What problem does it solve? AI coding agents start every session from zero, so hard-won debugging insights and workarounds are lost when the session ends. This Skill captures non-obvious solutions discovered during work and saves them as new reusable skills that surface automatically in future sessions. ## Core Features & Use Cases - Automatic Skill Extraction: Evaluates completed tasks for extractable knowledge (debugging techniques, workarounds, project-specific patterns) and writes new SKILL.md files with retrieval-optimized descriptions. - Retrospective Mode: Run /retrospective at the end of a session to review the conversation, prioritize the highest-value discoveries, and extract 1-3 new skills. - Quality Gates & Deduplication: Verifies solutions actually worked, checks existing skills for near-matches, and confirms with a dry-run preview before writing anything. - Use Case: After spending an hour diagnosing a Prisma "too many connections" error in a serverless deployment, the Skill saves a prisma-connection-pool-exhaustion skill with exact trigger errors and the verified fix, so the next occurrence is resolved in minutes. ## Quick Start Ask the agent to run a retrospective and save what we learned from this debugging session as a reusable skill.
