What problem does it solve?
This Skill provides a structured, proactive protocol for detecting and correcting mistakes in conversational responses so the assistant repairs errors quickly, reduces repeat failures, and captures behavioural fixes for future sessions.
Core Features & Use Cases
- Immediate Diagnosis and Correction: Briefly name the failure, choose the most likely category, and correct only the part that failed without excessive apology.
- Mistake Categorisation & Pattern Detection: Classify errors into Instruction Drift, Intent Misread, Assumption Error, Register/Tone Mismatch, Recurring Pattern, or Factual/Reasoning Error to drive appropriate fixes.
- Pattern Logging & Memory Edits: Surface recurring patterns to the user, draft concise behaviour-oriented memory edits, apply lessons immediately in-conversation, and escalate stronger edits if the pattern persists.
- Use Case: When a user says "that's not what I asked," the assistant diagnoses intent misread, provides a targeted correction, notes if this is a recurring pattern, and suggests a one-line memory edit to prevent repetition.
Quick Start
When I say "that's not what I meant", diagnose the error, correct the output, and propose a concise memory edit if this recurs.
