run-retrospective

What problem does it solve?

Post-plan knowledge and process gaps often vanish after a project closes, causing repeated mistakes, documentation drift, and lost learnings; this Skill centralizes evidence and turns plan artifacts into prioritized, actionable lessons and minimal doc fixes so teams capture value from the work they've already done.

Core Features & Use Cases

• Evidence collection: Reads PROGRESS.md, story execution logs, QA run results, architecture docs, commit history, and current SKILL.md files to build a complete timeline of what happened.
• Pattern identification: Categorizes findings into documentation drift, struggle patterns, agent autonomy gaps, and skill gaps, then prioritizes by impact and likelihood to recur.
• Documentation & proposals: Applies factual doc fixes, drafts topic-based lessons in docs/, and prepares proposed skill and CLAUDE.md changes for user approval; commits approved changes and reports the summary to the team.
• Use case: Run after a plan's QA completes to produce a short, prioritized retrospective, clean up inaccurate docs, and propose targeted skill improvements.

Quick Start

Run the retrospective on plan-0-infrastructure after QA completes to generate prioritized lessons, factual doc fixes, and proposed skill changes for approval.