How it works
From input to result in 3 steps
1
Paste
Paste an agent description and a sample trace.
2
Score
Get quality, faithfulness, and task-success scores plus a regression flag.
3
Review
Compare scores across runs; humans own the release gate.
Was this result helpful?Feedback