These Evals were written to validate an AIP agent behavior. Test cases are like below and parallelization was set to 1. Still, they are running in different sessions. It’s not a deal breaker as such but nice to see them in one session.
Hi @RajKarri,
It is currently not possible out-of-the-box.
You can however achieve this by wrapping the published agent function in a typescript function where the session rid can be persisted across multiple calls to the agent. And the evaluation suite would be set up on this typescript function (docs, see “Code-authored functions”).
Let me know if you encounter any issues with this approach.
1 Like
This works for my scenario. Thanks.