What problem does it solve? Post-incident reviews often end with vague conclusions like "someone forgot" and no lasting fixes. This Skill walks you through a disciplined Correction of Errors (COE) process that forces evidence-based 5-Whys analysis, structural root causes, and deterministic preventive mechanisms, producing a publishable COE document. ## Core Features & Use Cases - Structured 10-Phase Workflow: From pattern scan and severity classification through timeline building, 5-Whys analysis, corrective actions, and preventive mechanisms, ending in a saved COE document. - Evidence-Based Root Cause Analysis: Every Why requires a citation (file path, commit hash, log entry, or user-reported), and behavioral root causes are rejected in favor of structural ones. - Repeat Pattern Detection: Scans prior COEs and git history to flag recurring failure classes before analysis begins. - Use Case: After discovering that standing orders were dispatched to the wrong agent for two weeks, run the COE workflow to classify severity, reconstruct the timeline from git history, derive the structural root cause, and generate a COE document with deterministic preventive mechanisms and metrics to watch. ## Quick Start Ask the AI to run a COE on a recent incident, for example: run a COE on the bug where customer invoices were sent twice last Friday.
