How to specify the next agent？

[huangbhan]

generate_sql_agent = Agent(
        name="generate sql Agent",
        instructions='''
            1. You are responsible for generating SQL statements based on user requests..
           '''
        handoffs=[sql_checker_agent],
    )

 sql_checker_agent = Agent(
        name="SQL Checker Agent",
        handoff_description="Check the SQL statement generated based on the user's request.",
        instructions='''You are responsible for checking the SQL statement generated based on the user's request.
        ''',
        tools=[task_query],
    )

I am worried that handoffs may not be triggered. I need to ensure that sql_checker_agent is called after generate_sql_agent.
In the documentation(https://openai.github.io/openai-agents-python/multi_agent/) I found this: "Chaining multiple agents by transforming the output of one into the input of the next. You can decompose a task like writing a blog post into a series of steps - do research, write an outline, write the blog post, critique it, and then improve it."
But how to implement it with code?

[DanieleMorotti]

Hi, if I'm not wrong this example from the github repo may be what you're looking for.
It first plans searches, then it actually performs them and at the end it writes a report.

[tameverfit]

hi @DanieleMorotti, do you have any advice on ensuring the agent can be handed off correctly when the chat history is too long?

[DanieleMorotti]

Hi, I believe this issue can be addressed using standard chat techniques.
It might be useful to summarize the conversation history to reduce token count. Perhaps you could implement an agent that automatically condenses the history once it exceeds a certain threshold.

These are just my opinions, so take them with a grain of salt :)

[huangbhan]

Yes, that's exactly what I want. Do you have any plans to add some code examples like this to the documentation? I think this will help people in need more quickly.

Hi, if I'm not wrong this example from the github repo may be what you're looking for.It first plans searches, then it actually performs them and at the end it writes a report.

[DanieleMorotti]

@rm-openai may answer to this. But if you're looking for other examples, there are some in the examples directory.