Correction Learning Framework

When This Activates

This skill activates when you detect the user is correcting your output:

• Explicit disagreement
• Preference expressions
• Style corrections
• Terminology corrections

Correction Detection Patterns

Explicit Corrections:

• "no, I meant..."
• "that's wrong/incorrect/not right"
• "actually, I want/need/meant..."
• "I said X not Y"
• "no, not X"