fix: avoid Agent SDK race on greeting + clear session on abort

1. Change on_enter greeting from generate_reply() to session.say() with
   a static message — avoids spawning an Agent SDK task just for a greeting,
   which caused a race condition when the user speaks before it completes.

2. Clear agent session ID when receiving abort/exit errors so the next
   task starts a fresh session instead of trying to resume a dead process.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

packages/services/voice-agent/src/agent.py

@@ -93,10 +93,14 @@ class IT0VoiceAgent(Agent):
         )
 
     async def on_enter(self):
-        """Called when the agent becomes active — greet the user."""
-        self.session.generate_reply(
-            instructions="用一句简短的话打招呼，告诉用户你是IT0运维助手，可以帮助什么。"
-        )
+        """Called when the agent becomes active — greet the user.
+
+        Uses session.say() with a static message instead of generate_reply()
+        to avoid triggering the Agent SDK / LLM pipeline for a simple greeting.
+        This prevents a race condition when the user speaks before the
+        greeting LLM task completes.
+        """
+        self.session.say("你好，我是IT0运维助手，有什么可以帮你的？")
 
 
 # ---------------------------------------------------------------------------

packages/services/voice-agent/src/plugins/agent_llm.py

@@ -331,6 +331,11 @@ class AgentServiceLLMStream(llm.LLMStream):
                     elif evt_type == "error":
                         err_msg = evt_data.get("message", "Unknown error")
                         logger.error("Agent error: %s", err_msg)
+                        # Clear session so next task starts fresh
+                        # (don't try to resume a dead/aborted session)
+                        if "aborted" in err_msg.lower() or "exited" in err_msg.lower():
+                            logger.info("Clearing agent session after abort/exit")
+                            self._llm_instance._agent_session_id = None
                         self._event_ch.send_nowait(
                             llm.ChatChunk(
                                 id=request_id,