#testing

Design a rigorous, platform-aware evaluation for an AI agent - define what good looks like, pick the right grading method, build a test set, and turn results into a defensible go/no-go decision.