Scanning and caching

What gets scanned

Inside a git repository, hiarky scans only files git would show: tracked files plus untracked files that are not ignored. Generated clients and build output never reach a snapshot, whatever they are called.

On top of that, these are always skipped: node_modules, dist, build, out, .next, coverage, generated, .turbo, vendor, .hiarky, minified bundles and declaration files.

Test files are not skipped. They are part of the review.

The analysis cache

Analysing a range of commits re-reads mostly identical trees, so each file's analysis is cached by content under .hiarky/cache/. Fingerprints come from a single git ls-files -s call instead of re-reading the project. Only files git reports as modified are hashed directly.

On a nine-commit --per-commit review of a real monorepo, the cache cut the run from 4.8s to 2.3s.

Pass --no-cache to snap, review or backfill to bypass it, or delete .hiarky/cache/. It is derived data and has its own .gitignore. Neither the cache nor the snapshots count as working-tree changes, so a snapshot taken on a clean checkout is still recorded as clean.