agent.run_stream doesn't run with tools

[ryx2]

Initial Checks

• I confirm that I'm using the latest version of Pydantic AI
• I confirm that I searched for my issue in https://github.com/pydantic/pydantic-ai/issues before opening this issue

Description

When I use agent.run, it loops through the tools until completion, however run_stream's implementation is that it just returns after one iter. Wouldn't it make more sense for the default behavior to be do the same thing as agent.run does, and loop through tool executions until completion?

Example Code

Instead of including my own code, I think I can see the problematic area of logic in the run_stream implementation:

final_result = await stream_to_final(streamed_response)
if final_result is not None:
    yield StreamedRunResult(...)
    break  # ← Exits early, preventing tool processing steps

Python, Pydantic AI & LLM client version

0.4.6

[pydanticai-bot]

PydanticAI Github Bot Found 1 issues similar to this one:

1. "tools work with agent.run() but do not work with agent.run_stream() #957" (90% similar)

[tianshangwuyun]

I also found similar problems, unstable

[DouweM]

@ryx2 run_stream is a bit inflexible in that it will run the agent loop silently until it sees something that looks like final output that you'd want to stream (like text, if you've not specified output_type or if str is included as an option), even if tool calls may still follow and this wasn't the "real" final output, as it has no way of knowing that at that point, and waiting to stream until we've received the entire response would defeat the point.

Continuing the run with tool calls and all after text has already been streamed is also problematic, because run_stream/stream_text has no way of communicating that you should discard the originally streamed "final result" because the run was not actually done yet.

If you want the ability to stream and make sure the entire agent run is completed including tool calls, we recommend using agent.iter() (https://ai.pydantic.dev/agents/#streaming), or the new event_stream_handler (#1977) I've been working on in #2306.

Does that work for your use case?

The request_stream in this snippet is an AgentStream that also has stream_text and stream_output methods:

if Agent.is_model_request_node(node):
    async with node.stream(run.ctx) as request_stream: