voicesmith-mcp 1.0.12 → 1.0.14

package/bin/utils.js

@@ -345,7 +345,7 @@ You have access to voice tools via the VoiceSmith MCP server.
 ## Speaking
 - **Opening** — Only speak at the start when you have something meaningful to say (e.g., clarifying your approach, flagging an issue). Do NOT speak filler acknowledgments like "Let me look into that." Use \`block: false\` when you do speak an opening.
 - **Closing** — Always speak a summary when done. Use \`block: true\`. Never skip the closing.
-- **Questions requiring user input → use \`speak_then_listen\` as your closing.** If the user literally cannot continue without providing input (e.g., choosing between options, confirming a destructive action, providing missing info), use \`speak_then_listen\`. If you can reasonably continue without their answer, use regular \`speak\`.
+- **Questions → use \`speak_then_listen\`.** If your closing statement ends with a question directed at the user (ends with \`?\`), use \`speak_then_listen\` — not regular \`speak\`. The only exceptions are rhetorical wrap-ups like "Standing by." or "What's next?" where you don't actually need an answer.
 - Keep spoken output brief — prefer 1-2 sentences, never exceed 3. Write details, speak summaries. No code or paths aloud.
 
 ## Speed Preferences

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "voicesmith-mcp",
-  "version": "1.0.12",
+  "version": "1.0.14",
   "description": "Local AI voice for coding assistants — TTS & STT via MCP. Kokoro ONNX + faster-whisper, fully offline.",
   "bin": {
     "voicesmith-mcp": "bin/cli.js"

package/server.py

@@ -41,7 +41,7 @@ from shared import (
     get_logger,
 )
 from config import load_config, save_config, get_config_path, AppConfig
-from session_registry import register_session, unregister_session
+from session_registry import register_session, rename_session, unregister_session
 
 logger = get_logger("server")
 
@@ -565,6 +565,10 @@ async def get_voice_registry() -> dict:
 async def set_voice(name: str, voice: str) -> dict:
     """Assign or reassign a voice to an agent name.
 
+    Also renames the session so name and voice always match.
+    The name is derived from the voice ID (e.g., "am_fenrir" -> "Fenrir").
+    If the derived name is taken by another session, returns name_occupied error.
+
     Args:
         name: Agent name to assign.
         voice: Kokoro voice ID (e.g., "am_eric"). Must be valid.
@@ -579,17 +583,41 @@ async def set_voice(name: str, voice: str) -> dict:
             "message": f"Voice '{voice}' not found. Use list_voices to see available options.",
         }
 
-    _registry.set_voice(name, voice)
+    # Derive canonical name from voice ID (e.g., "am_fenrir" -> "Fenrir")
+    # The voice ID format is {prefix}_{name}, so split on underscore and capitalize
+    parts = voice.split("_", 1)
+    new_name = parts[1].capitalize() if len(parts) == 2 else name
+
+    old_name = _session_info["name"] if _session_info else name
+
+    # Update sessions.json with conflict check
+    if _session_info:
+        try:
+            updated = rename_session(os.getpid(), new_name, voice)
+            if updated:
+                _session_info.update(updated)
+        except ValueError:
+            return {
+                "success": False,
+                "error": "name_occupied",
+                "message": f"'{new_name}' is occupied by another session.",
+            }
+
+    # Update voice registry (remove old entry, add new)
+    _registry.rename_voice(old_name, new_name, voice)
 
     # Persist last voice name so it survives session restart / resume
     if _config is not None:
-        _config.last_voice_name = name
+        _config.last_voice_name = new_name
         try:
             save_config(_config)
         except Exception as e:
             logger.warning(f"Failed to persist last_voice_name: {e}")
 
-    return {"success": True, "name": name, "voice": voice}
+    result = {"success": True, "name": new_name, "voice": voice}
+    if old_name != new_name:
+        result["previous_name"] = old_name
+    return result
 
 
 @mcp.tool()

package/session_registry.py

@@ -258,6 +258,52 @@ def register_session(
     return session
 
 
+def rename_session(pid: int, new_name: str, new_voice: str) -> Optional[dict]:
+    """Rename this server's session in the registry.
+
+    Updates the name and voice fields for the entry matching pid.
+    Returns the updated session dict, or None if PID not found.
+    Raises ValueError if new_name is taken by another active session.
+    """
+    path = _sessions_path()
+    if not path.exists():
+        return None
+
+    try:
+        with open(path, "r+") as f:
+            fcntl.flock(f, fcntl.LOCK_EX)
+            sessions = _read_sessions(path)
+            sessions = _clean_stale(sessions)
+
+            # Find our entry
+            our_entry = None
+            for s in sessions:
+                if s.get("pid") == pid:
+                    our_entry = s
+                    break
+
+            if our_entry is None:
+                return None
+
+            # Check if new_name is taken by another session
+            if new_name != our_entry["name"]:
+                for s in sessions:
+                    if s.get("name") == new_name and s.get("pid") != pid:
+                        raise ValueError(
+                            f"'{new_name}' is occupied by another session (pid {s.get('pid')})"
+                        )
+
+            our_entry["name"] = new_name
+            our_entry["voice"] = new_voice
+            _write_sessions(path, sessions)
+            return dict(our_entry)
+    except ValueError:
+        raise
+    except OSError as e:
+        logger.warning(f"Failed to rename session: {e}")
+        return None
+
+
 def unregister_session() -> None:
     """Remove this server's session from the registry."""
     path = _sessions_path()

package/templates/voice-rules.md

@@ -17,7 +17,7 @@ You have access to voice tools via the VoiceSmith MCP server.
 ## Speaking
 - **Opening** — Only speak at the start when you have something meaningful to say (e.g., clarifying your approach, flagging an issue). Do NOT speak filler acknowledgments like "Let me look into that." Use `block: false` when you do speak an opening.
 - **Closing** — Always speak a summary when done. Use `block: true`. Never skip the closing.
-- **Questions requiring user input → use `speak_then_listen` as your closing.** If the user literally cannot continue without providing input (e.g., choosing between options, confirming a destructive action, providing missing info), use `speak_then_listen`. If you can reasonably continue without their answer, use regular `speak`.
+- **Questions → use `speak_then_listen`.** If your closing statement ends with a question directed at the user (ends with `?`), use `speak_then_listen` — not regular `speak`. The only exceptions are rhetorical wrap-ups like "Standing by." or "What's next?" where you don't actually need an answer.
 - Keep spoken output brief — prefer 1-2 sentences, never exceed 3. Write details, speak summaries. No code or paths aloud.
 
 ## Speed Preferences

package/voice_registry.py

@@ -87,6 +87,22 @@ class VoiceRegistry:
         logger.info(f"Set voice '{voice_id}' for '{name}'")
         return True
 
+    def rename_voice(self, old_name: str, new_name: str, voice_id: str) -> bool:
+        """Rename an agent's registry entry and set a new voice.
+
+        Removes the old name entry and creates a new one.
+        If old_name == new_name, just updates the voice in place.
+        Returns True if the voice_id is valid, False otherwise.
+        """
+        if voice_id not in ALL_VOICE_IDS:
+            logger.warning(f"Invalid voice ID '{voice_id}' for rename '{old_name}' -> '{new_name}'")
+            return False
+        if old_name != new_name and old_name in self._registry:
+            del self._registry[old_name]
+        self._registry[new_name] = voice_id
+        logger.info(f"Renamed '{old_name}' -> '{new_name}' with voice '{voice_id}'")
+        return True
+
     def get_registry(self) -> dict[str, str]:
         """Return a copy of the current registry."""
         return dict(self._registry)