Feature Request

[PKQ1688]

When I run an agent, the tool is a python function written by myself. If an error is reported inside, I hope to be able to obtain the error content outside, and then handle it in different ways. I How to implement it through code.

[dirkbrnd]

@PKQ1688 I made a note to work on this.
Can you possibly give me a code sample of what you are trying to do? Just so I can make sure I find the best solution.

[PKQ1688]

try:
     answer_id_res = self.exec_agent.run(question)
 except Exception as e:
     logger.error(f'Error in exec_agent, {e}')
     logger.error(f'Error in exec_agent: {question}')
     pass

maybe like this? I'm not sure either.
The main reason is that when calling the API in the tool, there may be some problems with the API itself, and the current agent is easy to keep retrying. I hope to get the error and then execute my workflow without being affected by a certain agent.

[dirkbrnd]

Yes that makes sense. Thanks, I'll look into it.