Most developers test their code

But do you test (aka evaluate) your agents?

Comments