What problem does it solve? After structural or architectural changes, a repository's docs, skills, and README often silently fall out of sync with reality — referencing deleted scripts, moved files, or stale workflows that only fail when someone follows them. This Skill systematically verifies every self-description in the repo against how the code actually works. ## Core Features & Use Cases - Execution-based verification: Runs the commands and recipes docs prescribe (npm scripts, validator chains, multi-step workflows) rather than just reading them, catching instructions that would break on the next run. - Mechanical cross-checks: Greps for referenced commands, file paths, import specifiers, internal links, and hardcoded counts, then diffs them against package.json scripts, the exports map, and registries. - Skills self-audit: Reviews every SKILL.md for broken commands, stale paths, duplicated registry facts, and missing coverage of repeated workflows. - Use Case: After migrating a build step or renaming a route, run the audit to get a severity-grouped report (broken, stale, gaps, verified) with file paths, line numbers, and concrete fixes before the drift misleads contributors or consumers. ## Quick Start Ask the AI to run a drift audit on this repository to check whether the docs and skills are still up to date.
