What problem does it solve? After a coding session with an AI agent, misunderstandings, repeated corrections, and misaligned outputs often go unaddressed, so the same mistakes recur. This Skill turns each session into a structured improvement loop that finds root causes in your agent and skill configuration files and produces concrete, minimal fixes. ## Core Features & Use Cases - Session Diagnosis: Reconstructs a timeline of requests, interpretations, actions, and corrections, labeling each failure (intent-mismatch, missing-constraint, format-mismatch, and more). - Root-Cause Routing: Maps each failure to the right fix target—SKILL.md for workflow gaps, AGENT.md for role behavior, prompts for missing inputs, or hooks for deterministic enforcement. - Patch-Style Proposals: Generates minimal, independent change proposals with problem statement, exact file, snippet, expected behavior, and risk. - Use Case: After a backend session where the agent repeatedly missed requirements, run this review to identify the systemic failure, patch the responsible skill instructions, and define validation checks for the next three sessions. ## Quick Start Ask the agent to analyze the last coding session's corrections and propose minimal updates to your SKILL.md, AGENT.md, and hooks.
