What problem does it solve? Lessons learned during code review and retry cycles usually stay buried in PR comments, so agents repeat the same mistakes. This Skill turns resolved review/retry loops and other readable inputs into generalized, durable knowledge written into each agent's own persistent memory. ## Core Features & Use Cases - Memory writing by the agent itself: Dispatches target agents as subagents so memory writes happen under their own memory: permission, then verifies what was actually created. - Target agent resolution: Derives which agents an input concerns from the task, the review record, or project and plugin agent definitions, checking memory: and write-tool prerequisites before dispatch. - Consolidation and verification: Enforces memory conventions (MEMORY.md index, typed topic files, Why/How to apply lines) and reports created, updated, deleted, skipped, and non-conforming entries. - Use Case: After a review round on issue #42 found a bug and a retry fixed it, run the Skill so the reviewing agent records a generalized pitfall rule in its memory instead of leaving the lesson in a PR comment. ## Quick Start Ask the assistant to have the tech-lead agent record the lesson learned from the review loop on issue #42 into its memory.
