What problem does it solve? Skill documentation drifts out of sync with the real codebase as code evolves, leaving stale file paths, renamed classes, and dead API references that mislead the next developer or agent. This Skill keeps SKILL.md files accurate by auditing references against the actual code and synchronizing them after every feature change. ## Core Features & Use Cases - Create: Traverse any codebase directory (language-agnostic), analyze module dependencies, and generate an anti-corruption SKILL.md with executable detection rules and good/bad examples. - Audit & Sync: Extract every file path, class, and function reference from an existing skill and verify each one exists in the code with grep and test -f, then fix or remove broken references. - Post-Feature Sync: After a PR or commit, classify the git diff (new files, renames, signature changes, new dependencies) and update affected SKILL.md files and references/ docs in the same PR. - Reflection Loop: After using a skill, convert lessons learned and pitfalls into structured good_eg/bad_eg entries so the skill accumulates project experience. - Use Case: You just merged a branch that renamed class UserModel to class Account and moved auth/login.py. Run this Skill to find every skill referencing the old names, sed-replace them, verify all paths exist, and commit the doc sync alongside the code. ## Quick Start Ask the AI to audit the skills under .claude/skills against the current codebase and fix any references that no longer exist.
