What problem does it solve? After finishing significant work, valuable insights like gotchas, patterns, and architectural rationale are easily lost. This Skill provides a decision framework for determining what is worth documenting, where each type of knowledge belongs, and how to format it so future developers and AI sessions benefit. ## Core Features & Use Cases - Documentation Triage: Decides whether a learning belongs in the project CLAUDE.md, an ADR in docs/adr/, an active plan file, auto-memory, or the README based on its type and audience. - Learning Classification: Distinguishes gotchas, patterns, anti-patterns, architectural decisions, edge cases, and tool knowledge, with explicit criteria for what is worth recording versus what the repo already shows. - Standard Format: Provides a scannable Context/Issue/Solution template with correct and wrong code examples so entries can be grasped in under ten seconds. - Use Case: After debugging a subtle API behaviour where an endpoint returns null instead of an empty array, use this Skill to record the gotcha in the project CLAUDE.md with context, issue, and solution before the session ends. ## Quick Start Ask the assistant to document this learning or record this gotcha from the work just completed, and it will place the entry in the right documentation file.
