Is there a way to access eval results directly in an agent’s response when using the OpenAI Agent SDK - similar to how Azure Foundry Agent SDK supports evaluations like task adherence, intent resolution, fluency, etc.?
I’m looking for samples or guidance on integrating such quality metrics into the agent’s behavior loop. The current OpenAI evals docs only cover model-level evaluation, not integration with the agent SDK runtime. Any pointers would be appreciated.