Hey guys I am using a Mcp server for my tools in livekit agent, I am facing a problem where after user confirms their name LLM is calling a tool to create_appointment, but the problem is if user is saying name multiple times and speech is getting transcribed, llm is calling tool multiple times which is fine but livekit is discarding the first tool call result and call in context which is causing problem because llm can only see response of second tool call, has anybody faced this issue and solved it, it would be really helpful.

You might want to take a look at tasks, https://docs.livekit.io/agents/build/tasks/, this lets you retrieve the user's name before moving onto the create_appointment stage in your workflow.