Return

[ChrisDelClea]

Please read this first

• Have you read the docs? yes
• Have you searched for related issues? yes

Question

During streaming a response with :

result = Runner.run_streamed(starting_agent=agent, input=inputs)
async for event in result.stream_events():
data = event.data

i found multiple Delta and Event objects from openai.types.responses that can be used e.g.:

 ResponseFunctionCallArgumentsDeltaEvent, ResponseFunctionCallArgumentsDoneEvent, ResponseFunctionToolCall

Yet, after the run when i do:

        for msg in result.to_input_list():
            print(msg)

(ToolCallOutputItem)

i get a entry:

{'call_id': 'call_xxxx', 'output': "{'assistant': 'Assistent'}", 'type': 'function_call_output'}

So my question is, how can i capture the function_call_output via event during streaming?

[DanieleMorotti]

Hi, I think this example in the documentation may be what you're looking for. Hope it's helpful!

[ChrisDelClea]

Great, thank you.

[ChrisDelClea]

But well, what actually would be great too: if i can output the result from the called function tool directly. Does that make sense?

[rm-openai]

But well, what actually would be great too: if i can output the result from the called function tool directly. Does that make sense?

Does this code from the example not work?

        # When items are generated, print them
        elif event.type == "run_item_stream_event":
            elif event.item.type == "tool_call_output_item":
                print(f"-- Tool output: {event.item.output}"

[ChrisDelClea]

Thanks, @rm-openai . It works. Just another question:
During handoff i can see some sort of reasoning generated, getting it with:

            if event.type == "raw_response_event" and isinstance(event.data, ResponseFunctionCallArgumentsDeltaEvent):
                parsed_json = json.loads(event.data.delta)
                status_queue.put(f"- {parsed_json.get('message', parsed_json)}")

e.g.
Transferring you to the Service Agent to help you.
However, this will also output all other function call arguments .
Do you know if there's a way to only output the message from the handoff function call?
I assume what i see here as message is the reasoning from the llm after the handoff function call right?