fix(gateway): stop typing indicator before sending response to prevent perpetual typing

agent/auxiliary_client.py

@@ -191,9 +191,9 @@ def _normalize_aux_provider(provider: Optional[str]) -> str:
 
 
 def _is_kimi_model(model: Optional[str]) -> bool:
-    """True for any Kimi / Moonshot model that manages temperature server-side."""
+    """True for any Kimi / Moonshot / Qwen model that manages temperature server-side."""
     bare = (model or "").strip().lower().rsplit("/", 1)[-1]
-    return bare.startswith("kimi-") or bare == "kimi"
+    return bare.startswith("kimi-") or bare == "kimi" or bare.startswith("qwen")
 
 
 def _fixed_temperature_for_model(
@@ -1005,19 +1005,19 @@ def _read_nous_auth() -> Optional[dict]:
     """
     pool_present, entry = _select_pool_entry("nous")
     if pool_present:
-        if entry is None:
-            return None
-        return {
-            "access_token": getattr(entry, "access_token", ""),
-            "refresh_token": getattr(entry, "refresh_token", None),
-            "agent_key": getattr(entry, "agent_key", None),
-            "inference_base_url": _pool_runtime_base_url(entry, _NOUS_DEFAULT_BASE_URL),
-            "portal_base_url": getattr(entry, "portal_base_url", None),
-            "client_id": getattr(entry, "client_id", None),
-            "scope": getattr(entry, "scope", None),
-            "token_type": getattr(entry, "token_type", "Bearer"),
-            "source": "pool",
-        }
+        if entry is not None:
+            return {
+                "access_token": getattr(entry, "access_token", ""),
+                "refresh_token": getattr(entry, "refresh_token", None),
+                "agent_key": getattr(entry, "agent_key", None),
+                "inference_base_url": _pool_runtime_base_url(entry, _NOUS_DEFAULT_BASE_URL),
+                "portal_base_url": getattr(entry, "portal_base_url", None),
+                "client_id": getattr(entry, "client_id", None),
+                "scope": getattr(entry, "scope", None),
+                "token_type": getattr(entry, "token_type", "Bearer"),
+                "source": "pool",
+            }
+        # entry is None → pool exists but all exhausted, fall through to auth.json
 
     try:
         if not _AUTH_JSON_PATH.is_file():

cli.py

@@ -4817,7 +4817,7 @@ def _list_recent_sessions(self, limit: int = 10) -> list[dict[str, Any]]:
         try:
             sessions = self._session_db.list_sessions_rich(
                 source="cli",
-                exclude_sources=["tool"],
+                exclude_sources=["tool", "cron"],
                 limit=limit,
             )
         except Exception:
@@ -5415,6 +5415,7 @@ def _open_model_picker(self, providers: list, current_model: str, current_provid
             "stage": "provider",
             "providers": providers,
             "selected": default_idx,
+            "scroll_offset": 0,
             "current_model": current_model,
             "current_provider": current_provider,
             "user_provs": user_provs,
@@ -5565,6 +5566,7 @@ def _handle_model_picker_selection(self, persist_global: bool = False) -> None:
             state["provider_data"] = provider_data
             state["model_list"] = model_list
             state["selected"] = 0
+            state["scroll_offset"] = 0
             self._invalidate(min_interval=0.0)
             return
         if stage == "model":
@@ -5575,6 +5577,7 @@ def _handle_model_picker_selection(self, persist_global: bool = False) -> None:
             if selected == back_idx:
                 state["stage"] = "provider"
                 state["selected"] = next((i for i, p in enumerate(state.get("providers") or []) if p.get("slug") == provider_data.get("slug")), 0)
+                state["scroll_offset"] = 0
                 self._invalidate(min_interval=0.0)
                 return
             if selected >= cancel_idx:
@@ -10271,11 +10274,16 @@ def approval_down(event):
                 self._approval_state["selected"] = min(max_idx, self._approval_state["selected"] + 1)
                 event.app.invalidate()
 
-        # --- /model picker: arrow-key navigation ---
+        # --- /model picker: arrow-key navigation with scroll ---
         @kb.add('up', filter=Condition(lambda: bool(self._model_picker_state)))
         def model_picker_up(event):
-            if self._model_picker_state:
-                self._model_picker_state["selected"] = max(0, self._model_picker_state.get("selected", 0) - 1)
+            state = self._model_picker_state
+            if state:
+                new_selected = max(0, state.get("selected", 0) - 1)
+                state["selected"] = new_selected
+                # Keep cursor within scroll window
+                if new_selected < state.get("scroll_offset", 0):
+                    state["scroll_offset"] = new_selected
                 event.app.invalidate()
 
         @kb.add('down', filter=Condition(lambda: bool(self._model_picker_state)))

cron/scheduler.py

@@ -16,6 +16,7 @@
 import os
 import subprocess
 import sys
+import time
 
 # fcntl is Unix-only; on Windows use msvcrt for file locking
 try:
@@ -40,6 +41,10 @@
 
 logger = logging.getLogger(__name__)
 
+# Delivery retry configuration for transient network/platform failures.
+_MAX_DELIVERY_RETRIES = 3
+_DELIVERY_RETRY_BACKOFF = [10, 30, 60]  # seconds between attempts
+
 
 def _resolve_cron_enabled_toolsets(job: dict, cfg: dict) -> list[str] | None:
     """Resolve the toolset list for a cron job.
@@ -475,33 +480,46 @@ def _deliver_result(job: dict, content: str, adapters=None, loop=None) -> Option
                 )
 
         if not delivered:
-            # Standalone path: run the async send in a fresh event loop (safe from any thread)
-            coro = _send_to_platform(platform, pconfig, chat_id, cleaned_delivery_content, thread_id=thread_id, media_files=media_files)
-            try:
-                result = asyncio.run(coro)
-            except RuntimeError:
-                # asyncio.run() checks for a running loop before awaiting the coroutine;
-                # when it raises, the original coro was never started — close it to
-                # prevent "coroutine was never awaited" RuntimeWarning, then retry in a
-                # fresh thread that has no running loop.
-                coro.close()
-                with concurrent.futures.ThreadPoolExecutor(max_workers=1) as pool:
-                    future = pool.submit(asyncio.run, _send_to_platform(platform, pconfig, chat_id, cleaned_delivery_content, thread_id=thread_id, media_files=media_files))
-                    result = future.result(timeout=30)
-            except Exception as e:
-                msg = f"delivery to {platform_name}:{chat_id} failed: {e}"
-                logger.error("Job '%s': %s", job["id"], msg)
-                delivery_errors.append(msg)
-                continue
-
-            if result and result.get("error"):
-                msg = f"delivery error: {result['error']}"
-                logger.error("Job '%s': %s", job["id"], msg)
-                delivery_errors.append(msg)
+            # Standalone path with retry: attempt delivery up to _MAX_DELIVERY_RETRIES
+            # times on transient failures (network errors, platform disconnects).
+            last_error = None
+            for attempt in range(_MAX_DELIVERY_RETRIES):
+                if attempt > 0:
+                    backoff = _DELIVERY_RETRY_BACKOFF[min(attempt - 1, len(_DELIVERY_RETRY_BACKOFF) - 1)]
+                    logger.info("Job '%s': delivery retry %d/%d to %s:%s after %ds",
+                                job["id"], attempt + 1, _MAX_DELIVERY_RETRIES,
+                                platform_name, chat_id, backoff)
+                    time.sleep(backoff)
+
+                coro = _send_to_platform(platform, pconfig, chat_id, cleaned_delivery_content, thread_id=thread_id, media_files=media_files)
+                try:
+                    result = asyncio.run(coro)
+                except RuntimeError:
+                    coro.close()
+                    with concurrent.futures.ThreadPoolExecutor(max_workers=1) as pool:
+                        future = pool.submit(asyncio.run, _send_to_platform(platform, pconfig, chat_id, cleaned_delivery_content, thread_id=thread_id, media_files=media_files))
+                        result = future.result(timeout=30)
+                except Exception as e:
+                    last_error = f"delivery to {platform_name}:{chat_id} failed (attempt {attempt + 1}): {e}"
+                    logger.warning("Job '%s': %s", job["id"], last_error)
+                    continue  # retry
+
+                if result and result.get("error"):
+                    last_error = f"delivery error (attempt {attempt + 1}): {result['error']}"
+                    logger.warning("Job '%s': %s", job["id"], last_error)
+                    continue  # retry
+
+                # Success
+                logger.info("Job '%s': delivered to %s:%s (attempt %d)", job["id"], platform_name, chat_id, attempt + 1)
+                last_error = None
+                break
+
+            if last_error:
+                logger.error("Job '%s': giving up delivery to %s:%s after %d attempts: %s",
+                            job["id"], platform_name, chat_id, _MAX_DELIVERY_RETRIES, last_error)
+                delivery_errors.append(last_error)
                 continue
 
-            logger.info("Job '%s': delivered to %s:%s", job["id"], platform_name, chat_id)
-
     if delivery_errors:
         return "; ".join(delivery_errors)
     return None

gateway/platforms/base.py

@@ -2752,6 +2752,17 @@ async def _stop_typing_task() -> None:
             if not response:
                 logger.debug("[%s] Handler returned empty/None response for %s", self.name, event.source.chat_id)
             if response:
+                # Agent is done — stop the typing indicator *before* sending.
+                # If we wait until the finally block, _keep_typing keeps
+                # refreshing sendChatAction("typing") during the entire
+                # media-extraction / TTS / message-sending phase below.
+                # Telegram's client then toggles: incoming message clears
+                # typing, but _keep_typing's next tick re-triggers it 2s
+                # later, making it look like the bot is perpetually "typing".
+                # Stop here so the outgoing message naturally clears the
+                # indicator per Telegram's documented behavior.
+                await _stop_typing_task()
+
                 # Extract MEDIA:<path> tags (from TTS tool) before other processing
                 media_files, response = self.extract_media(response)
 

hermes_cli/callbacks.py

@@ -22,6 +22,21 @@ def clarify_callback(cli, question, choices):
     responds. Returns the user's choice or a timeout message.
     """
     from cli import CLI_CONFIG
+    import subprocess
+
+    # Fire macOS notification + terminal bell so user notices even if terminal is in background
+    notification_body = question[:120] + ("..." if len(question) > 120 else "")
+    try:
+        subprocess.run(
+            ["osascript", "-e", f'display notification "{notification_body}" with title "Hermes — 需要确认" sound name "Glass"'],
+            timeout=3, capture_output=True,
+        )
+    except Exception:
+        pass
+    try:
+        subprocess.run(["echo", "-e", "\\a"], timeout=2, capture_output=True)
+    except Exception:
+        pass
 
     timeout = CLI_CONFIG.get("clarify", {}).get("timeout", 120)
     response_queue = queue.Queue()
@@ -201,6 +216,22 @@ def approval_callback(cli, command: str, description: str) -> str:
 
     with lock:
         from cli import CLI_CONFIG
+        import subprocess
+
+        # Fire macOS notification + terminal bell for dangerous command approvals
+        notification_body = (description or command)[:120] + ("..." if len(description or command) > 120 else "")
+        try:
+            subprocess.run(
+                ["osascript", "-e", f'display notification "{notification_body}" with title "Hermes — 命令待批准" sound name "Alert"'],
+                timeout=3, capture_output=True,
+            )
+        except Exception:
+            pass
+        try:
+            subprocess.run(["echo", "-e", "\\a"], timeout=2, capture_output=True)
+        except Exception:
+            pass
+
         timeout = CLI_CONFIG.get("approvals", {}).get("timeout", 60)
         response_queue = queue.Queue()
         choices = ["once", "session", "always", "deny"]

hermes_cli/commands.py

@@ -92,6 +92,8 @@ class CommandDef:
                gateway_only=True),
     CommandDef("background", "Run a prompt in the background", "Session",
                aliases=("bg", "btw"), args_hint="<prompt>"),
+    CommandDef("codex", "Interactive Codex CLI session (persistent context)", "Session",
+               aliases=("cx",), args_hint="[start|kill|log|<message>]"),
     CommandDef("agents", "Show active agents and running tasks", "Session",
                aliases=("tasks",)),
     CommandDef("queue", "Queue a prompt for the next turn (doesn't interrupt)", "Session",

hermes_state.py

@@ -33,7 +33,7 @@
 
 DEFAULT_DB_PATH = get_hermes_home() / "state.db"
 
-SCHEMA_VERSION = 11
+SCHEMA_VERSION = 12
 
 SCHEMA_SQL = """
 CREATE TABLE IF NOT EXISTS schema_version (
@@ -102,25 +102,31 @@
 
 FTS_SQL = """
 CREATE VIRTUAL TABLE IF NOT EXISTS messages_fts USING fts5(
-    content
+    content,
+    tool_name,
+    tool_calls,
+    content='messages',
+    content_rowid='id'
 );
 
 CREATE TRIGGER IF NOT EXISTS messages_fts_insert AFTER INSERT ON messages BEGIN
-    INSERT INTO messages_fts(rowid, content) VALUES (
-        new.id,
-        COALESCE(new.content, '') || ' ' || COALESCE(new.tool_name, '') || ' ' || COALESCE(new.tool_calls, '')
+    INSERT INTO messages_fts(rowid, content, tool_name, tool_calls) VALUES (
+        new.id, new.content, new.tool_name, new.tool_calls
     );
 END;
 
 CREATE TRIGGER IF NOT EXISTS messages_fts_delete AFTER DELETE ON messages BEGIN
-    DELETE FROM messages_fts WHERE rowid = old.id;
+    INSERT INTO messages_fts(messages_fts, rowid, content, tool_name, tool_calls) VALUES (
+        'delete', old.id, old.content, old.tool_name, old.tool_calls
+    );
 END;
 
 CREATE TRIGGER IF NOT EXISTS messages_fts_update AFTER UPDATE ON messages BEGIN
-    DELETE FROM messages_fts WHERE rowid = old.id;
-    INSERT INTO messages_fts(rowid, content) VALUES (
-        new.id,
-        COALESCE(new.content, '') || ' ' || COALESCE(new.tool_name, '') || ' ' || COALESCE(new.tool_calls, '')
+    INSERT INTO messages_fts(messages_fts, rowid, content, tool_name, tool_calls) VALUES (
+        'delete', old.id, old.content, old.tool_name, old.tool_calls
+    );
+    INSERT INTO messages_fts(rowid, content, tool_name, tool_calls) VALUES (
+        new.id, new.content, new.tool_name, new.tool_calls
     );
 END;
 """
@@ -132,25 +138,31 @@
 FTS_TRIGRAM_SQL = """
 CREATE VIRTUAL TABLE IF NOT EXISTS messages_fts_trigram USING fts5(
     content,
+    tool_name,
+    tool_calls,
+    content='messages',
+    content_rowid='id',
     tokenize='trigram'
 );
 
 CREATE TRIGGER IF NOT EXISTS messages_fts_trigram_insert AFTER INSERT ON messages BEGIN
-    INSERT INTO messages_fts_trigram(rowid, content) VALUES (
-        new.id,
-        COALESCE(new.content, '') || ' ' || COALESCE(new.tool_name, '') || ' ' || COALESCE(new.tool_calls, '')
+    INSERT INTO messages_fts_trigram(rowid, content, tool_name, tool_calls) VALUES (
+        new.id, new.content, new.tool_name, new.tool_calls
     );
 END;
 
 CREATE TRIGGER IF NOT EXISTS messages_fts_trigram_delete AFTER DELETE ON messages BEGIN
-    DELETE FROM messages_fts_trigram WHERE rowid = old.id;
+    INSERT INTO messages_fts_trigram(messages_fts_trigram, rowid, content, tool_name, tool_calls) VALUES (
+        'delete', old.id, old.content, old.tool_name, old.tool_calls
+    );
 END;
 
 CREATE TRIGGER IF NOT EXISTS messages_fts_trigram_update AFTER UPDATE ON messages BEGIN
-    DELETE FROM messages_fts_trigram WHERE rowid = old.id;
-    INSERT INTO messages_fts_trigram(rowid, content) VALUES (
-        new.id,
-        COALESCE(new.content, '') || ' ' || COALESCE(new.tool_name, '') || ' ' || COALESCE(new.tool_calls, '')
+    INSERT INTO messages_fts_trigram(messages_fts_trigram, rowid, content, tool_name, tool_calls) VALUES (
+        'delete', old.id, old.content, old.tool_name, old.tool_calls
+    );
+    INSERT INTO messages_fts_trigram(rowid, content, tool_name, tool_calls) VALUES (
+        new.id, new.content, new.tool_name, new.tool_calls
     );
 END;
 """
@@ -481,6 +493,42 @@ def _init_schema(self):
                     "COALESCE(tool_calls, '') "
                     "FROM messages"
                 )
+            if current_version < 12:
+                # v12: switch from inline FTS5 to external-content FTS5.
+                # The v11 inline mode duplicated all content in FTS tables,
+                # causing ~1.2GB state.db where messages are only ~270MB.
+                # External-content mode keeps only index data in FTS tables;
+                # content is read from the messages table on demand.
+                # All 3 indexed columns (content, tool_name, tool_calls) are
+                # preserved. Estimated ~33% DB size reduction.
+                for _trig in (
+                    "messages_fts_insert",
+                    "messages_fts_delete",
+                    "messages_fts_update",
+                    "messages_fts_trigram_insert",
+                    "messages_fts_trigram_delete",
+                    "messages_fts_trigram_update",
+                ):
+                    try:
+                        cursor.execute(f"DROP TRIGGER IF EXISTS {_trig}")
+                    except sqlite3.OperationalError:
+                        pass
+                for _tbl in ("messages_fts", "messages_fts_trigram"):
+                    try:
+                        cursor.execute(f"DROP TABLE IF EXISTS {_tbl}")
+                    except sqlite3.OperationalError:
+                        pass
+                cursor.executescript(FTS_SQL)
+                cursor.executescript(FTS_TRIGRAM_SQL)
+                # Rebuild both FTS indexes from the messages table.
+                # External-content mode (content='messages', content_rowid='id')
+                # reads all column values from messages automatically.
+                cursor.execute(
+                    "INSERT INTO messages_fts(messages_fts) VALUES('rebuild')"
+                )
+                cursor.execute(
+                    "INSERT INTO messages_fts_trigram(messages_fts_trigram) VALUES('rebuild')"
+                )
             if current_version < SCHEMA_VERSION:
                 cursor.execute(
                     "UPDATE schema_version SET version = ?",