what's the purpose of MAX_TOOL_STEPS ? explain in detail

Thanks for reaching out, Purna. I'm an AI assistant and I'm here to help. Give me a moment while I research this topic.

I have a function call which takes 30 seconds to finish, now the agent currently executed this function_call. The problem is when the user speaks anything when the function_call is running, the agent will call the function again. how can I prevent that?

To prevent the agent from calling a function again while it is still executing, you can use the

max_nested_fnc_calls

parameter. This parameter controls the maximum depth of nested function calls allowed. When the nested function depth reaches or exceeds the configured maximum depth, the function context is dropped, and additional function calls are prevented. For your scenario, you can set

max_nested_fnc_calls

to a lower value or even to 0 to completely disable nested function calls. This ensures that the agent will not call the function again while it is still running. Sources: livekit.agents.pipeline.pipeline_agent API documentation

Thanks, will test it out.

You're welcome! Feel free to reach out if you have any further questions or need assistance. Good luck with your testing!

is max_nested_fnc_calls present in V1

Yes, the

max_nested_fnc_calls

parameter is available in V1. It is part of the

VoicePipelineAgent

class constructor and controls the maximum depth of nested function calls allowed. By default, its value is set to 1. Sources: livekit.agents.pipeline API documentation