Triage agent can not delegate task to handoff agent

[eccozhou]

Question

I want to output which agent is used for reponse, but it seems not working.

It seems triage_agent is not delegate task to any handoff agent, he output the response by himself.

Code

class HomeworkOutput(BaseModel):
    is_homework: bool
    reasoning: str

guardrail_agent = Agent(
    name="Guardrail check",
    instructions="Check if the user is asking about homework.",
    output_type=HomeworkOutput,
)

math_tutor_agent = Agent(
    name="Math Tutor",
    handoff_description="Specialist agent for math questions",
    instructions="You provide help with math problems. Explain your reasoning at each step and include examples",
)

history_tutor_agent = Agent(
    name="History Tutor",
    handoff_description="Specialist agent for historical questions",
    instructions="You provide assistance with historical queries. Explain important events and context clearly.",
)

def on_handoff(ctx: RunContextWrapper):
    print("Math handoff triggered")

math_handoff = handoff(
    agent=math_tutor_agent,
    on_handoff=on_handoff,
)

triage_agent = Agent(
    name="Triage Agent",
    instructions="You determine which agent to use based on the user's homework question",
    handoffs=[history_tutor_agent, math_handoff],
)

async def main():
    result = await Runner.run(triage_agent, "who was the first president of the united states?")
    print(result.final_output)

if __name__ == "__main__":
    asyncio.run(main())

[rm-openai]

Can you try using the recommended prompts? that may help

https://openai.github.io/openai-agents-python/handoffs/#recommended-prompts

[eccozhou]

It's still not working.

[rm-openai]

It works fine for me, see script below. I think the problem may be your on_handoff - it is only provided to the math_handoff, but the question is a history question.

import asyncio

from pydantic import BaseModel

from agents import Agent, RunContextWrapper, Runner, handoff
from agents.extensions.handoff_prompt import prompt_with_handoff_instructions


class HomeworkOutput(BaseModel):
    is_homework: bool
    reasoning: str


guardrail_agent = Agent(
    name="Guardrail check",
    instructions="Check if the user is asking about homework.",
    output_type=HomeworkOutput,
)

math_tutor_agent = Agent(
    name="Math Tutor",
    handoff_description="Specialist agent for math questions",
    instructions=prompt_with_handoff_instructions(
        "You provide help with math problems. Explain your reasoning at each step and include examples"
    ),
)

history_tutor_agent = Agent(
    name="History Tutor",
    handoff_description="Specialist agent for historical questions",
    instructions=prompt_with_handoff_instructions(
        "You provide assistance with historical queries. Explain important events and context clearly."
    ),
)


def on_handoff(ctx: RunContextWrapper):
    print("Math handoff triggered")


math_handoff = handoff(
    agent=math_tutor_agent,
    on_handoff=on_handoff,
)

triage_agent = Agent(
    name="Triage Agent",
    instructions=prompt_with_handoff_instructions(
        "You determine which agent to use based on the user's homework question"
    ),
    handoffs=[history_tutor_agent, math_handoff],
)


async def main():
    result = await Runner.run(triage_agent, "who was the first president of the united states?")
    print(result.last_agent.name)
    print(result.final_output)


if __name__ == "__main__":
    asyncio.run(main())

Output

History Tutor
The first president of the United States was George Washington. He served from 1789 to 1797 and is often called the "Father of His Country" for his pivotal role in the founding of the nation. Washington set many important precedents for the national government and the presidency in particular.

[nvenkat94]

Hi @eccozhou ,

Could you try result.last_agent.name instead of result.final_output. This will print the last used agent.

[github-actions]

This issue is stale because it has been open for 7 days with no activity.

[github-actions]

This issue was closed because it has been inactive for 3 days since being marked as stale.