Question

Hello 👋!

I have an agent that has multiple tools available. It's an agent with no deps and it returns a string (conversational).

Rather than statically checking for a specific answer or using a LLMJudge, I'd like to evaluate whether the agent used a specific tool during the process. For this I do not care about the agent output.
Is it possible with the current implementation of Evals?

Thanks in advance!

Additional Context

No response