[sanderjd]

Maybe. But that sounds like a large amount of bespoke work for what seems like a common problem?

[manojlds]

I was talking about enterprise agents and then realized the question is more about coding agents.

[sanderjd]

Ah I see! Yes, I was talking about a coding harness, not an enterprise agent. I entirely agree with you that your suggestion of driving it via evals is the right thing for that use case!