What problem does it solve? Renames and removals often leave stray references behind, and self-reported "we got everything" claims are unreliable. This Skill provides a deterministic, repeatable sweep that verifies whether old class names, retired tooling vocabulary, deprecated env vars, or renamed fields actually disappeared from every file. ## Core Features & Use Cases - Parallel deterministic scanning: A dependency-free Node script uses worker_threads to scan every non-binary file across CPU cores, producing identical sorted output (file, line, column) on every rerun. - Boundary matching for short terms: The --boundary flag applies word-boundary matching plus a PascalCase segment variant, so a term like id catches IdToken without matching inside hidden. - Sensible exclusions: Skips build output, lockfiles, worktrees, and binary assets by default, with per-run --exclude paths for directories expected to cite retired names. - Use Case: After renaming a service class across a monorepo, run the scanner with the old name to confirm zero remaining references before calling the cleanup complete. ## Quick Start Ask the AI to run the scan-vocabulary script against the repo root with your retired terms, for example to verify that every reference to an old service name has been removed.
