Refactor WebSocket authentication handling by removing auth requirements from the hello message. Update related documentation and schemas to reflect the changes in authentication strategy, simplifying the connection process.

engine/core/session.py

@@ -124,7 +124,6 @@ class Session:
         self.ws_state = WsSessionState.WAIT_HELLO
         self._pipeline_started = False
         self.protocol_version: Optional[str] = None
-        self.authenticated: bool = False
 
         # Track IDs
         self.current_track_id: str = self.TRACK_CONTROL
@@ -264,7 +263,7 @@ class Session:
             await self._send_error("client", f"Unsupported message type: {msg_type}", "protocol.unsupported")
 
     async def _handle_hello(self, message: HelloMessage) -> None:
-        """Handle initial hello/auth/version negotiation."""
+        """Handle initial hello/version negotiation."""
         if self.ws_state != WsSessionState.WAIT_HELLO:
             await self._send_error("client", "Duplicate hello", "protocol.order")
             return
@@ -279,23 +278,6 @@ class Session:
             self.ws_state = WsSessionState.STOPPED
             return
 
-        auth_payload = message.auth
-        api_key = auth_payload.apiKey if auth_payload else None
-        jwt = auth_payload.jwt if auth_payload else None
-
-        if settings.ws_api_key:
-            if api_key != settings.ws_api_key:
-                await self._send_error("auth", "Invalid API key", "auth.invalid_api_key")
-                await self.transport.close()
-                self.ws_state = WsSessionState.STOPPED
-                return
-        elif settings.ws_require_auth and not (api_key or jwt):
-            await self._send_error("auth", "Authentication required", "auth.required")
-            await self.transport.close()
-            self.ws_state = WsSessionState.STOPPED
-            return
-
-        self.authenticated = True
         self.protocol_version = message.version
         self.ws_state = WsSessionState.WAIT_START
         await self._send_event(
@@ -701,8 +683,6 @@ class Session:
             return self.TRACK_AUDIO_IN
         if stage in {"llm", "tts", "tool"}:
             return self.TRACK_AUDIO_OUT
-        if str(code or "").strip().lower().startswith("auth."):
-            return self.TRACK_CONTROL
         return self.TRACK_CONTROL
 
     def _envelope_event(self, event: Dict[str, Any]) -> Dict[str, Any]: