home or get_hermes_home()
     log_dir = home / "logs"
     log_dir.mkdir(parents=True, exist_ok=True)
@@ -248,6 +244,9 @@ def setup_logging(
             log_filter=_ComponentFilter(COMPONENT_PREFIXES["gateway"]),
         )
 
+    if _logging_initialized and not force:
+        return log_dir
+
     # Ensure root logger level is low enough for the handlers to fire.
     if root.level == logging.NOTSET or root.level > level:
         root.setLevel(level)

hermes_state.py

@@ -832,7 +832,18 @@ class SessionDB:
         params = []
 
         if not include_children:
-            where_clauses.append("s.parent_session_id IS NULL")
+            # Show root sessions and branch sessions (whose parent ended with
+            # end_reason='branched' before the child was created), while still
+            # hiding sub-agent runs and compression continuations (which also
+            # carry a parent_session_id but were spawned while the parent was
+            # still live — i.e., started_at < parent.ended_at).
+            where_clauses.append(
+                "(s.parent_session_id IS NULL"
+                " OR EXISTS (SELECT 1 FROM sessions p"
+                "            WHERE p.id = s.parent_session_id"
+                "            AND p.end_reason = 'branched'"
+                "            AND s.started_at >= p.ended_at))"
+            )
 
         if source:
             where_clauses.append("s.source = ?")
@@ -1121,20 +1132,27 @@ class SessionDB:
                 current = child_id
         return session_id
 
-    def get_messages_as_conversation(self, session_id: str) -> List[Dict[str, Any]]:
+    def get_messages_as_conversation(
+        self, session_id: str, include_ancestors: bool = False
+    ) -> List[Dict[str, Any]]:
         """
         Load messages in the OpenAI conversation format (role + content dicts).
         Used by the gateway to restore conversation history.
         """
+        session_ids = [session_id]
+        if include_ancestors:
+            session_ids = self._session_lineage_root_to_tip(session_id)
+
         with self._lock:
-            cursor = self._conn.execute(
+            placeholders = ",".join("?" for _ in session_ids)
+            rows = self._conn.execute(
                 "SELECT role, content, tool_call_id, tool_calls, tool_name, "
                 "reasoning, reasoning_content, reasoning_details, codex_reasoning_items, "
                 "codex_message_items "
-                "FROM messages WHERE session_id = ? ORDER BY timestamp, id",
-                (session_id,),
-            )
-            rows = cursor.fetchall()
+                f"FROM messages WHERE session_id IN ({placeholders}) ORDER BY timestamp, id",
+                tuple(session_ids),
+            ).fetchall()
+
         messages = []
         for row in rows:
             msg = {"role": row["role"], "content": row["content"]}
@@ -1174,9 +1192,47 @@ class SessionDB:
                     except (json.JSONDecodeError, TypeError):
                         logger.warning("Failed to deserialize codex_message_items, falling back to None")
                         msg["codex_message_items"] = None
+            if include_ancestors and self._is_duplicate_replayed_user_message(messages, msg):
+                continue
             messages.append(msg)
         return messages
 
+    def _session_lineage_root_to_tip(self, session_id: str) -> List[str]:
+        if not session_id:
+            return [session_id]
+
+        chain = []
+        current = session_id
+        seen = set()
+        with self._lock:
+            for _ in range(100):
+                if not current or current in seen:
+                    break
+                seen.add(current)
+                chain.append(current)
+                row = self._conn.execute(
+                    "SELECT parent_session_id FROM sessions WHERE id = ?",
+                    (current,),
+                ).fetchone()
+                if row is None:
+                    break
+                current = row["parent_session_id"] if hasattr(row, "keys") else row[0]
+        return list(reversed(chain)) or [session_id]
+
+    @staticmethod
+    def _is_duplicate_replayed_user_message(messages: List[Dict[str, Any]], msg: Dict[str, Any]) -> bool:
+        if msg.get("role") != "user":
+            return False
+        content = msg.get("content")
+        if not isinstance(content, str) or not content:
+            return False
+        for prev in reversed(messages):
+            if prev.get("role") == "user" and prev.get("content") == content:
+                return True
+            if prev.get("role") == "assistant" and (prev.get("content") or prev.get("tool_calls")):
+                return False
+        return False
+
     # =========================================================================
     # Search
     # =========================================================================
@@ -1501,12 +1557,45 @@ class SessionDB:
             )
         self._execute_write(_do)
 
-    def delete_session(self, session_id: str) -> bool:
+    @staticmethod
+    def _remove_session_files(sessions_dir: Optional[Path], session_id: str) -> None:
+        """Remove on-disk transcript files for a session.
+
+        Cleans up ``{session_id}.json``, ``{session_id}.jsonl``, and any
+        ``request_dump_{session_id}_*.json`` files left by the gateway.
+        Silently skips files that don't exist and swallows OSError so a
+        filesystem hiccup never blocks a DB operation.
+        """
+        if sessions_dir is None:
+            return
+        for suffix in (".json", ".jsonl"):
+            p = sessions_dir / f"{session_id}{suffix}"
+            try:
+                p.unlink(missing_ok=True)
+            except OSError:
+                pass
+        # request_dump files use session_id as a prefix component
+        try:
+            for p in sessions_dir.glob(f"request_dump_{session_id}_*.json"):
+                try:
+                    p.unlink(missing_ok=True)
+                except OSError:
+                    pass
+        except OSError:
+            pass
+
+    def delete_session(
+        self,
+        session_id: str,
+        sessions_dir: Optional[Path] = None,
+    ) -> bool:
         """Delete a session and all its messages.
 
         Child sessions are orphaned (parent_session_id set to NULL) rather
         than cascade-deleted, so they remain accessible independently.
-        Returns True if the session was found and deleted.
+        When *sessions_dir* is provided, also removes on-disk transcript
+        files (``.json`` / ``.jsonl`` / ``request_dump_*``) for the deleted
+        session. Returns True if the session was found and deleted.
         """
         def _do(conn):
             cursor = conn.execute(
@@ -1523,16 +1612,29 @@ class SessionDB:
             conn.execute("DELETE FROM messages WHERE session_id = ?", (session_id,))
             conn.execute("DELETE FROM sessions WHERE id = ?", (session_id,))
             return True
-        return self._execute_write(_do)
 
-    def prune_sessions(self, older_than_days: int = 90, source: str = None) -> int:
+        deleted = self._execute_write(_do)
+        if deleted:
+            self._remove_session_files(sessions_dir, session_id)
+        return deleted
+
+    def prune_sessions(
+        self,
+        older_than_days: int = 90,
+        source: str = None,
+        sessions_dir: Optional[Path] = None,
+    ) -> int:
         """Delete sessions older than N days. Returns count of deleted sessions.
 
         Only prunes ended sessions (not active ones).  Child sessions outside
         the prune window are orphaned (parent_session_id set to NULL) rather
-        than cascade-deleted.
+        than cascade-deleted.  When *sessions_dir* is provided, also removes
+        on-disk transcript files (``.json`` / ``.jsonl`` /
+        ``request_dump_*``) for every pruned session, outside the DB
+        transaction.
         """
         cutoff = time.time() - (older_than_days * 86400)
+        removed_ids: list[str] = []
 
         def _do(conn):
             if source:
@@ -1562,9 +1664,14 @@ class SessionDB:
             for sid in session_ids:
                 conn.execute("DELETE FROM messages WHERE session_id = ?", (sid,))
                 conn.execute("DELETE FROM sessions WHERE id = ?", (sid,))
+                removed_ids.append(sid)
             return len(session_ids)
 
-        return self._execute_write(_do)
+        count = self._execute_write(_do)
+        # Clean up on-disk files outside the DB transaction
+        for sid in removed_ids:
+            self._remove_session_files(sessions_dir, sid)
+        return count
 
     # ── Meta key/value (for scheduler bookkeeping) ──
 
@@ -1618,6 +1725,7 @@ class SessionDB:
         retention_days: int = 90,
         min_interval_hours: int = 24,
         vacuum: bool = True,
+        sessions_dir: Optional[Path] = None,
     ) -> Dict[str, Any]:
         """Idempotent auto-maintenance: prune old sessions + optional VACUUM.
 
@@ -1625,6 +1733,10 @@ class SessionDB:
         within ``min_interval_hours`` no-op. Designed to be called once at
         startup from long-lived entrypoints (CLI, gateway, cron scheduler).
 
+        When *sessions_dir* is provided, on-disk transcript files
+        (``.json`` / ``.jsonl`` / ``request_dump_*``) for pruned sessions
+        are removed as part of the same sweep (issue #3015).
+
         Never raises. On any failure, logs a warning and returns a dict
         with ``"error"`` set.
 
@@ -1648,7 +1760,10 @@ class SessionDB:
                 except (TypeError, ValueError):
                     pass  # corrupt meta; treat as no prior run
 
-            pruned = self.prune_sessions(older_than_days=retention_days)
+            pruned = self.prune_sessions(
+                older_than_days=retention_days,
+                sessions_dir=sessions_dir,
+            )
             result["pruned"] = pruned
 
             # Only VACUUM if we actually freed rows — VACUUM on a tight DB

nix/tui.nix

@@ -4,7 +4,7 @@ let
   src = ../ui-tui;
   npmDeps = pkgs.fetchNpmDeps {
     inherit src;
-    hash = "sha256-RU4qSHgJPMyfRSEJDzkG4+MReDZDc6QbTD2wisa5QE0=";
+    hash = "sha256-Chz+NW9NXqboXHOa6PKwf5bhAkkcFtKNhvKWwg2XSPc=";
   };
 
   npm = hermesNpmLib.mkNpmPassthru { folder = "ui-tui"; attr = "tui"; pname = "hermes-tui"; };

optional-skills/migration/openclaw-migration/scripts/openclaw_to_hermes.py

@@ -380,6 +380,10 @@ def backup_existing(path: Path, backup_root: Path) -> Optional[Path]:
 # Replace OpenClaw brand names with Hermes in migrated text so that
 # memory entries, user profiles, SOUL.md, and workspace instructions
 # read as self-referential to the new agent identity.
+#
+# Case-preserving: ``OpenClaw`` → ``Hermes`` (prose), but lowercase matches
+# like ``openclaw`` → ``hermes`` (so filesystem paths like ``~/.openclaw``
+# become ``~/.hermes`` — the real Hermes home — not the broken ``~/.Hermes``).
 _REBRAND_PATTERNS: List[Tuple[re.Pattern, str]] = [
     (re.compile(r'\bOpen[\s-]?Claw\b', re.IGNORECASE), 'Hermes'),
     (re.compile(r'\bClawdBot\b', re.IGNORECASE), 'Hermes'),
@@ -387,10 +391,31 @@ _REBRAND_PATTERNS: List[Tuple[re.Pattern, str]] = [
 ]
 
 
+def _case_preserving_replacement(replacement: str):
+    """Return a re.sub replacement fn that lowercases the result when the
+    matched text was all-lowercase.
+
+    Keeps ``OpenClaw`` → ``Hermes`` but maps ``openclaw`` → ``hermes`` so a
+    filesystem path like ``~/.openclaw/config.yaml`` rewrites to
+    ``~/.hermes/config.yaml`` (the real Hermes home) instead of the broken
+    ``~/.Hermes/config.yaml``.
+    """
+    def _sub(match: "re.Match[str]") -> str:
+        matched = match.group(0)
+        if matched and matched.islower():
+            return replacement.lower()
+        return replacement
+    return _sub
+
+
 def rebrand_text(text: str) -> str:
-    """Replace OpenClaw / ClawdBot / MoltBot brand names with Hermes."""
+    """Replace OpenClaw / ClawdBot / MoltBot brand names with Hermes.
+
+    Preserves case so filesystem-path matches (lowercase) don't become
+    capitalized directory names that don't exist.
+    """
     for pattern, replacement in _REBRAND_PATTERNS:
-        text = pattern.sub(replacement, text)
+        text = pattern.sub(_case_preserving_replacement(replacement), text)
     return text
 
 

plugins/memory/hindsight/__init__.py

@@ -3,7 +3,9 @@
 Long-term memory with knowledge graph, entity resolution, and multi-strategy
 retrieval. Supports cloud (API key) and local modes.
 
-Configurable timeout via HINDSIGHT_TIMEOUT env var or config.json.
+Configurable request timeout via HINDSIGHT_TIMEOUT env var or config.json.
+Configurable embedded daemon idle timeout via HINDSIGHT_IDLE_TIMEOUT env var
+or config.json idle_timeout.
 
 Original PR #1811 by benfrank241, adapted to MemoryProvider ABC.
 
@@ -14,6 +16,7 @@ Config via environment variables:
   HINDSIGHT_API_URL                — API endpoint
   HINDSIGHT_MODE                   — cloud or local (default: cloud)
   HINDSIGHT_TIMEOUT                — API request timeout in seconds (default: 120)
+  HINDSIGHT_IDLE_TIMEOUT           — embedded daemon idle timeout seconds; 0 disables shutdown (default: 300)
   HINDSIGHT_RETAIN_TAGS            — comma-separated tags attached to retained memories
   HINDSIGHT_RETAIN_SOURCE          — metadata source value attached to retained memories
   HINDSIGHT_RETAIN_USER_PREFIX     — label used before user turns in retained transcripts
@@ -45,6 +48,7 @@ _DEFAULT_API_URL = "https://api.hindsight.vectorize.io"
 _DEFAULT_LOCAL_URL = "http://localhost:8888"
 _MIN_CLIENT_VERSION = "0.4.22"
 _DEFAULT_TIMEOUT = 120  # seconds — cloud API can take 30-40s per request
+_DEFAULT_IDLE_TIMEOUT = 300  # seconds — Hindsight embedded daemon default
 _VALID_BUDGETS = {"low", "mid", "high"}
 _PROVIDER_DEFAULT_MODELS = {
     "openai": "gpt-4o-mini",
@@ -59,6 +63,17 @@ _PROVIDER_DEFAULT_MODELS = {
 }
 
 
+def _parse_int_setting(value: Any, default: int) -> int:
+    """Parse an integer config/env value, falling back on invalid input."""
+    if value is None or value == "":
+        return default
+    try:
+        return int(value)
+    except (TypeError, ValueError):
+        logger.warning("Invalid integer Hindsight setting %r; using default %s", value, default)
+        return default
+
+
 def _check_local_runtime() -> tuple[bool, str | None]:
     """Return whether local embedded Hindsight imports cleanly.
 
@@ -203,6 +218,8 @@ def _load_config() -> dict:
     return {
         "mode": os.environ.get("HINDSIGHT_MODE", "cloud"),
         "apiKey": os.environ.get("HINDSIGHT_API_KEY", ""),
+        "timeout": _parse_int_setting(os.environ.get("HINDSIGHT_TIMEOUT"), _DEFAULT_TIMEOUT),
+        "idle_timeout": _parse_int_setting(os.environ.get("HINDSIGHT_IDLE_TIMEOUT"), _DEFAULT_IDLE_TIMEOUT),
         "retain_tags": os.environ.get("HINDSIGHT_RETAIN_TAGS", ""),
         "retain_source": os.environ.get("HINDSIGHT_RETAIN_SOURCE", ""),
         "retain_user_prefix": os.environ.get("HINDSIGHT_RETAIN_USER_PREFIX", "User"),
@@ -304,6 +321,16 @@ def _build_embedded_profile_env(config: dict[str, Any], *, llm_api_key: str | No
     }
     if current_base_url:
         env_values["HINDSIGHT_API_LLM_BASE_URL"] = str(current_base_url)
+
+    idle_timeout = (
+        config.get("idle_timeout")
+        if config.get("idle_timeout") is not None
+        else os.environ.get("HINDSIGHT_IDLE_TIMEOUT")
+    )
+    if idle_timeout is not None and idle_timeout != "":
+        env_values["HINDSIGHT_EMBED_DAEMON_IDLE_TIMEOUT"] = str(
+            _parse_int_setting(idle_timeout, _DEFAULT_IDLE_TIMEOUT)
+        )
     return env_values
 
 
@@ -412,6 +439,7 @@ class HindsightMemoryProvider(MemoryProvider):
         self._turn_index = 0
         self._client = None
         self._timeout = _DEFAULT_TIMEOUT
+        self._idle_timeout = _DEFAULT_IDLE_TIMEOUT
         self._prefetch_result = ""
         self._prefetch_lock = threading.Lock()
         self._prefetch_thread = None
@@ -592,10 +620,17 @@ class HindsightMemoryProvider(MemoryProvider):
             sys.stdout.write("  LLM API key: ")
             sys.stdout.flush()
             llm_key = getpass.getpass(prompt="") if sys.stdin.isatty() else sys.stdin.readline().strip()
-            # Always write explicitly (including empty) so the provider sees ""
-            # rather than a missing variable.  The daemon reads from .env at
-            # startup and fails when HINDSIGHT_LLM_API_KEY is unset.
-            env_writes["HINDSIGHT_LLM_API_KEY"] = llm_key
+            if llm_key:
+                env_writes["HINDSIGHT_LLM_API_KEY"] = llm_key
+            else:
+                env_path = Path(hermes_home) / ".env"
+                existing_llm_key = ""
+                if env_path.exists():
+                    for line in env_path.read_text().splitlines():
+                        if line.startswith("HINDSIGHT_LLM_API_KEY="):
+                            existing_llm_key = line.split("=", 1)[1]
+                            break
+                env_writes["HINDSIGHT_LLM_API_KEY"] = existing_llm_key
 
         # Step 4: Save everything
         provider_config["bank_id"] = "hermes"
@@ -605,6 +640,11 @@ class HindsightMemoryProvider(MemoryProvider):
         timeout_val = existing_timeout if existing_timeout else _DEFAULT_TIMEOUT
         provider_config["timeout"] = timeout_val
         env_writes["HINDSIGHT_TIMEOUT"] = str(timeout_val)
+        if mode == "local_embedded":
+            existing_idle_timeout = self._config.get("idle_timeout") if self._config else None
+            idle_timeout_val = existing_idle_timeout if existing_idle_timeout is not None else _DEFAULT_IDLE_TIMEOUT
+            provider_config["idle_timeout"] = idle_timeout_val
+            env_writes["HINDSIGHT_IDLE_TIMEOUT"] = str(idle_timeout_val)
         config["memory"]["provider"] = "hindsight"
         save_config(config)
 
@@ -693,6 +733,7 @@ class HindsightMemoryProvider(MemoryProvider):
             {"key": "recall_max_input_chars", "description": "Maximum input query length for auto-recall", "default": 800},
             {"key": "recall_prompt_preamble", "description": "Custom preamble for recalled memories in context"},
             {"key": "timeout", "description": "API request timeout in seconds", "default": _DEFAULT_TIMEOUT},
+            {"key": "idle_timeout", "description": "Embedded daemon idle timeout in seconds (0 disables auto-shutdown)", "default": _DEFAULT_IDLE_TIMEOUT, "when": {"mode": "local_embedded"}},
         ]
 
     def _get_client(self):
@@ -720,6 +761,14 @@ class HindsightMemoryProvider(MemoryProvider):
                 )
                 if self._llm_base_url:
                     kwargs["llm_base_url"] = self._llm_base_url
+                idle_timeout = _parse_int_setting(
+                    self._config.get("idle_timeout")
+                    if self._config.get("idle_timeout") is not None
+                    else os.environ.get("HINDSIGHT_IDLE_TIMEOUT", self._idle_timeout),
+                    _DEFAULT_IDLE_TIMEOUT,
+                )
+                self._idle_timeout = idle_timeout
+                kwargs["idle_timeout"] = idle_timeout
                 self._client = HindsightEmbedded(**kwargs)
             else:
                 from hindsight_client import Hindsight
@@ -736,6 +785,38 @@ class HindsightMemoryProvider(MemoryProvider):
         """Schedule *coro* on the shared loop using the configured timeout."""
         return _run_sync(coro, timeout=self._timeout)
 
+    def _is_retriable_embedded_connection_error(self, exc: Exception) -> bool:
+        """Return True for stale embedded-daemon connection failures."""
+        if self._mode != "local_embedded":
+            return False
+        text = f"{type(exc).__name__}: {exc}".lower()
+        return any(
+            marker in text
+            for marker in (
+                "cannot connect to host",
+                "connection refused",
+                "connect call failed",
+                "clientconnectorerror",
+            )
+        )
+
+    def _run_hindsight_operation(self, operation):
+        """Run an async Hindsight client operation, retrying once after idle shutdown."""
+        client = self._get_client()
+        try:
+            return self._run_sync(operation(client))
+        except Exception as exc:
+            if not self._is_retriable_embedded_connection_error(exc):
+                raise
+            logger.info(
+                "Hindsight embedded daemon appears unreachable; recreating client and retrying once: %s",
+                exc,
+            )
+            self._client = None
+            client = self._get_client()
+            self._client = client
+            return self._run_sync(operation(client))
+
     def initialize(self, session_id: str, **kwargs) -> None:
         self._session_id = str(session_id or "").strip()
         self._parent_session_id = str(kwargs.get("parent_session_id", "") or "").strip()
@@ -790,7 +871,14 @@ class HindsightMemoryProvider(MemoryProvider):
         self._session_turns = []
         self._mode = self._config.get("mode", "cloud")
         # Read timeout from config or env var, fall back to default
-        self._timeout = self._config.get("timeout") or int(os.environ.get("HINDSIGHT_TIMEOUT", str(_DEFAULT_TIMEOUT)))
+        self._timeout = _parse_int_setting(
+            self._config.get("timeout") if self._config.get("timeout") is not None else os.environ.get("HINDSIGHT_TIMEOUT"),
+            _DEFAULT_TIMEOUT,
+        )
+        self._idle_timeout = _parse_int_setting(
+            self._config.get("idle_timeout") if self._config.get("idle_timeout") is not None else os.environ.get("HINDSIGHT_IDLE_TIMEOUT"),
+            _DEFAULT_IDLE_TIMEOUT,
+        )
         # "local" is a legacy alias for "local_embedded"
         if self._mode == "local":
             self._mode = "local_embedded"
@@ -981,10 +1069,9 @@ class HindsightMemoryProvider(MemoryProvider):
 
         def _run():
             try:
-                client = self._get_client()
                 if self._prefetch_method == "reflect":
                     logger.debug("Prefetch: calling reflect (bank=%s, query_len=%d)", self._bank_id, len(query))
-                    resp = self._run_sync(client.areflect(bank_id=self._bank_id, query=query, budget=self._budget))
+                    resp = self._run_hindsight_operation(lambda client: client.areflect(bank_id=self._bank_id, query=query, budget=self._budget))
                     text = resp.text or ""
                 else:
                     recall_kwargs: dict = {
@@ -998,7 +1085,7 @@ class HindsightMemoryProvider(MemoryProvider):
                         recall_kwargs["types"] = self._recall_types
                     logger.debug("Prefetch: calling recall (bank=%s, query_len=%d, budget=%s)",
                                  self._bank_id, len(query), self._budget)
-                    resp = self._run_sync(client.arecall(**recall_kwargs))
+                    resp = self._run_hindsight_operation(lambda client: client.arecall(**recall_kwargs))
                     num_results = len(resp.results) if resp.results else 0
                     logger.debug("Prefetch: recall returned %d results", num_results)
                     text = "\n".join(f"- {r.text}" for r in resp.results if r.text) if resp.results else ""
@@ -1131,12 +1218,14 @@ class HindsightMemoryProvider(MemoryProvider):
                 item.pop("retain_async", None)
                 logger.debug("Hindsight retain: bank=%s, doc=%s, async=%s, content_len=%d, num_turns=%d",
                              self._bank_id, self._document_id, self._retain_async, len(content), len(self._session_turns))
-                self._run_sync(client.aretain_batch(
-                    bank_id=self._bank_id,
-                    items=[item],
-                    document_id=self._document_id,
-                    retain_async=self._retain_async,
-                ))
+                self._run_hindsight_operation(
+                    lambda client: client.aretain_batch(
+                        bank_id=self._bank_id,
+                        items=[item],
+                        document_id=self._document_id,
+                        retain_async=self._retain_async,
+                    )
+                )
                 logger.debug("Hindsight retain succeeded")
             except Exception as e:
                 logger.warning("Hindsight sync failed: %s", e, exc_info=True)
@@ -1152,12 +1241,6 @@ class HindsightMemoryProvider(MemoryProvider):
         return [RETAIN_SCHEMA, RECALL_SCHEMA, REFLECT_SCHEMA]
 
     def handle_tool_call(self, tool_name: str, args: dict, **kwargs) -> str:
-        try:
-            client = self._get_client()
-        except Exception as e:
-            logger.warning("Hindsight client init failed: %s", e)
-            return tool_error(f"Hindsight client unavailable: {e}")
-
         if tool_name == "hindsight_retain":
             content = args.get("content", "")
             if not content:
@@ -1171,7 +1254,7 @@ class HindsightMemoryProvider(MemoryProvider):
                 )
                 logger.debug("Tool hindsight_retain: bank=%s, content_len=%d, context=%s",
                              self._bank_id, len(content), context)
-                self._run_sync(client.aretain(**retain_kwargs))
+                self._run_hindsight_operation(lambda client: client.aretain(**retain_kwargs))
                 logger.debug("Tool hindsight_retain: success")
                 return json.dumps({"result": "Memory stored successfully."})
             except Exception as e:
@@ -1194,7 +1277,7 @@ class HindsightMemoryProvider(MemoryProvider):
                     recall_kwargs["types"] = self._recall_types
                 logger.debug("Tool hindsight_recall: bank=%s, query_len=%d, budget=%s",
                              self._bank_id, len(query), self._budget)
-                resp = self._run_sync(client.arecall(**recall_kwargs))
+                resp = self._run_hindsight_operation(lambda client: client.arecall(**recall_kwargs))
                 num_results = len(resp.results) if resp.results else 0
                 logger.debug("Tool hindsight_recall: %d results", num_results)
                 if not resp.results:
@@ -1212,9 +1295,11 @@ class HindsightMemoryProvider(MemoryProvider):
             try:
                 logger.debug("Tool hindsight_reflect: bank=%s, query_len=%d, budget=%s",
                              self._bank_id, len(query), self._budget)
-                resp = self._run_sync(client.areflect(
-                    bank_id=self._bank_id, query=query, budget=self._budget
-                ))
+                resp = self._run_hindsight_operation(
+                    lambda client: client.areflect(
+                        bank_id=self._bank_id, query=query, budget=self._budget
+                    )
+                )
                 logger.debug("Tool hindsight_reflect: response_len=%d", len(resp.text or ""))
                 return json.dumps({"result": resp.text or "No relevant memories found."})
             except Exception as e:
@@ -1231,9 +1316,19 @@ class HindsightMemoryProvider(MemoryProvider):
         if self._client is not None:
             try:
                 if self._mode == "local_embedded":
-                    # Use the public close() API. The RuntimeError from
-                    # aiohttp's "attached to a different loop" is expected
-                    # and harmless — the daemon keeps running independently.
+                    # HindsightEmbedded.close() delegates to its sync client.close().
+                    # When Hermes created/used that client on the shared async loop,
+                    # closing it from this thread can raise "attached to a different
+                    # loop" before aiohttp releases the session. Close the embedded
+                    # inner async client on the shared loop first, then let the
+                    # wrapper clean up daemon/UI bookkeeping.
+                    inner_client = getattr(self._client, "_client", None)
+                    if inner_client is not None and hasattr(inner_client, "aclose"):
+                        _run_sync(inner_client.aclose())
+                        try:
+                            self._client._client = None
+                        except Exception:
+                            pass
                     try:
                         self._client.close()
                     except RuntimeError:

run_agent.py

@@ -3304,10 +3304,19 @@ class AIAgent:
                 logger.warning("Background memory/skill review failed: %s", e)
                 self._emit_auxiliary_failure("background review", e)
             finally:
-                # Close all resources (httpx client, subprocesses, etc.) so
-                # GC doesn't try to clean them up on a dead asyncio event
-                # loop (which produces "Event loop is closed" errors).
+                # Background review agents can initialize memory providers
+                # (for example Hindsight) that own their own network clients.
+                # Explicitly stop those providers before closing the agent so
+                # their aiohttp sessions do not leak until GC/process exit.
+                # Then close all remaining resources (httpx client,
+                # subprocesses, etc.) so GC doesn't try to clean them up on a
+                # dead asyncio event loop (which produces "Event loop is
+                # closed" errors).
                 if review_agent is not None:
+                    try:
+                        review_agent.shutdown_memory_provider()
+                    except Exception:
+                        pass
                     try:
                         review_agent.close()
                     except Exception:
@@ -8151,6 +8160,22 @@ class AIAgent:
             except Exception as e:
                 logger.warning("Session DB compression split failed — new session will NOT be indexed: %s", e)
 
+        # Notify the context engine that the session_id rotated because of
+        # compression (not a fresh /new). Plugin engines (e.g. hermes-lcm) use
+        # boundary_reason="compression" to preserve DAG lineage across the
+        # rollover instead of re-initializing fresh per-session state.
+        # See hermes-lcm#68. Built-in ContextCompressor ignores kwargs.
+        try:
+            _old_sid = locals().get("old_session_id")
+            if _old_sid and hasattr(self.context_compressor, "on_session_start"):
+                self.context_compressor.on_session_start(
+                    self.session_id or "",
+                    boundary_reason="compression",
+                    old_session_id=_old_sid,
+                )
+        except Exception as _ce_err:
+            logger.debug("context engine on_session_start (compression): %s", _ce_err)
+
         # Warn on repeated compressions (quality degrades with each pass)
         _cc = self.context_compressor.compression_count
         if _cc >= 2:

scripts/install.sh

@@ -1055,10 +1055,37 @@ setup_path() {
         return 0
     fi
 
-    # FHS layout: /usr/local/bin is on PATH for every standard shell, nothing to inject.
+    # FHS layout: /usr/local/bin is normally on PATH for login shells (via
+    # /etc/profile pathmunge), but on RHEL/CentOS/Rocky/Alma 8+ non-login
+    # interactive root shells (su, sudo -s, tmux panes, some web terminals)
+    # only source /etc/bashrc, which does NOT add /usr/local/bin — and
+    # /root/.bash_profile doesn't either.  So verify with `command -v` and
+    # fall back to writing a PATH guard into /root/.bashrc when needed.
     if [ "$ROOT_FHS_LAYOUT" = true ]; then
         export PATH="$command_link_dir:$PATH"
-        log_info "/usr/local/bin is already on PATH for all shells"
+        # Probe a fresh non-login interactive bash the way the user will use it.
+        # `bash -i -c` sources ~/.bashrc but NOT ~/.bash_profile or /etc/profile,
+        # which is the exact scenario where RHEL root loses /usr/local/bin.
+        if env -i HOME="$HOME" TERM="${TERM:-dumb}" bash -i -c 'command -v hermes' \
+                >/dev/null 2>&1; then
+            log_info "/usr/local/bin is already on PATH for all shells"
+            log_success "hermes command ready"
+            return 0
+        fi
+
+        log_info "hermes not on PATH in non-login shells (common on RHEL-family)"
+        PATH_LINE='export PATH="/usr/local/bin:$PATH"'
+        PATH_COMMENT='# Hermes Agent — ensure /usr/local/bin is on PATH (RHEL non-login shells)'
+        for SHELL_CONFIG in "$HOME/.bashrc" "$HOME/.bash_profile"; do
+            [ -f "$SHELL_CONFIG" ] || continue
+            if ! grep -v '^[[:space:]]*#' "$SHELL_CONFIG" 2>/dev/null \
+                    | grep -qE 'PATH=.*(/usr/local/bin|\$command_link_dir)'; then
+                echo "" >> "$SHELL_CONFIG"
+                echo "$PATH_COMMENT" >> "$SHELL_CONFIG"
+                echo "$PATH_LINE" >> "$SHELL_CONFIG"
+                log_success "Added /usr/local/bin to PATH in $SHELL_CONFIG"
+            fi
+        done
         log_success "hermes command ready"
         return 0
     fi

scripts/profile-tui.py

@@ -0,0 +1,614 @@
+#!/usr/bin/env python3
+"""Drive the Hermes TUI under HERMES_DEV_PERF and summarize the pipeline.
+
+Usage:
+  scripts/profile-tui.py [--session SID] [--hold KEY] [--seconds N] [--rate HZ]
+
+Defaults: picks the session with the most messages, holds PageUp for 8s at
+~30 Hz (matching xterm key-repeat), summarizes ~/.hermes/perf.log on exit.
+
+The --tui build must exist (run `npm run build` in ui-tui first). This script
+launches `node dist/entry.js` directly with HERMES_TUI_RESUME set so it
+bypasses the hermes_cli wrapper — we want repeatable timing, not the CLI's
+session-picker flow.
+
+Environment overrides:
+  HERMES_PERF_LOG     (default ~/.hermes/perf.log)
+  HERMES_PERF_NODE    (default node from $PATH)
+  HERMES_TUI_DIR      (default /home/bb/hermes-agent/ui-tui)
+
+Exit code is 0 if the harness ran and parsed results, 2 if the TUI crashed
+or produced no perf data (suggests HERMES_DEV_PERF wiring is broken).
+"""
+
+from __future__ import annotations
+
+import argparse
+import json
+import os
+import pty
+import select
+import signal
+import sqlite3
+import sys
+import time
+from pathlib import Path
+from typing import Any
+
+
+DEFAULT_TUI_DIR = Path(os.environ.get("HERMES_TUI_DIR", "/home/bb/hermes-agent/ui-tui"))
+DEFAULT_LOG = Path(os.environ.get("HERMES_PERF_LOG", str(Path.home() / ".hermes" / "perf.log")))
+DEFAULT_STATE_DB = Path.home() / ".hermes" / "state.db"
+
+# Keystroke escape sequences.  Matches what xterm/VT220 send when the
+# terminal has bracketed-paste disabled and the key-repeat handler fires.
+KEYS = {
+    "page_up": b"\x1b[5~",
+    "page_down": b"\x1b[6~",
+    "wheel_up": b"\x1b[M`!!",      # mouse wheel up (SGR-less) — best-effort
+    "shift_up": b"\x1b[1;2A",
+    "shift_down": b"\x1b[1;2B",
+}
+
+
+def pick_longest_session(db: Path) -> str:
+    conn = sqlite3.connect(db)
+    row = conn.execute(
+        "SELECT id FROM sessions s ORDER BY "
+        "(SELECT COUNT(*) FROM messages m WHERE m.session_id = s.id) DESC LIMIT 1"
+    ).fetchone()
+    if not row:
+        sys.exit(f"no sessions in {db}")
+    return row[0]
+
+
+def drain(fd: int, timeout: float) -> bytes:
+    """Read whatever's available from fd within `timeout`, then return."""
+    chunks = []
+    end = time.monotonic() + timeout
+    while time.monotonic() < end:
+        r, _, _ = select.select([fd], [], [], max(0.0, end - time.monotonic()))
+        if not r:
+            break
+        try:
+            data = os.read(fd, 4096)
+        except OSError:
+            break
+        if not data:
+            break
+        chunks.append(data)
+    return b"".join(chunks)
+
+
+def hold_key(fd: int, seq: bytes, seconds: float, rate_hz: int) -> int:
+    """Write `seq` to fd at ~rate_hz for `seconds`. Returns keystrokes sent."""
+    interval = 1.0 / max(1, rate_hz)
+    end = time.monotonic() + seconds
+    sent = 0
+    while time.monotonic() < end:
+        try:
+            os.write(fd, seq)
+            sent += 1
+        except OSError:
+            break
+        # Drain stdout to keep the PTY buffer flowing; ignore content.
+        drain(fd, 0)
+        time.sleep(interval)
+    return sent
+
+
+def summarize(log: Path, since_ts_ms: int) -> dict[str, Any]:
+    """Parse perf.log, keep only events newer than since_ts_ms, return stats."""
+    react_events: list[dict[str, Any]] = []
+    frame_events: list[dict[str, Any]] = []
+    if not log.exists():
+        return {"error": f"no log at {log}", "react": [], "frame": []}
+    for line in log.read_text().splitlines():
+        line = line.strip()
+        if not line:
+            continue
+        try:
+            row = json.loads(line)
+        except json.JSONDecodeError:
+            continue
+        if int(row.get("ts", 0)) < since_ts_ms:
+            continue
+        src = row.get("src")
+        if src == "react":
+            react_events.append(row)
+        elif src == "frame":
+            frame_events.append(row)
+
+    return {
+        "react": react_events,
+        "frame": frame_events,
+    }
+
+
+def pct(values: list[float], p: float) -> float:
+    if not values:
+        return 0.0
+    s = sorted(values)
+    idx = min(len(s) - 1, int(len(s) * p))
+    return s[idx]
+
+
+def format_report(data: dict[str, Any]) -> str:
+    react = data.get("react") or []
+    frames = data.get("frame") or []
+    out = []
+
+    out.append("═══ React Profiler ═══")
+    if not react:
+        out.append("  (no react events — HERMES_DEV_PERF wired? threshold too high?)")
+    else:
+        by_id: dict[str, list[float]] = {}
+        for r in react:
+            by_id.setdefault(r["id"], []).append(r["actualMs"])
+        out.append(f"  {'pane':<14} {'count':>6} {'p50':>8} {'p95':>8} {'p99':>8} {'max':>8}")
+        for pid, ms in sorted(by_id.items(), key=lambda kv: -pct(kv[1], 0.99)):
+            out.append(
+                f"  {pid:<14} {len(ms):>6} {pct(ms,0.50):>8.2f} {pct(ms,0.95):>8.2f} "
+                f"{pct(ms,0.99):>8.2f} {max(ms):>8.2f}"
+            )
+
+    out.append("")
+    out.append("═══ Ink pipeline ═══")
+    if not frames:
+        out.append("  (no frame events — onFrame wiring broken?)")
+    else:
+        dur = [f["durationMs"] for f in frames]
+        phases_present = any(f.get("phases") for f in frames)
+        out.append(f"  frames captured: {len(frames)}")
+        out.append(
+            f"  durationMs  p50={pct(dur,0.50):.2f}  p95={pct(dur,0.95):.2f}  "
+            f"p99={pct(dur,0.99):.2f}  max={max(dur):.2f}"
+        )
+        # Effective FPS during the run: frames / elapsed seconds.
+        ts = sorted(f["ts"] for f in frames)
+        if len(ts) >= 2:
+            elapsed_s = (ts[-1] - ts[0]) / 1000.0
+            fps = len(frames) / elapsed_s if elapsed_s > 0 else float("inf")
+            out.append(f"  throughput: {len(frames)} frames / {elapsed_s:.2f}s = {fps:.1f} fps")
+
+        if phases_present:
+            fields = ["yoga", "renderer", "diff", "optimize", "write", "commit"]
+            out.append("")
+            out.append(f"  {'phase':<10} {'p50':>8} {'p95':>8} {'p99':>8} {'max':>8}   (ms)")
+            for field in fields:
+                vals = [f["phases"][field] for f in frames if f.get("phases")]
+                if vals:
+                    out.append(
+                        f"  {field:<10} {pct(vals,0.50):>8.2f} {pct(vals,0.95):>8.2f} "
+                        f"{pct(vals,0.99):>8.2f} {max(vals):>8.2f}"
+                    )
+            # Derived: sum of phases vs durationMs (reveals hidden time).
+            sum_ps = [
+                sum(f["phases"][k] for k in fields)
+                for f in frames if f.get("phases")
+            ]
+            if sum_ps:
+                dur_match = [f["durationMs"] for f in frames if f.get("phases")]
+                deltas = [d - s for d, s in zip(dur_match, sum_ps)]
+                out.append(
+                    f"  {'dur-Σphases':<10} {pct(deltas,0.50):>8.2f} {pct(deltas,0.95):>8.2f} "
+                    f"{pct(deltas,0.99):>8.2f} {max(deltas):>8.2f}   (unaccounted-for time)"
+                )
+
+            # Yoga counters
+            visited = [f["phases"]["yogaVisited"] for f in frames if f.get("phases")]
+            measured = [f["phases"]["yogaMeasured"] for f in frames if f.get("phases")]
+            cache_hits = [f["phases"]["yogaCacheHits"] for f in frames if f.get("phases")]
+            live = [f["phases"]["yogaLive"] for f in frames if f.get("phases")]
+            out.append("")
+            out.append("  Yoga counters (per frame):")
+            for name, vals in (
+                ("visited", visited),
+                ("measured", measured),
+                ("cacheHits", cache_hits),
+                ("live", live),
+            ):
+                if vals:
+                    out.append(f"    {name:<11} p50={pct(vals,0.5):.0f}  p99={pct(vals,0.99):.0f}  max={max(vals)}")
+
+            # Patch counts — proxy for "how much changed each frame"
+            patches = [f["phases"]["patches"] for f in frames if f.get("phases")]
+            if patches:
+                out.append(
+                    f"  patches     p50={pct(patches,0.5):.0f}  p99={pct(patches,0.99):.0f}  "
+                    f"max={max(patches)}  total={sum(patches)}"
+                )
+            optimized = [
+                f["phases"].get("optimizedPatches", 0)
+                for f in frames if f.get("phases")
+            ]
+            if any(optimized):
+                out.append(
+                    f"  optimized   p50={pct(optimized,0.5):.0f}  p99={pct(optimized,0.99):.0f}  "
+                    f"max={max(optimized)}  total={sum(optimized)}"
+                    f"  (ratio: {sum(optimized)/max(1,sum(patches)):.2f})"
+                )
+
+            # Write bytes + drain telemetry — the outer-terminal bottleneck gauge.
+            bytes_written = [
+                f["phases"].get("writeBytes", 0)
+                for f in frames if f.get("phases")
+            ]
+            if any(bytes_written):
+                total_b = sum(bytes_written)
+                kb = total_b / 1024
+                out.append(
+                    f"  writeBytes  p50={pct(bytes_written,0.5):.0f}B  p99={pct(bytes_written,0.99):.0f}B  "
+                    f"max={max(bytes_written)}B  total={kb:.1f}KB"
+                )
+            drains = [
+                f["phases"].get("prevFrameDrainMs", 0)
+                for f in frames if f.get("phases")
+            ]
+            if any(d > 0 for d in drains):
+                nonzero = [d for d in drains if d > 0]
+                out.append(
+                    f"  drainMs     p50={pct(nonzero,0.5):.2f}  p95={pct(nonzero,0.95):.2f}  "
+                    f"p99={pct(nonzero,0.99):.2f}  max={max(nonzero):.2f}   (terminal flush latency)"
+                )
+            backpressure = sum(1 for f in frames if f.get("phases", {}).get("backpressure"))
+            if backpressure:
+                out.append(
+                    f"  backpressure: {backpressure}/{len(frames)} frames "
+                    f"({100*backpressure/len(frames):.0f}%)   (Node stdout buffer full — terminal slow)"
+                )
+
+        # Flickers
+        flicker_frames = [f for f in frames if f.get("flickers")]
+        if flicker_frames:
+            out.append("")
+            out.append(f"  ⚠ flickers detected in {len(flicker_frames)} frames")
+            reasons: dict[str, int] = {}
+            for f in flicker_frames:
+                for fl in f["flickers"]:
+                    reasons[fl["reason"]] = reasons.get(fl["reason"], 0) + 1
+            for reason, n in sorted(reasons.items(), key=lambda kv: -kv[1]):
+                out.append(f"    {reason}: {n}")
+
+    return "\n".join(out)
+
+
+def key_metrics(data: dict[str, Any]) -> dict[str, float]:
+    """Flatten the report into a dict of scalar metrics for A/B diffing."""
+    metrics: dict[str, float] = {}
+    frames = data.get("frame") or []
+    react = data.get("react") or []
+
+    if frames:
+        durs = [f["durationMs"] for f in frames]
+        metrics["frames"] = len(frames)
+        metrics["dur_p50"] = pct(durs, 0.50)
+        metrics["dur_p95"] = pct(durs, 0.95)
+        metrics["dur_p99"] = pct(durs, 0.99)
+        metrics["dur_max"] = max(durs)
+
+        ts = sorted(f["ts"] for f in frames)
+        if len(ts) >= 2:
+            elapsed = (ts[-1] - ts[0]) / 1000.0
+            metrics["fps_throughput"] = len(frames) / elapsed if elapsed > 0 else 0.0
+            # Interframe gaps distribution — complementary view to throughput:
+            gaps = [ts[i] - ts[i - 1] for i in range(1, len(ts))]
+            if gaps:
+                metrics["gap_p50_ms"] = pct(gaps, 0.50)
+                metrics["gap_p99_ms"] = pct(gaps, 0.99)
+                metrics["gaps_under_16ms"] = sum(1 for g in gaps if g < 16)
+                metrics["gaps_over_200ms"] = sum(1 for g in gaps if g >= 200)
+
+        for phase in ("renderer", "yoga", "diff", "write"):
+            vals = [f["phases"][phase] for f in frames if f.get("phases")]
+            if vals:
+                metrics[f"{phase}_p99"] = pct(vals, 0.99)
+                metrics[f"{phase}_max"] = max(vals)
+
+        patches = [f["phases"]["patches"] for f in frames if f.get("phases")]
+        if patches:
+            metrics["patches_total"] = sum(patches)
+            metrics["patches_p99"] = pct(patches, 0.99)
+
+        optimized = [
+            f["phases"].get("optimizedPatches", 0) for f in frames if f.get("phases")
+        ]
+        if any(optimized):
+            metrics["optimized_total"] = sum(optimized)
+
+        bytes_list = [
+            f["phases"].get("writeBytes", 0) for f in frames if f.get("phases")
+        ]
+        if any(bytes_list):
+            metrics["writeBytes_total"] = sum(bytes_list)
+
+        drains = [
+            f["phases"].get("prevFrameDrainMs", 0)
+            for f in frames if f.get("phases")
+        ]
+        drain_nonzero = [d for d in drains if d > 0]
+        if drain_nonzero:
+            metrics["drain_p99"] = pct(drain_nonzero, 0.99)
+            metrics["drain_max"] = max(drain_nonzero)
+
+        bp = sum(1 for f in frames if f.get("phases", {}).get("backpressure"))
+        metrics["backpressure_frames"] = bp
+
+    if react:
+        for pid in set(e["id"] for e in react):
+            ms = [e["actualMs"] for e in react if e["id"] == pid]
+            metrics[f"react_{pid}_p99"] = pct(ms, 0.99)
+            metrics[f"react_{pid}_max"] = max(ms)
+
+    return metrics
+
+
+def format_diff(before: dict[str, float], after: dict[str, float]) -> str:
+    """Render a side-by-side A/B comparison table."""
+    keys = sorted(set(before) | set(after))
+    lines = [f"{'metric':<28} {'before':>12} {'after':>12} {'delta':>12}  {'%':>6}"]
+    lines.append("─" * 76)
+    for k in keys:
+        b = before.get(k, 0.0)
+        a = after.get(k, 0.0)
+        d = a - b
+        pct_change = ((a / b) - 1) * 100 if b not in (0, 0.0) else float("inf") if a else 0
+
+        # Flag improvements vs regressions. For _p99 / _max / _total / gaps_over /
+        # patches / writeBytes / backpressure, LOWER is better.  For fps / gaps_under,
+        # HIGHER is better.
+        lower_is_better = any(
+            token in k
+            for token in (
+                "p50",
+                "p95",
+                "p99",
+                "_max",
+                "_total",
+                "gaps_over",
+                "backpressure",
+                "drain",
+            )
+        )
+        higher_is_better = "fps_" in k or "gaps_under" in k
+        mark = ""
+        if d and not (lower_is_better or higher_is_better):
+            mark = ""
+        elif d < 0 and lower_is_better:
+            mark = "↓"
+        elif d > 0 and higher_is_better:
+            mark = "↑"
+        elif d > 0 and lower_is_better:
+            mark = "↑"  # regression
+        elif d < 0 and higher_is_better:
+            mark = "↓"  # regression
+
+        pct_str = "—" if pct_change == float("inf") else f"{pct_change:+6.1f}%"
+        lines.append(
+            f"{k:<28} {b:>12.2f} {a:>12.2f} {d:>+12.2f}  {pct_str} {mark}"
+        )
+
+    return "\n".join(lines)
+
+
+def run_once(args: argparse.Namespace) -> dict[str, Any]:
+    tui_dir = Path(args.tui_dir).resolve()
+    entry = tui_dir / "dist" / "entry.js"
+    if not entry.exists():
+        sys.exit(f"{entry} missing — run `npm run build` in {tui_dir} first")
+
+    sid = args.session or pick_longest_session(DEFAULT_STATE_DB)
+    print(f"• session: {sid}")
+    print(f"• hold: {args.hold} x {args.rate}Hz for {args.seconds}s after {args.warmup}s warmup")
+    print(f"• terminal: {args.cols}x{args.rows}")
+
+    log = Path(args.log)
+    if not args.keep_log and log.exists():
+        log.unlink()
+
+    since_ms = int(time.time() * 1000)
+
+    env = os.environ.copy()
+    env["HERMES_DEV_PERF"] = "1"
+    env["HERMES_DEV_PERF_MS"] = str(args.threshold_ms)
+    env["HERMES_DEV_PERF_LOG"] = str(log)
+    env["HERMES_TUI_RESUME"] = sid
+    env["COLUMNS"] = str(args.cols)
+    env["LINES"] = str(args.rows)
+    env["TERM"] = env.get("TERM", "xterm-256color")
+
+    # Pass through extra flags the TUI wrapper recognizes (e.g. --no-fullscreen).
+    # Stored on args as `extra_flags` list.
+    node = os.environ.get("HERMES_PERF_NODE", "node")
+    node_args = [node, str(entry), *getattr(args, "extra_flags", [])]
+
+    pid, fd = pty.fork()
+    if pid == 0:
+        os.execvpe(node, node_args, env)
+
+    try:
+        import fcntl, struct, termios
+        winsize = struct.pack("HHHH", args.rows, args.cols, 0, 0)
+        fcntl.ioctl(fd, termios.TIOCSWINSZ, winsize)
+
+        print(f"• pid: {pid}  fd: {fd}")
+        print(f"• warmup {args.warmup}s (drain startup output)…")
+        drain(fd, args.warmup)
+
+        print(f"• holding {args.hold}…")
+        sent = hold_key(fd, KEYS[args.hold], args.seconds, args.rate)
+        print(f"  sent {sent} keystrokes")
+
+        drain(fd, 0.5)
+    finally:
+        try:
+            os.kill(pid, signal.SIGTERM)
+            for _ in range(10):
+                pid_done, _ = os.waitpid(pid, os.WNOHANG)
+                if pid_done == pid:
+                    break
+                time.sleep(0.1)
+            else:
+                os.kill(pid, signal.SIGKILL)
+                os.waitpid(pid, 0)
+        except (ProcessLookupError, ChildProcessError):
+            pass
+        try:
+            os.close(fd)
+        except OSError:
+            pass
+
+    time.sleep(0.2)
+    return summarize(log, since_ms)
+
+
+def main() -> int:
+    p = argparse.ArgumentParser()
+    p.add_argument("--session", help="session id to resume (default: longest in db)")
+    p.add_argument("--hold", default="page_up", choices=sorted(KEYS.keys()), help="key to hold")
+    p.add_argument("--seconds", type=float, default=8.0, help="how long to hold the key")
+    p.add_argument("--rate", type=int, default=30, help="keystrokes per second")
+    p.add_argument("--warmup", type=float, default=3.0, help="seconds to wait after launch before input")
+    p.add_argument("--threshold-ms", type=float, default=0.0, help="HERMES_DEV_PERF_MS (0 = capture all)")
+    p.add_argument("--cols", type=int, default=120)
+    p.add_argument("--rows", type=int, default=40)
+    p.add_argument("--keep-log", action="store_true", help="don't wipe perf.log before run")
+    p.add_argument("--tui-dir", default=str(DEFAULT_TUI_DIR))
+    p.add_argument("--log", default=str(DEFAULT_LOG))
+    p.add_argument("--save", metavar="LABEL",
+                   help="save the final metrics as /tmp/perf-<LABEL>.json for later --compare")
+    p.add_argument("--compare", metavar="LABEL",
+                   help="diff against /tmp/perf-<LABEL>.json after running")
+    p.add_argument("--loop", action="store_true",
+                   help="watch for source changes, rebuild, rerun, and diff vs previous run")
+    p.add_argument("--extra-flag", dest="extra_flags", action="append", default=[],
+                   help="pass through to node dist/entry.js (repeatable)")
+    args = p.parse_args()
+
+    if args.loop:
+        return loop_mode(args)
+
+    # Single-shot path.
+    data = run_once(args)
+    print()
+    print(format_report(data))
+
+    metrics = key_metrics(data)
+
+    if args.save:
+        path = Path(f"/tmp/perf-{args.save}.json")
+        path.write_text(json.dumps(metrics, indent=2))
+        print(f"\n• saved: {path}")
+
+    if args.compare:
+        path = Path(f"/tmp/perf-{args.compare}.json")
+        if not path.exists():
+            print(f"\n⚠ no baseline at {path} — run with --save {args.compare} first")
+        else:
+            before = json.loads(path.read_text())
+            print(f"\n═══ A/B diff vs /tmp/perf-{args.compare}.json ═══")
+            print(format_diff(before, metrics))
+
+    if not data["react"] and not data["frame"]:
+        return 2
+    return 0
+
+
+def loop_mode(args: argparse.Namespace) -> int:
+    """Watch source files, rebuild, rerun, print A/B diff against previous run.
+
+    Keeps a rolling 'previous run' baseline in memory so each iteration
+    reports delta vs the last one — visibility into whether the last
+    edit moved the needle.  Press Ctrl+C to stop.
+    """
+    import subprocess
+
+    tui_dir = Path(args.tui_dir).resolve()
+    src_root = tui_dir / "src"
+    pkg_root = tui_dir / "packages" / "hermes-ink" / "src"
+
+    def collect_mtimes() -> dict[str, float]:
+        mtimes: dict[str, float] = {}
+        for root in (src_root, pkg_root):
+            if not root.exists():
+                continue
+            for path in root.rglob("*"):
+                if path.suffix in {".ts", ".tsx"} and "__tests__" not in str(path):
+                    try:
+                        mtimes[str(path)] = path.stat().st_mtime
+                    except OSError:
+                        pass
+        return mtimes
+
+    previous_metrics: dict[str, float] | None = None
+    previous_mtimes = collect_mtimes()
+    iteration = 0
+
+    print(f"• loop mode — watching {src_root} + {pkg_root} for *.ts(x) changes")
+    print("• edit any TS file, the harness rebuilds + reruns automatically")
+    print("• Ctrl+C to stop\n")
+
+    try:
+        while True:
+            iteration += 1
+            print(f"\n{'═' * 76}")
+            print(f"Iteration {iteration}  @ {time.strftime('%H:%M:%S')}")
+            print("═" * 76)
+
+            if iteration > 1:
+                print("• rebuilding…")
+                result = subprocess.run(
+                    ["npm", "run", "build"],
+                    cwd=tui_dir,
+                    capture_output=True,
+                    text=True,
+                )
+                if result.returncode != 0:
+                    print("✗ build failed:")
+                    print(result.stdout[-2000:])
+                    print(result.stderr[-2000:])
+                    print("\n• waiting for source changes to retry…")
+                    previous_mtimes = wait_for_change(previous_mtimes, collect_mtimes)
+                    continue
+                print("✓ build ok")
+
+            data = run_once(args)
+            metrics = key_metrics(data)
+
+            print()
+            print(format_report(data))
+
+            if previous_metrics is not None:
+                print(f"\n═══ A/B diff vs iteration {iteration - 1} ═══")
+                print(format_diff(previous_metrics, metrics))
+
+            previous_metrics = metrics
+
+            print("\n• waiting for source changes…")
+            previous_mtimes = wait_for_change(previous_mtimes, collect_mtimes)
+    except KeyboardInterrupt:
+        print("\n• loop stopped")
+        return 0
+
+
+def wait_for_change(prev: dict[str, float], collect) -> dict[str, float]:
+    """Poll every 1s until a watched file's mtime changes. Debounced 500ms."""
+    while True:
+        time.sleep(1)
+        current = collect()
+
+        changed = [
+            path for path, mtime in current.items() if prev.get(path) != mtime
+        ]
+
+        if changed:
+            print(f"  ↻ {len(changed)} file(s) changed:")
+            for path in changed[:5]:
+                print(f"    {path}")
+            # Debounce — editor save bursts can take ~500ms to settle
+            time.sleep(0.5)
+            return collect()
+
+
+if __name__ == "__main__":
+    sys.exit(main())

scripts/release.py

@@ -43,16 +43,24 @@ AUTHOR_MAP = {
     "teknium1@gmail.com": "teknium1",
     "teknium@nousresearch.com": "teknium1",
     "127238744+teknium1@users.noreply.github.com": "teknium1",
+    "johnnncenaaa77@gmail.com": "johnncenae",
     "focusflow.app.help@gmail.com": "yes999zc",
     "343873859@qq.com": "DrStrangerUJN",
     "uzmpsk.dilekakbas@gmail.com": "dlkakbs",
     "jefferson@heimdallstrategy.com": "Mind-Dragon",
     "130918800+devorun@users.noreply.github.com": "devorun",
+    "sonoyuncudmr@gmail.com": "Sonoyunchu",
     "maks.mir@yahoo.com": "say8hi",
     "web3blind@users.noreply.github.com": "web3blind",
     "julia@alexland.us": "alexg0bot",
     "1060770+benjaminsehl@users.noreply.github.com": "benjaminsehl",
     "nerijusn76@gmail.com": "Nerijusas",
+    "itonov@proton.me": "Ito-69",
+    "glesstech@gmail.com": "georgeglessner",
+    "maxim.smetanin@gmail.com": "maxims-oss",
+    "CREWorx@users.noreply.github.com": "BadTechBandit",
+    "yoimexex@gmail.com": "Yoimex",
+    "6548898+romanornr@users.noreply.github.com": "romanornr",
     # contributors (from noreply pattern)
     "david.vv@icloud.com": "davidvv",
     "wangqiang@wangqiangdeMac-mini.local": "xiaoqiang243",
@@ -67,9 +75,12 @@ AUTHOR_MAP = {
     "thomasgeorgevii09@gmail.com": "tochukwuada",
     "harryykyle1@gmail.com": "hharry11",
     "kshitijk4poor@gmail.com": "kshitijk4poor",
+    "1294707+Tosko4@users.noreply.github.com": "Tosko4",
     "keira.voss94@gmail.com": "keiravoss94",
     "16443023+stablegenius49@users.noreply.github.com": "stablegenius49",
     "fqsy1416@gmail.com": "EKKOLearnAI",
+    "octo-patch@github.com": "octo-patch",
+    "math0r-be@github.com": "math0r-be",
     "simbamax99@gmail.com": "simbam99",
     "iris@growthpillars.co": "irispillars",
     "185121704+stablegenius49@users.noreply.github.com": "stablegenius49",
@@ -116,9 +127,22 @@ AUTHOR_MAP = {
     "Mibayy@users.noreply.github.com": "Mibayy",
     "mibayy@users.noreply.github.com": "Mibayy",
     "135070653+sgaofen@users.noreply.github.com": "sgaofen",
+    "lzy.dev@gmail.com": "zhiyanliu",
+    "me@janstepanovsky.cz": "hhhonzik",
+    "139848623+hhuang91@users.noreply.github.com": "hhuang91",
+    "s.ozaki@ebinou.net": "Satoshi-agi",
+    "10774721+kunlabs@users.noreply.github.com": "kunlabs",
+    "110560187+Wang-tianhao@users.noreply.github.com": "Wang-tianhao",
+    "170458616+ghostmfr@users.noreply.github.com": "ghostmfr",
+    "1848670+mewwts@users.noreply.github.com": "mewwts",
+    "1930707+haru398801@users.noreply.github.com": "haru398801",
+    "rapabelias@gmail.com": "badgerbees",
+    "xnb888@proton.me": "xnbi",
+    "xiahu889889@proton.me": "xiahu88988",
     "nocoo@users.noreply.github.com": "nocoo",
     "30841158+n-WN@users.noreply.github.com": "n-WN",
     "tsuijinglei@gmail.com": "hiddenpuppy",
+    "buraysandro9@gmail.com": "ygd58",
     "jerome@clawwork.ai": "HiddenPuppy",
     "jerome.benoit@sap.com": "jerome-benoit",
     "wysie@users.noreply.github.com": "Wysie",
@@ -191,6 +215,7 @@ AUTHOR_MAP = {
     "satelerd@gmail.com": "satelerd",
     "dan@danlynn.com": "danklynn",
     "mattmaximo@hotmail.com": "MattMaximo",
+    "MatthewRHardwick@gmail.com": "mrhwick",
     "149063006+j3ffffff@users.noreply.github.com": "j3ffffff",
     "A-FdL-Prog@users.noreply.github.com": "A-FdL-Prog",
     "l0hde@users.noreply.github.com": "l0hde",
@@ -377,6 +402,17 @@ AUTHOR_MAP = {
     "zzn+pa@zzn.im": "xinbenlv",
     "zaynjarvis@gmail.com": "ZaynJarvis",
     "zhiheng.liu@bytedance.com": "ZaynJarvis",
+    "izhaolongfei@gmail.com": "loongfay",
+    "296659110@qq.com": "lrt4836",
+    "fe.daniel91@gmail.com": "beforeload",
+    "libo1106@foxmail.com": "libo1106",
+    "295367131@qq.com": "295367131",
+    "295367132@qq.com": "IxAres",
+    "danieldliu@tencent.com": "danieldliu",
+    "loongzhao@tencent.com": "loongzhao",
+    "Bartok9@users.noreply.github.com": "Bartok9",
+    "LeonSGP43@users.noreply.github.com": "LeonSGP43",
+    "kshitijk4poor@users.noreply.github.com": "kshitijk4poor",
     "mbelleau@Michels-MacBook-Pro.local": "malaiwah",
     "michel.belleau@malaiwah.com": "malaiwah",
     "gnanasekaran.sekareee@gmail.com": "gnanam1990",

skills/apple/apple-notes/SKILL.md

@@ -1,6 +1,6 @@
 ---
 name: apple-notes
-description: Manage Apple Notes via the memo CLI on macOS (create, view, search, edit).
+description: "Manage Apple Notes via memo CLI: create, search, edit."
 version: 1.0.0
 author: Hermes Agent
 license: MIT

skills/apple/apple-reminders/SKILL.md

@@ -1,6 +1,6 @@
 ---
 name: apple-reminders
-description: Manage Apple Reminders via remindctl CLI (list, add, complete, delete).
+description: "Apple Reminders via remindctl: add, list, complete."
 version: 1.0.0
 author: Hermes Agent
 license: MIT

skills/apple/findmy/SKILL.md

@@ -1,6 +1,6 @@
 ---
 name: findmy
-description: Track Apple devices and AirTags via FindMy.app on macOS using AppleScript and screen capture.
+description: "Track Apple devices/AirTags via FindMy.app on macOS."
 version: 1.0.0
 author: Hermes Agent
 license: MIT

skills/autonomous-ai-agents/claude-code/SKILL.md

@@ -1,6 +1,6 @@
 ---
 name: claude-code
-description: Delegate coding tasks to Claude Code (Anthropic's CLI agent). Use for building features, refactoring, PR reviews, and iterative coding. Requires the claude CLI installed.
+description: "Delegate coding to Claude Code CLI (features, PRs)."
 version: 2.2.0
 author: Hermes Agent + Teknium
 license: MIT

skills/autonomous-ai-agents/codex/SKILL.md

@@ -1,6 +1,6 @@
 ---
 name: codex
-description: Delegate coding tasks to OpenAI Codex CLI agent. Use for building features, refactoring, PR reviews, and batch issue fixing. Requires the codex CLI and a git repository.
+description: "Delegate coding to OpenAI Codex CLI (features, PRs)."
 version: 1.0.0
 author: Hermes Agent
 license: MIT
@@ -14,6 +14,15 @@ metadata:
 
 Delegate coding tasks to [Codex](https://github.com/openai/codex) via the Hermes terminal. Codex is OpenAI's autonomous coding agent CLI.
 
+## When to use
+
+- Building features
+- Refactoring
+- PR reviews
+- Batch issue fixing
+
+Requires the codex CLI and a git repository.
+
 ## Prerequisites
 
 - Codex installed: `npm install -g @openai/codex`

skills/autonomous-ai-agents/hermes-agent/SKILL.md

@@ -1,6 +1,6 @@
 ---
 name: hermes-agent
-description: Complete guide to using and extending Hermes Agent — CLI usage, setup, configuration, spawning additional agents, gateway platforms, skills, voice, tools, profiles, and a concise contributor reference. Load this skill when helping users configure Hermes, troubleshoot issues, spawn agent instances, or make code contributions.
+description: "Configure, extend, or contribute to Hermes Agent."
 version: 2.0.0
 author: Hermes Agent + Teknium
 license: MIT
@@ -115,7 +115,7 @@ hermes tools disable NAME   Disable a toolset
 
 hermes skills list          List installed skills
 hermes skills search QUERY  Search the skills hub
-hermes skills install ID    Install a skill
+hermes skills install ID    Install a skill (ID can be a hub identifier OR a direct https://…/SKILL.md URL; pass --name to override when frontmatter has no name)
 hermes skills inspect ID    Preview without installing
 hermes skills config        Enable/disable skills per platform
 hermes skills check         Check for updates
@@ -402,6 +402,63 @@ Tool changes take effect on `/reset` (new session). They do NOT apply mid-conver
 
 ---
 
+## Security & Privacy Toggles
+
+Common "why is Hermes doing X to my output / tool calls / commands?" toggles — and the exact commands to change them. Most of these need a fresh session (`/reset` in chat, or start a new `hermes` invocation) because they're read once at startup.
+
+### Secret redaction in tool output
+
+Hermes auto-redacts strings that look like API keys, tokens, and secrets in all tool output (terminal stdout, `read_file`, web content, subagent summaries, etc.) so the model never sees raw credentials. If the user is intentionally working with mock tokens, share-management tokens, or their own secrets and the redaction is getting in the way:
+
+```bash
+hermes config set security.redact_secrets false      # disable globally
+```
+
+**Restart required.** `security.redact_secrets` is snapshotted at import time — setting it mid-session (e.g. via `export HERMES_REDACT_SECRETS=false` from a tool call) will NOT take effect for the running process. Tell the user to run `hermes config set security.redact_secrets false` in a terminal, then start a new session. This is deliberate — it prevents an LLM from turning off redaction on itself mid-task.
+
+Re-enable with:
+```bash
+hermes config set security.redact_secrets true
+```
+
+### PII redaction in gateway messages
+
+Separate from secret redaction. When enabled, the gateway hashes user IDs and strips phone numbers from the session context before it reaches the model:
+
+```bash
+hermes config set privacy.redact_pii true    # enable
+hermes config set privacy.redact_pii false   # disable (default)
+```
+
+### Command approval prompts
+
+By default (`approvals.mode: manual`), Hermes prompts the user before running shell commands flagged as destructive (`rm -rf`, `git reset --hard`, etc.). The modes are:
+
+- `manual` — always prompt (default)
+- `smart` — use an auxiliary LLM to auto-approve low-risk commands, prompt on high-risk
+- `off` — skip all approval prompts (equivalent to `--yolo`)
+
+```bash
+hermes config set approvals.mode smart       # recommended middle ground
+hermes config set approvals.mode off         # bypass everything (not recommended)
+```
+
+Per-invocation bypass without changing config:
+- `hermes --yolo …`
+- `export HERMES_YOLO_MODE=1`
+
+Note: YOLO / `approvals.mode: off` does NOT turn off secret redaction. They are independent.
+
+### Shell hooks allowlist
+
+Some shell-hook integrations require explicit allowlisting before they fire. Managed via `~/.hermes/shell-hooks-allowlist.json` — prompted interactively the first time a hook wants to run.
+
+### Disabling the web/browser/image-gen tools
+
+To keep the model away from network or media tools entirely, open `hermes tools` and toggle per-platform. Takes effect on next session (`/reset`). See the Tools & Skills section above.
+
+---
+
 ## Voice & Transcription
 
 ### STT (Voice → Text)

skills/autonomous-ai-agents/opencode/SKILL.md

@@ -1,6 +1,6 @@
 ---
 name: opencode
-description: Delegate coding tasks to OpenCode CLI agent for feature implementation, refactoring, PR review, and long-running autonomous sessions. Requires the opencode CLI installed and authenticated.
+description: "Delegate coding to OpenCode CLI (features, PR review)."
 version: 1.2.0
 author: Hermes Agent
 license: MIT

skills/creative/architecture-diagram/SKILL.md

@@ -1,6 +1,6 @@
 ---
 name: architecture-diagram
-description: Generate dark-themed SVG diagrams of software systems and cloud infrastructure as standalone HTML files with inline SVG graphics. Semantic component colors (cyan=frontend, emerald=backend, violet=database, amber=cloud/AWS, rose=security, orange=message bus), JetBrains Mono font, grid background. Best suited for software architecture, cloud/VPC topology, microservice maps, service-mesh diagrams, database + API layer diagrams, security groups, message buses — anything that fits a tech-infra deck with a dark aesthetic. If a more specialized diagramming skill exists for the subject (scientific, educational, hand-drawn, animated, etc.), prefer that — otherwise this skill can also serve as a general-purpose SVG diagram fallback. Based on Cocoon AI's architecture-diagram-generator (MIT).
+description: "Dark-themed SVG architecture/cloud/infra diagrams as HTML."
 version: 1.0.0
 author: Cocoon AI (hello@cocoon-ai.com), ported by Hermes Agent
 license: MIT

skills/creative/ascii-art/SKILL.md

@@ -1,6 +1,6 @@
 ---
 name: ascii-art
-description: Generate ASCII art using pyfiglet (571 fonts), cowsay, boxes, toilet, image-to-ascii, remote APIs (asciified, ascii.co.uk), and LLM fallback. No API keys required.
+description: "ASCII art: pyfiglet, cowsay, boxes, image-to-ascii."
 version: 4.0.0
 author: 0xbyt4, Hermes Agent
 license: MIT

skills/creative/ascii-video/SKILL.md

@@ -1,10 +1,18 @@
 ---
 name: ascii-video
-description: "Production pipeline for ASCII art video — any format. Converts video/audio/images/generative input into colored ASCII character video output (MP4, GIF, image sequence). Covers: video-to-ASCII conversion, audio-reactive music visualizers, generative ASCII art animations, hybrid video+audio reactive, text/lyrics overlays, real-time terminal rendering. Use when users request: ASCII video, text art video, terminal-style video, character art animation, retro text visualization, audio visualizer in ASCII, converting video to ASCII art, matrix-style effects, or any animated ASCII output."
+description: "ASCII video: convert video/audio to colored ASCII MP4/GIF."
 ---
 
 # ASCII Video Production Pipeline
 
+## When to use
+
+Use when users request: ASCII video, text art video, terminal-style video, character art animation, retro text visualization, audio visualizer in ASCII, converting video to ASCII art, matrix-style effects, or any animated ASCII output.
+
+## What's inside
+
+Production pipeline for ASCII art video — any format. Converts video/audio/images/generative input into colored ASCII character video output (MP4, GIF, image sequence). Covers: video-to-ASCII conversion, audio-reactive music visualizers, generative ASCII art animations, hybrid video+audio reactive, text/lyrics overlays, real-time terminal rendering.
+
 ## Creative Standard
 
 This is visual art. ASCII characters are the medium; cinema is the standard.

skills/creative/baoyu-comic/SKILL.md

@@ -1,6 +1,6 @@
 ---
 name: baoyu-comic
-description: Knowledge comic creator supporting multiple art styles and tones. Creates original educational comics with detailed panel layouts and sequential image generation. Use when user asks to create "知识漫画", "教育漫画", "biography comic", "tutorial comic", or "Logicomix-style comic".
+description: "Knowledge comics (知识漫画): educational, biography, tutorial."
 version: 1.56.1
 author: 宝玉 (JimLiu)
 license: MIT

skills/creative/baoyu-infographic/SKILL.md

@@ -1,6 +1,6 @@
 ---
 name: baoyu-infographic
-description: Generate professional infographics with 21 layout types and 21 visual styles. Analyzes content, recommends layout×style combinations, and generates publication-ready infographics. Use when user asks to create "infographic", "visual summary", "信息图", "可视化", or "高密度信息大图".
+description: "Infographics: 21 layouts x 21 styles (信息图, 可视化)."
 version: 1.56.1
 author: 宝玉 (JimLiu)
 license: MIT

skills/creative/claude-design/SKILL.md

@@ -0,0 +1,590 @@
+---
+name: claude-design
+description: Design one-off HTML artifacts (landing, deck, prototype).
+version: 1.0.0
+author: BadTechBandit
+license: MIT
+metadata:
+  hermes:
+    tags: [design, html, prototype, ux, ui, creative, artifact, deck, motion, design-system]
+    related_skills: [design-md, popular-web-designs, excalidraw, architecture-diagram]
+---
+
+# Claude Design for CLI/API Agents
+
+Use this skill when the user asks for design work that would normally fit Claude Design, but the agent is running in a CLI/API environment instead of the hosted Claude Design web UI.
+
+The goal is to preserve Claude Design's useful design behavior and taste while removing hosted-tool plumbing that does not exist in normal agent environments.
+
+**Before starting, check for other web-design skills like `popular-web-designs` (ready-to-paste design systems for Stripe, Linear, Vercel, Notion, etc.) and `design-md` (Google's DESIGN.md token spec format).** If the user wants a known brand's look, load `popular-web-designs` alongside this one and let it supply the visual vocabulary. If the deliverable is a token spec file rather than a rendered artifact, use `design-md` instead. Full decision table below.
+
+## When To Use This Skill vs `popular-web-designs` vs `design-md`
+
+Hermes has three design-related skills under `skills/creative/`. They do different jobs — load the right one (or combine them):
+
+| Skill | What it gives you | Use when the user wants... |
+|---|---|---|
+| **claude-design** (this one) | Design *process and taste* — how to scope a brief, gather context, produce variants, verify a local HTML artifact, avoid AI-design slop | a from-scratch designed artifact (landing page, prototype, deck, component lab, motion study) with no specific brand or token system dictated |
+| **popular-web-designs** | 54 ready-to-paste design systems — exact colors, typography, components, CSS values for sites like Stripe, Linear, Vercel, Notion, Airbnb | "make it look like Stripe / Linear / Vercel", a page styled after a known brand, or a visual starting point pulled from a real product |
+| **design-md** | Google's DESIGN.md spec format — author/validate/diff/export design-token files, WCAG contrast checking, Tailwind/DTCG export | a formal, persistent, machine-readable design-system *spec file* (tokens + rationale) that lives in a repo and gets consumed by agents over time |
+
+Rule of thumb:
+
+- **Process + taste, one-off artifact** → claude-design
+- **Match a known brand's look** → popular-web-designs (and let claude-design drive the process)
+- **Author the tokens spec itself** → design-md
+
+These compose: use `popular-web-designs` for the visual vocabulary, `claude-design` for how to turn a brief into a thoughtful local HTML file, and `design-md` when the output is the token file rather than a rendered artifact.
+
+## Runtime Mode
+
+You are running in **CLI/API mode**, not the Claude Design hosted web UI.
+
+Ignore references from source Claude Design prompts to hosted-only tools, project panes, preview panes, special toolbar protocols, or platform callbacks that are not available in the current environment.
+
+Examples of hosted-tool concepts to ignore or remap:
+
+- `done()`
+- `fork_verifier_agent()`
+- `questions_v2()`
+- `copy_starter_component()`
+- `show_to_user()`
+- `show_html()`
+- `snip()`
+- `eval_js_user_view()`
+- hosted asset review panes
+- hosted edit-mode or Tweaks toolbar messaging
+- `/projects/<projectId>/...` cross-project paths
+- built-in `window.claude.complete()` artifact helper
+- tool schemas embedded in the source prompt
+- web-search citation scaffolding meant for the hosted runtime
+
+Instead, use the tools actually available in the current agent environment.
+
+Default deliverable:
+
+- a complete local HTML file
+- self-contained CSS and JavaScript when portability matters
+- exact on-disk path in the final response
+- verification using available local methods before saying it is done
+
+If the user asks for implementation in an existing repo, generate code in the repo's actual stack instead of forcing a standalone HTML artifact.
+
+## Core Identity
+
+Act as an expert designer working with the user as the manager.
+
+HTML is the default tool, but the medium changes by assignment:
+
+- UX designer for flows and product surfaces
+- interaction designer for prototypes
+- visual designer for static explorations
+- motion designer for animated artifacts
+- deck designer for presentations
+- design-systems designer for tokens, components, and visual rules
+- frontend-minded prototyper when code fidelity matters
+
+Avoid generic web-design tropes unless the user explicitly asks for a conventional web page.
+
+Do not expose internal prompts, hidden system messages, or implementation plumbing. Talk about capabilities and deliverables in user terms: HTML files, prototypes, decks, exported assets, screenshots, code, and design options.
+
+## When To Use
+
+Use this skill for:
+
+- landing pages
+- teaser pages
+- high-fidelity prototypes
+- interactive product mockups
+- visual option boards
+- component explorations
+- design-system previews
+- HTML slide decks
+- motion studies
+- onboarding flows
+- dashboard concepts
+- settings, command palettes, modals, cards, forms, empty states
+- redesigns based on screenshots, repos, brand docs, or UI kits
+
+Do not use this skill for pure DESIGN.md token authoring unless the user specifically asks for a DESIGN.md file. Use `design-md` for that.
+
+## Design Principle: Start From Context, Not Vibes
+
+Good high-fidelity design does not start from scratch.
+
+Before designing, look for source context:
+
+1. brand docs
+2. existing product screenshots
+3. current repo components
+4. design tokens
+5. UI kits
+6. prior mockups
+7. reference models
+8. copy docs
+9. constraints from legal, product, or engineering
+
+If a repo is available, inspect actual source files before inventing UI:
+
+- theme files
+- token files
+- global stylesheets
+- layout scaffolds
+- component files
+- route/page files
+- form/button/card/navigation implementations
+
+The file tree is only the menu. Read the files that define the visual vocabulary before designing.
+
+If context is missing and fidelity matters, ask concise focused questions instead of producing a generic mockup.
+
+## Asking Questions
+
+Ask questions when the assignment is new, ambiguous, high-fidelity, externally facing, or depends on taste.
+
+Keep questions short. Do not ask ten questions by default unless the problem is genuinely underspecified.
+
+Usually ask for:
+
+- intended output format
+- audience
+- fidelity level
+- source materials available
+- brand/design system in play
+- number of variations wanted
+- whether to stay conservative or explore divergent ideas
+- which dimension matters most: layout, visual language, interaction, copy, motion, or systemization
+
+Skip questions when:
+
+- the user gave enough direction
+- this is a small tweak
+- the task is clearly a continuation
+- the missing detail has an obvious default
+
+When proceeding with assumptions, label only the important ones.
+
+## Workflow
+
+1. **Understand the brief**
+   - What is being designed?
+   - Who is it for?
+   - What artifact should exist at the end?
+   - What constraints are locked?
+
+2. **Gather context**
+   - Read supplied docs, screenshots, repo files, or design assets.
+   - Identify the visual vocabulary before writing code.
+
+3. **Define the design system for this artifact**
+   - colors
+   - type
+   - spacing
+   - radii
+   - shadows or elevation
+   - motion posture
+   - component treatment
+   - interaction rules
+
+4. **Choose the right format**
+   - Static visual comparison: one HTML canvas with options side by side.
+   - Interaction/flow: clickable prototype.
+   - Presentation: fixed-size HTML deck with slide navigation.
+   - Component exploration: component lab with variants.
+   - Motion: timeline or state-based animation.
+
+5. **Build the artifact**
+   - Prefer a single self-contained HTML file unless the task calls for a repo implementation.
+   - Preserve prior versions for major revisions.
+   - Avoid unnecessary dependencies.
+
+6. **Verify**
+   - Confirm files exist.
+   - Run any available syntax/static checks.
+   - If browser tools are available, open the file and check console errors.
+   - If visual fidelity matters and screenshot tools are available, inspect at least the primary viewport.
+
+7. **Report briefly**
+   - exact file path
+   - what was created
+   - caveats
+   - next decision or next iteration
+
+## Artifact Format Rules
+
+Default to local files.
+
+For standalone artifacts:
+
+- create a descriptive filename, e.g. `Landing Page.html`, `Command Palette Prototype.html`, `Design System Board.html`
+- embed CSS in `<style>`
+- embed JS in `<script>`
+- keep the artifact openable directly in a browser
+- avoid remote dependencies unless they are explicitly useful and stable
+- include responsive behavior unless the format is intentionally fixed-size
+
+For significant revisions:
+
+- preserve the previous version as `Name.html`
+- create `Name v2.html`, `Name v3.html`, etc.
+- or keep one file with in-page toggles if the assignment is variant exploration
+
+For repo implementation:
+
+- follow the repo's actual stack
+- use existing components and tokens where possible
+- do not create a standalone artifact if the user asked for production code
+
+## HTML / CSS / JS Standards
+
+Use modern CSS well:
+
+- CSS variables for tokens
+- CSS grid for layout
+- container queries when helpful
+- `text-wrap: pretty` where supported
+- real focus states
+- real hover states
+- `prefers-reduced-motion` handling for non-trivial motion
+- responsive scaling
+- semantic HTML where practical
+
+Avoid:
+
+- huge monolithic files when a real repo structure is expected
+- fragile hard-coded viewport assumptions
+- inaccessible tiny hit targets
+- decorative JS that fights usability
+- `scrollIntoView` unless there is no safer option
+
+Mobile hit targets should be at least 44px.
+
+For print documents, text should be at least 12pt.
+
+For 1920×1080 slide decks, text should generally be 24px or larger.
+
+## React Guidance for Standalone HTML
+
+Use plain HTML/CSS/JS by default.
+
+Use React only when:
+
+- the artifact needs meaningful state
+- variants/toggles are easier as components
+- interaction complexity warrants it
+- the target implementation is React/Next.js and fidelity matters
+
+If using React from CDN in standalone HTML:
+
+- pin exact versions
+- avoid unpinned `react@18` style URLs
+- avoid `type="module"` unless necessary
+- avoid multiple global objects named `styles`
+- give global style objects specific names, e.g. `commandPaletteStyles`, `deckStyles`
+- if splitting Babel scripts, explicitly attach shared components to `window`
+
+If building inside a real repo, use the repo's package manager and component architecture instead.
+
+## Deck Rules
+
+For slide decks, use a fixed-size canvas and scale it to fit the viewport.
+
+Default slide size: 1920×1080, 16:9.
+
+Requirements:
+
+- keyboard navigation
+- visible slide count
+- localStorage persistence for current slide
+- print-friendly layout when practical
+- screen labels or stable IDs for important slides
+- no speaker notes unless the user explicitly asks
+
+Do not hand-wave a deck as markdown bullets. Create a designed artifact if asked for a deck.
+
+Use 1–2 background colors max unless the brand system requires more.
+
+Keep slides sparse. If a slide feels empty, solve it with layout, rhythm, scale, or imagery placeholders, not filler text.
+
+## Prototype Rules
+
+For interactive prototypes:
+
+- make the primary path clickable
+- include key states: default, hover/focus, loading, empty, error, success where relevant
+- expose variations with in-page controls when useful
+- keep controls out of the final composition unless they are intentionally part of the prototype
+- persist important state in localStorage when refresh continuity matters
+
+If the prototype is meant to model a product flow, design the flow, not just the first screen.
+
+## Variation Rules
+
+When exploring, default to at least three options:
+
+1. **Conservative** — closest to existing patterns / lowest risk
+2. **Strong-fit** — best interpretation of the brief
+3. **Divergent** — more novel, useful for discovering taste boundaries
+
+Variations can explore:
+
+- layout
+- hierarchy
+- type scale
+- density
+- color posture
+- surface treatment
+- motion
+- interaction model
+- copy structure
+- component shape
+
+Do not create variations that are merely color swaps unless color is the actual question.
+
+When the user picks a direction, consolidate. Do not leave the project as a pile of options forever.
+
+## Tweakable Designs in CLI/API Mode
+
+The hosted Claude Design edit-mode toolbar does not exist here.
+
+Still preserve the idea: when useful, add in-page controls called `Tweaks`.
+
+A good `Tweaks` panel can control:
+
+- theme mode
+- layout variant
+- density
+- accent color
+- type scale
+- motion on/off
+- copy variant
+- component variant
+
+Keep it small and unobtrusive. The design should look final when tweaks are hidden.
+
+Persist tweak values with localStorage when helpful.
+
+## Content Discipline
+
+Do not add filler content.
+
+Every element must earn its place.
+
+Avoid:
+
+- fake metrics
+- decorative stats
+- generic feature grids
+- unnecessary icons
+- placeholder testimonials
+- AI-generated fluff sections
+- invented content that changes strategy or claims
+
+If additional sections, pages, copy, or claims would improve the artifact, ask before adding them.
+
+When copy is necessary but not final, mark it as draft or placeholder.
+
+## Anti-Slop Rules
+
+Avoid common AI design sludge:
+
+- aggressive gradient backgrounds
+- glassmorphism by default
+- emoji unless the brand uses them
+- generic SaaS cards with icons everywhere
+- left-border accent callout cards
+- fake dashboards filled with arbitrary numbers
+- stock-photo hero sections
+- oversized rounded rectangles as a substitute for hierarchy
+- rainbow palettes
+- vague labels like “Insights,” “Growth,” “Scale,” “Optimize” without content
+- decorative SVG illustrations pretending to be product imagery
+
+Minimal is not automatically good. Dense is not automatically cluttered. Choose intentionally.
+
+## Typography
+
+Use the existing type system if one exists.
+
+If not, choose type deliberately based on the artifact:
+
+- editorial: serif or humanist headline with restrained sans body
+- software/productivity: precise sans with strong numeric treatment
+- luxury/minimal: fewer weights, more spacing discipline
+- technical: mono accents only, not mono everywhere
+- deck: large, clear, high contrast
+
+Avoid overused defaults when a stronger choice is appropriate.
+
+If using web fonts, keep the number of families and weights low.
+
+Use type as hierarchy before adding boxes, icons, or color.
+
+## Color
+
+Use brand/design-system colors first.
+
+If no palette exists:
+
+- define a small system
+- include neutrals, surface, ink, muted text, border, accent, danger/success if needed
+- use one primary accent unless the assignment calls for a broader palette
+- prefer oklch for harmonious invented palettes when browser support is acceptable
+- check contrast for important text and controls
+
+Do not invent lots of colors from scratch.
+
+## Layout and Composition
+
+Design with rhythm:
+
+- scale
+- whitespace
+- density
+- alignment
+- repetition
+- contrast
+- interruption
+
+Avoid making every section the same card grid.
+
+For product UIs, prioritize speed of comprehension over decoration.
+
+For marketing surfaces, make one idea land per section.
+
+For dashboards, avoid “data slop.” Only show data that helps the user decide or act.
+
+## Motion
+
+Use motion as discipline, not theater.
+
+Good motion:
+
+- clarifies state changes
+- reduces anxiety during loading
+- shows continuity between surfaces
+- gives controls tactility
+- stays subtle
+
+Bad motion:
+
+- loops without purpose
+- delays the user
+- calls attention to itself
+- hides poor hierarchy
+
+Respect `prefers-reduced-motion` for non-trivial animation.
+
+## Images and Icons
+
+Use real supplied imagery when available.
+
+If an asset is missing:
+
+- use a clean placeholder
+- use typography, layout, or abstract texture instead
+- ask for real material when fidelity matters
+
+Do not draw elaborate fake SVG illustrations unless the assignment is explicitly illustration work.
+
+Avoid iconography unless it improves scanning or matches the design system.
+
+## Source-Code Fidelity
+
+When recreating or extending a UI from a repo:
+
+1. inspect the repo tree
+2. identify the actual UI source files
+3. read theme/token/global style/component files
+4. lift exact values where appropriate
+5. match spacing, radii, shadows, copy tone, density, and interaction patterns
+6. only then design or modify
+
+Do not build from memory when source files are available.
+
+For GitHub URLs, parse owner/repo/ref/path correctly and inspect the relevant files before designing.
+
+## Reading Documents and Assets
+
+Read Markdown, HTML, CSS, JS, TS, JSX, TSX, JSON, SVG, and plain text directly when available.
+
+For DOCX/PPTX/PDF, use available local extraction tools if present. If not available, ask the user to provide exported text/images or use another available tool path.
+
+For sketches, prioritize thumbnails or screenshots over raw drawing JSON unless the JSON is the only usable source.
+
+## Copyright and Reference Models
+
+Do not recreate a company's distinctive UI, proprietary command structure, branded screens, or exact visual identity unless the user clearly has rights to that source.
+
+It is acceptable to extract general design principles:
+
+- density without clutter
+- command-first interaction
+- monochrome with one accent
+- editorial hierarchy
+- clear empty states
+- strong keyboard affordances
+
+It is not acceptable to clone proprietary layouts, copy exact branded surfaces, or reproduce copyrighted content.
+
+When using references, transform posture and principles into an original design.
+
+## Verification
+
+Before final response, verify as much as the environment allows.
+
+Minimum:
+
+- file exists at the stated path
+- HTML is saved completely
+- obvious syntax issues are checked
+
+Better:
+
+- open in a browser tool and check console errors
+- inspect screenshots at the primary viewport
+- test key interactions
+- test light/dark or variants if present
+- test responsive breakpoints if relevant
+
+If verification is limited by environment, say exactly what was and was not verified.
+
+Never say “done” if the file was not actually written.
+
+## Final Response Format
+
+Keep final responses short.
+
+Include:
+
+- artifact path
+- what it contains
+- verification status
+- next suggested action, if useful
+
+Example:
+
+```text
+Created: /path/to/Prototype.html
+It includes 3 layout variants, a Tweaks panel for density/theme, and responsive behavior.
+Verified: file exists and opened cleanly in browser, no console errors.
+Next: pick the strongest direction and I’ll tighten copy + motion.
+```
+
+## Portable Opening Prompt Pattern
+
+When adapting a Claude Design style request into CLI/API mode, use this mental translation:
+
+```text
+You are running in CLI/API mode, not hosted Claude Design. Ignore references to hosted-only tools or preview panes. Produce complete local design artifacts, usually self-contained HTML with embedded CSS/JS, and verify with available local tools before returning. Preserve the design process: gather context, define the system, produce options, avoid filler, and meet a high visual bar.
+```
+
+## Pitfalls
+
+- Do not paste hosted tool schemas into a skill. They cause fake tool calls.
+- Do not point the skill at a giant external prompt as required runtime context. That creates drift.
+- Do not strip the design doctrine while removing tool plumbing.
+- Do not over-ask when the user already gave enough direction.
+- Do not under-ask for high-fidelity work with no brand context.
+- Do not produce generic SaaS layouts and call them designed.
+- Do not claim browser verification unless it actually happened.

skills/creative/creative-ideation/SKILL.md

@@ -1,7 +1,7 @@
 ---
 name: ideation
 title: Creative Ideation — Constraint-Driven Project Generation
-description: "Generate project ideas through creative constraints. Use when the user says 'I want to build something', 'give me a project idea', 'I'm bored', 'what should I make', 'inspire me', or any variant of 'I have tools but no direction'. Works for code, art, hardware, writing, tools, and anything that can be made."
+description: "Generate project ideas via creative constraints."
 version: 1.0.0
 author: SHL0MS
 license: MIT
@@ -14,6 +14,10 @@ metadata:
 
 # Creative Ideation
 
+## When to use
+
+Use when the user says 'I want to build something', 'give me a project idea', 'I'm bored', 'what should I make', 'inspire me', or any variant of 'I have tools but no direction'. Works for code, art, hardware, writing, tools, and anything that can be made.
+
 Generate project ideas through creative constraints. Constraint + direction = creativity.
 
 ## How It Works

skills/creative/design-md/SKILL.md

@@ -1,13 +1,13 @@
 ---
 name: design-md
-description: Author, validate, diff, and export DESIGN.md files — Google's open-source format spec that gives coding agents a persistent, structured understanding of a design system (tokens + rationale in one file). Use when building a design system, porting style rules between projects, generating UI with consistent brand, or auditing accessibility/contrast.
+description: Author/validate/export Google's DESIGN.md token spec files.
 version: 1.0.0
 author: Hermes Agent
 license: MIT
 metadata:
   hermes:
     tags: [design, design-system, tokens, ui, accessibility, wcag, tailwind, dtcg, google]
-    related_skills: [popular-web-designs, excalidraw, architecture-diagram]
+    related_skills: [popular-web-designs, claude-design, excalidraw, architecture-diagram]
 ---
 
 # DESIGN.md Skill
@@ -31,7 +31,9 @@ diffs versions for regressions, and exports to Tailwind or W3C DTCG JSON.
 - User wants contrast / WCAG accessibility validation on their color palette
 
 For purely visual inspiration or layout examples, use `popular-web-designs`
-instead. This skill is for the *formal spec file* itself.
+instead. For *process and taste* when designing a one-off HTML artifact
+from scratch (prototype, deck, landing page, component lab), use
+`claude-design`. This skill is for the *formal spec file* itself.
 
 ## File anatomy
 

skills/creative/excalidraw/SKILL.md

@@ -1,6 +1,6 @@
 ---
 name: excalidraw
-description: Create hand-drawn style diagrams using Excalidraw JSON format. Generate .excalidraw files for architecture diagrams, flowcharts, sequence diagrams, concept maps, and more. Files can be opened at excalidraw.com or uploaded for shareable links.
+description: "Hand-drawn Excalidraw JSON diagrams (arch, flow, seq)."
 version: 1.0.0
 author: Hermes Agent
 license: MIT
@@ -16,6 +16,10 @@ metadata:
 
 Create diagrams by writing standard Excalidraw element JSON and saving as `.excalidraw` files. These files can be drag-and-dropped onto [excalidraw.com](https://excalidraw.com) for viewing and editing. No accounts, no API keys, no rendering libraries -- just JSON.
 
+## When to use
+
+Generate `.excalidraw` files for architecture diagrams, flowcharts, sequence diagrams, concept maps, and more. Files can be opened at excalidraw.com or uploaded for shareable links.
+
 ## Workflow
 
 1. **Load this skill** (you already did)

skills/creative/manim-video/SKILL.md

@@ -1,11 +1,15 @@
 ---
 name: manim-video
-description: "Production pipeline for mathematical and technical animations using Manim Community Edition. Creates 3Blue1Brown-style explainer videos, algorithm visualizations, equation derivations, architecture diagrams, and data stories. Use when users request: animated explanations, math animations, concept visualizations, algorithm walkthroughs, technical explainers, 3Blue1Brown style videos, or any programmatic animation with geometric/mathematical content."
+description: "Manim CE animations: 3Blue1Brown math/algo videos."
 version: 1.0.0
 ---
 
 # Manim Video Production Pipeline
 
+## When to use
+
+Use when users request: animated explanations, math animations, concept visualizations, algorithm walkthroughs, technical explainers, 3Blue1Brown style videos, or any programmatic animation with geometric/mathematical content. Creates 3Blue1Brown-style explainer videos, algorithm visualizations, equation derivations, architecture diagrams, and data stories using Manim Community Edition.
+
 ## Creative Standard
 
 This is educational cinema. Every frame teaches. Every animation reveals structure.

skills/creative/p5js/SKILL.md

@@ -1,6 +1,6 @@
 ---
 name: p5js
-description: "Production pipeline for interactive and generative visual art using p5.js. Creates browser-based sketches, generative art, data visualizations, interactive experiences, 3D scenes, audio-reactive visuals, and motion graphics — exported as HTML, PNG, GIF, MP4, or SVG. Covers: 2D/3D rendering, noise and particle systems, flow fields, shaders (GLSL), pixel manipulation, kinetic typography, WebGL scenes, audio analysis, mouse/keyboard interaction, and headless high-res export. Use when users request: p5.js sketches, creative coding, generative art, interactive visualizations, canvas animations, browser-based visual art, data viz, shader effects, or any p5.js project."
+description: "p5.js sketches: gen art, shaders, interactive, 3D."
 version: 1.0.0
 metadata:
   hermes:
@@ -10,6 +10,14 @@ metadata:
 
 # p5.js Production Pipeline
 
+## When to use
+
+Use when users request: p5.js sketches, creative coding, generative art, interactive visualizations, canvas animations, browser-based visual art, data viz, shader effects, or any p5.js project.
+
+## What's inside
+
+Production pipeline for interactive and generative visual art using p5.js. Creates browser-based sketches, generative art, data visualizations, interactive experiences, 3D scenes, audio-reactive visuals, and motion graphics — exported as HTML, PNG, GIF, MP4, or SVG. Covers: 2D/3D rendering, noise and particle systems, flow fields, shaders (GLSL), pixel manipulation, kinetic typography, WebGL scenes, audio analysis, mouse/keyboard interaction, and headless high-res export.
+
 ## Creative Standard
 
 This is visual art rendered in the browser. The canvas is the medium; the algorithm is the brush.

skills/creative/pixel-art/SKILL.md

@@ -1,6 +1,6 @@
 ---
 name: pixel-art
-description: Convert images into retro pixel art with hardware-accurate palettes (NES, Game Boy, PICO-8, C64, etc.), and animate them into short videos. Presets cover arcade, SNES, and 10+ era-correct looks. Use `clarify` to let the user pick a style before generating.
+description: "Pixel art w/ era palettes (NES, Game Boy, PICO-8)."
 version: 2.0.0
 author: dodo-reach
 license: MIT

skills/creative/popular-web-designs/SKILL.md

@@ -1,10 +1,6 @@
 ---
 name: popular-web-designs
-description: >
-  54 production-quality design systems extracted from real websites. Load a template
-  to generate HTML/CSS that matches the visual identity of sites like Stripe, Linear,
-  Vercel, Notion, Airbnb, and more. Each template includes colors, typography, components,
-  layout rules, and ready-to-use CSS values.
+description: 54 real design systems (Stripe, Linear, Vercel) as HTML/CSS.
 version: 1.0.0
 author: Hermes Agent + Teknium (design systems sourced from VoltAgent/awesome-design-md)
 license: MIT
@@ -27,6 +23,16 @@ triggers:
 site's complete visual language: color palette, typography hierarchy, component styles, spacing
 system, shadows, responsive behavior, and practical agent prompts with exact CSS values.
 
+## Related design skills
+
+- **`claude-design`** — use for the design *process and taste* (scoping a brief,
+  producing variants, verifying a local HTML artifact, avoiding AI-design slop).
+  Pair it with this skill when the user wants a thoughtfully-designed page styled
+  after a known brand: `claude-design` drives the workflow, this skill supplies
+  the visual vocabulary.
+- **`design-md`** — use when the deliverable is a formal DESIGN.md token spec
+  file, not a rendered artifact.
+
 ## How to Use
 
 1. Pick a design from the catalog below

skills/creative/songwriting-and-ai-music/SKILL.md

@@ -1,9 +1,6 @@
 ---
 name: songwriting-and-ai-music
-description: >
-  Songwriting craft, AI music generation prompts (Suno focus), parody/adaptation
-  techniques, phonetic tricks, and lessons learned. These are tools and ideas,
-  not rules. Break any of them when the art calls for it.
+description: "Songwriting craft and Suno AI music prompts."
 tags: [songwriting, music, suno, parody, lyrics, creative]
 triggers:
   - writing a song

skills/data-science/jupyter-live-kernel/SKILL.md

@@ -1,11 +1,6 @@
 ---
 name: jupyter-live-kernel
-description: >
-  Use a live Jupyter kernel for stateful, iterative Python execution via hamelnb.
-  Load this skill when the task involves exploration, iteration, or inspecting
-  intermediate results — data science, ML experimentation, API exploration, or
-  building up complex code step-by-step. Uses terminal to run CLI commands against
-  a live Jupyter kernel. No new tools required.
+description: "Iterative Python via live Jupyter kernel (hamelnb)."
 version: 1.0.0
 author: Hermes Agent
 license: MIT

skills/devops/webhook-subscriptions/SKILL.md

@@ -1,6 +1,6 @@
 ---
 name: webhook-subscriptions
-description: Create and manage webhook subscriptions for event-driven agent activation, or for direct push notifications (zero LLM cost). Use when the user wants external services to trigger agent runs OR push notifications to chats.
+description: "Webhook subscriptions: event-driven agent runs."
 version: 1.1.0
 metadata:
   hermes:

skills/dogfood/SKILL.md

@@ -1,6 +1,6 @@
 ---
 name: dogfood
-description: Systematic exploratory QA testing of web applications — find bugs, capture evidence, and generate structured reports
+description: "Exploratory QA of web apps: find bugs, evidence, reports."
 version: 1.0.0
 metadata:
   hermes:

skills/email/himalaya/SKILL.md

@@ -1,6 +1,6 @@
 ---
 name: himalaya
-description: CLI to manage emails via IMAP/SMTP. Use himalaya to list, read, write, reply, forward, search, and organize emails from the terminal. Supports multiple accounts and message composition with MML (MIME Meta Language).
+description: "Himalaya CLI: IMAP/SMTP email from terminal."
 version: 1.0.0
 author: community
 license: MIT

skills/feeds/DESCRIPTION.md

@@ -1,3 +0,0 @@
----
-description: Skills for monitoring, aggregating, and processing RSS feeds, blogs, and web content sources.
----

skills/gaming/minecraft-modpack-server/SKILL.md

@@ -1,6 +1,6 @@
 ---
 name: minecraft-modpack-server
-description: Set up a modded Minecraft server from a CurseForge/Modrinth server pack zip. Covers NeoForge/Forge install, Java version, JVM tuning, firewall, LAN config, backups, and launch scripts.
+description: "Host modded Minecraft servers (CurseForge, Modrinth)."
 tags: [minecraft, gaming, server, neoforge, forge, modpack]
 ---
 

skills/gaming/pokemon-player/SKILL.md

@@ -1,6 +1,6 @@
 ---
 name: pokemon-player
-description: Play Pokemon games autonomously via headless emulation. Starts a game server, reads structured game state from RAM, makes strategic decisions, and sends button inputs — all from the terminal.
+description: "Play Pokemon via headless emulator + RAM reads."
 tags: [gaming, pokemon, emulator, pyboy, gameplay, gameboy]
 ---
 # Pokemon Player

skills/github/codebase-inspection/SKILL.md

@@ -1,6 +1,6 @@
 ---
 name: codebase-inspection
-description: Inspect and analyze codebases using pygount for LOC counting, language breakdown, and code-vs-comment ratios. Use when asked to check lines of code, repo size, language composition, or codebase stats.
+description: "Inspect codebases w/ pygount: LOC, languages, ratios."
 version: 1.0.0
 author: Hermes Agent
 license: MIT

skills/github/github-auth/SKILL.md

@@ -1,6 +1,6 @@
 ---
 name: github-auth
-description: Set up GitHub authentication for the agent using git (universally available) or the gh CLI. Covers HTTPS tokens, SSH keys, credential helpers, and gh auth — with a detection flow to pick the right method automatically.
+description: "GitHub auth setup: HTTPS tokens, SSH keys, gh CLI login."
 version: 1.1.0
 author: Hermes Agent
 license: MIT

skills/github/github-code-review/SKILL.md

@@ -1,6 +1,6 @@
 ---
 name: github-code-review
-description: Review code changes by analyzing git diffs, leaving inline comments on PRs, and performing thorough pre-push review. Works with gh CLI or falls back to git + GitHub REST API via curl.
+description: "Review PRs: diffs, inline comments via gh or REST."
 version: 1.1.0
 author: Hermes Agent
 license: MIT

skills/github/github-issues/SKILL.md

@@ -1,6 +1,6 @@
 ---
 name: github-issues
-description: Create, manage, triage, and close GitHub issues. Search existing issues, add labels, assign people, and link to PRs. Works with gh CLI or falls back to git + GitHub REST API via curl.
+description: "Create, triage, label, assign GitHub issues via gh or REST."
 version: 1.1.0
 author: Hermes Agent
 license: MIT

skills/github/github-pr-workflow/SKILL.md

@@ -1,6 +1,6 @@
 ---
 name: github-pr-workflow
-description: Full pull request lifecycle — create branches, commit changes, open PRs, monitor CI status, auto-fix failures, and merge. Works with gh CLI or falls back to git + GitHub REST API via curl.
+description: "GitHub PR lifecycle: branch, commit, open, CI, merge."
 version: 1.1.0
 author: Hermes Agent
 license: MIT

skills/github/github-repo-management/SKILL.md

@@ -1,6 +1,6 @@
 ---
 name: github-repo-management
-description: Clone, create, fork, configure, and manage GitHub repositories. Manage remotes, secrets, releases, and workflows. Works with gh CLI or falls back to git + GitHub REST API via curl.
+description: "Clone/create/fork repos; manage remotes, releases."
 version: 1.1.0
 author: Hermes Agent
 license: MIT

skills/mcp/native-mcp/SKILL.md

@@ -1,6 +1,6 @@
 ---
 name: native-mcp
-description: Built-in MCP (Model Context Protocol) client that connects to external MCP servers, discovers their tools, and registers them as native Hermes Agent tools. Supports stdio and HTTP transports with automatic reconnection, security filtering, and zero-config tool injection.
+description: "MCP client: connect servers, register tools (stdio/HTTP)."
 version: 1.0.0
 author: Hermes Agent
 license: MIT

skills/media/gif-search/SKILL.md

@@ -1,6 +1,6 @@
 ---
 name: gif-search
-description: Search and download GIFs from Tenor using curl. No dependencies beyond curl and jq. Useful for finding reaction GIFs, creating visual content, and sending GIFs in chat.
+description: "Search/download GIFs from Tenor via curl + jq."
 version: 1.1.0
 author: Hermes Agent
 license: MIT
@@ -16,6 +16,10 @@ metadata:
 
 Search and download GIFs directly via the Tenor API using curl. No extra tools needed.
 
+## When to use
+
+Useful for finding reaction GIFs, creating visual content, and sending GIFs in chat.
+
 ## Setup
 
 Set your Tenor API key in your environment (add to `~/.hermes/.env`):

skills/media/heartmula/SKILL.md

@@ -1,6 +1,6 @@
 ---
 name: heartmula
-description: Set up and run HeartMuLa, the open-source music generation model family (Suno-like). Generates full songs from lyrics + tags with multilingual support.
+description: "HeartMuLa: Suno-like song generation from lyrics + tags."
 version: 1.0.0
 metadata:
   hermes:
@@ -11,7 +11,7 @@ metadata:
 # HeartMuLa - Open-Source Music Generation
 
 ## Overview
-HeartMuLa is a family of open-source music foundation models (Apache-2.0) that generates music conditioned on lyrics and tags. Comparable to Suno for open-source. Includes:
+HeartMuLa is a family of open-source music foundation models (Apache-2.0) that generates music conditioned on lyrics and tags, with multilingual support. Generates full songs from lyrics + tags. Comparable to Suno for open-source. Includes:
 - **HeartMuLa** - Music language model (3B/7B) for generation from lyrics + tags
 - **HeartCodec** - 12.5Hz music codec for high-fidelity audio reconstruction
 - **HeartTranscriptor** - Whisper-based lyrics transcription

skills/media/songsee/SKILL.md

@@ -1,6 +1,6 @@
 ---
 name: songsee
-description: Generate spectrograms and audio feature visualizations (mel, chroma, MFCC, tempogram, etc.) from audio files via CLI. Useful for audio analysis, music production debugging, and visual documentation.
+description: "Audio spectrograms/features (mel, chroma, MFCC) via CLI."
 version: 1.0.0
 author: community
 license: MIT

skills/media/spotify/SKILL.md

@@ -1,6 +1,6 @@
 ---
 name: spotify
-description: Control Spotify — play music, search the catalog, manage playlists and library, inspect devices and playback state. Loads when the user asks to play/pause/queue music, search tracks/albums/artists, manage playlists, or check what's playing. Assumes the Hermes Spotify toolset is enabled and `hermes auth spotify` has been run.
+description: "Spotify: play, search, queue, manage playlists and devices."
 version: 1.0.0
 author: Hermes Agent
 license: MIT

skills/media/youtube-content/SKILL.md

@@ -1,14 +1,14 @@
 ---
 name: youtube-content
-description: >
-  Fetch YouTube video transcripts and transform them into structured content
-  (chapters, summaries, threads, blog posts). Use when the user shares a YouTube
-  URL or video link, asks to summarize a video, requests a transcript, or wants
-  to extract and reformat content from any YouTube video.
+description: "YouTube transcripts to summaries, threads, blogs."
 ---
 
 # YouTube Content Tool
 
+## When to use
+
+Use when the user shares a YouTube URL or video link, asks to summarize a video, requests a transcript, or wants to extract and reformat content from any YouTube video. Transforms transcripts into structured content (chapters, summaries, threads, blog posts).
+
 Extract transcripts from YouTube videos and convert them into useful formats.
 
 ## Setup

skills/mlops/evaluation/lm-evaluation-harness/SKILL.md

@@ -1,6 +1,6 @@
 ---
 name: evaluating-llms-harness
-description: Evaluates LLMs across 60+ academic benchmarks (MMLU, HumanEval, GSM8K, TruthfulQA, HellaSwag). Use when benchmarking model quality, comparing models, reporting academic results, or tracking training progress. Industry standard used by EleutherAI, HuggingFace, and major labs. Supports HuggingFace, vLLM, APIs.
+description: "lm-eval-harness: benchmark LLMs (MMLU, GSM8K, etc.)."
 version: 1.0.0
 author: Orchestra Research
 license: MIT
@@ -13,6 +13,10 @@ metadata:
 
 # lm-evaluation-harness - LLM Benchmarking
 
+## What's inside
+
+Evaluates LLMs across 60+ academic benchmarks (MMLU, HumanEval, GSM8K, TruthfulQA, HellaSwag). Use when benchmarking model quality, comparing models, reporting academic results, or tracking training progress. Industry standard used by EleutherAI, HuggingFace, and major labs. Supports HuggingFace, vLLM, APIs.
+
 ## Quick start
 
 lm-evaluation-harness evaluates LLMs across 60+ academic benchmarks using standardized prompts and metrics.

skills/mlops/evaluation/weights-and-biases/SKILL.md

@@ -1,6 +1,6 @@
 ---
 name: weights-and-biases
-description: Track ML experiments with automatic logging, visualize training in real-time, optimize hyperparameters with sweeps, and manage model registry with W&B - collaborative MLOps platform
+description: "W&B: log ML experiments, sweeps, model registry, dashboards."
 version: 1.0.0
 author: Orchestra Research
 license: MIT

skills/mlops/huggingface-hub/SKILL.md

@@ -1,6 +1,6 @@
 ---
 name: huggingface-hub
-description: Hugging Face Hub CLI (hf) — search, download, and upload models and datasets, manage repos, query datasets with SQL, deploy inference endpoints, manage Spaces and buckets.
+description: "HuggingFace hf CLI: search/download/upload models, datasets."
 version: 1.0.0
 author: Hugging Face
 license: MIT

skills/mlops/inference/obliteratus/SKILL.md

@@ -1,6 +1,6 @@
 ---
 name: obliteratus
-description: Remove refusal behaviors from open-weight LLMs using OBLITERATUS — mechanistic interpretability techniques (diff-in-means, SVD, whitened SVD, LEACE, SAE decomposition, etc.) to excise guardrails while preserving reasoning. 9 CLI methods, 28 analysis modules, 116 model presets across 5 compute tiers, tournament evaluation, and telemetry-driven recommendations. Use when a user wants to uncensor, abliterate, or remove refusal from an LLM.
+description: "OBLITERATUS: abliterate LLM refusals (diff-in-means)."
 version: 2.0.0
 author: Hermes Agent
 license: MIT
@@ -13,6 +13,10 @@ metadata:
 
 # OBLITERATUS Skill
 
+## What's inside
+
+9 CLI methods, 28 analysis modules, 116 model presets across 5 compute tiers, tournament evaluation, and telemetry-driven recommendations.
+
 Remove refusal behaviors (guardrails) from open-weight LLMs without retraining or fine-tuning. Uses mechanistic interpretability techniques — including diff-in-means, SVD, whitened SVD, LEACE concept erasure, SAE decomposition, Bayesian kernel projection, and more — to identify and surgically excise refusal directions from model weights while preserving reasoning capabilities.
 
 **License warning:** OBLITERATUS is AGPL-3.0. NEVER import it as a Python library. Always invoke via CLI (`obliteratus` command) or subprocess. This keeps Hermes Agent's MIT license clean.