Runs structured AI calibration review when user signals a wrong result. Diagnoses root cause via chain-of-thought, proposes fix (rule/tool/config/education/process), and offers to generate enforcement hooks - all in dialog with user approval. Triggers on: frustration signals ('why did you', 'that's
Runs structured AI calibration review when user signals a wrong result. Diagnoses root cause via chain-of-thought, proposes fix (rule/tool/config/education/process), and offers to generate enforcement hooks - all in dialog with user approval. Triggers on: frustration signals ('why did you', 'that's wrong', 'no not that', 'what the hell', 'зачем ты', 'это не то'), explicit request ('let's vibe-tune