How to track finish_reason when using the Runner class

[paulthemagno]

Hi! I'm trying to figure out how to track the finish_reason of my agents when using the Runner class.
Before switching to this repo, I was monitoring the finish_reason from each completion and logging them to one of my services. Is there a similar way to do this here?

I noticed these two documentation links about exceptions:

• https://openai.github.io/openai-agents-python/running_agents/#exceptions
• https://openai.github.io/openai-agents-python/ref/exceptions/

However, I couldn’t find any specific point where finish_reason is handled.
Are they automatically managed in some way by the Agents?

Thank you

[seratch]

When running an Agent, the agent loop automatically decides what to do next until the final output is populated. The finish_reason property in Chat Completions API is not exposed as part of the final response because the output data is compatible with Responses API format, which does not return "finish_reason" property. If there is a specific use case you're trying to implement with Agents SDK, sharing more details would be helpful for us to better understand the need.