rgitweb

A fully static, client-side Git repository browser — a cgit replacement with no server-side rendering at all.

Point it at the URL of a Git repository hosted as static files (the dumb-HTTP layout maintained by git update-server-info) and it browses branches, tags, history, commits with diffs, trees, and files — entirely in the browser.

How it works

The app fetches exactly what a dumb-HTTP git clone would read, but lazily: small text files (info/refs, HEAD, objects/info/packs) whole, loose objects individually, and pack files never in full — the .idx index locates an object and an HTTP Range request fetches just that object's bytes from the .pack. Packs are immutable, so the browser HTTP cache absorbs repeat reads. Deltified objects are resolved recursively client-side.

Hosting requirements for browsed repositories

• A bare repository served as static files, with git update-server-info run after each update (the post-update sample hook does this).
• CORS headers allowing the browsing origin, exposing Accept-Ranges and Content-Range.
• Range request support (standard on almost every static host); without it the app falls back to whole-file fetches.

Development

npm install
npm run dev       # local dev server
npm test          # Jest; fixtures are built with the real git CLI
npm run lint:fix  # ESLint (includes formatting)
npm run build     # static site in dist/

See AGENTS.md for architecture notes and conventions.

License

Apache-2.0