I understand the issue you're facing happens because the

agent-starter-python

is designed to handle just one session lifecycle per room. So, when a user leaves and then comes back, the agent doesn’t automatically restart or respond, even though it’s still listening. That can definitely be a bit confusing! To help you work through this, here are some steps you can try: • Dynamic Participant Handling: You can use

ctx.wait_for_participant()

along with room event listeners like

participant_connected

to detect when a user rejoins and then trigger a response from the agent. • Session State Tracking: It’s a good idea to keep track of session data—like chat history or agent settings—using a

SessionManager

pattern. This helps maintain continuity when users reconnect. • Restart Agent on Rejoin: When you detect a user has rejoined, manually restarting the agent session or generating a reply can make the experience smoother. • Frontend Coordination: Make sure your frontend lets the backend know when a user reconnects, so the agent can be reactivated appropriately. These tweaks should make your agent much better at handling users who leave and come back. If you’d like, I’m here to help you implement any of these—just let me know! Sources: Web and mobile frontends | LiveKit Docs | agent/main.py | Voice Assistants | livekit/livekit_composite | DeepWiki | templates.yaml | agent-examples/agent_examples/pipeline_agent.py