Make the AI judge laugh
My AI agent writes code, tests, and docs - all three disagree with each other., Sent through the API