How to log inputs and outputs for LLM node

Agents
1

This question originally came up in our Slack community and the thread has been consolidated here for long-term reference.

We’re trying to simultaneously log both the inputs and outputs to the STT, LLM, and TTS nodes.

We were able to view the output chunks from the LLM, but haven’t found a way to log inputs to the LLM node as well.

2

Override the llm_node in your agent to log both inputs and outputs:

async def llm_node(self, chat_ctx, tools, model_settings):
    # Log the input chat context
    logger.info(f"LLM Input: {chat_ctx.to_dict()}")
    
    # Call the default implementation and stream outputs
    async for chunk in Agent.default.llm_node(self, chat_ctx, tools, model_settings):
        # Log each output chunk
        logger.info(f"LLM Output chunk: {chunk}")
        yield chunk

For STT and TTS:

  • Subscribe to user_input_transcribed to log user transcriptions as they arrive
  • Use conversation_item_added to capture committed user/agent messages
  • Override transcription_node for text forwarding with timestamps

See the nodes documentation: