Turn any GitHub repo into an LLM prompt with one word swap

What it does

This tool takes a Git repository URL or a local directory and converts it into a single, structured text digest meant for LLM consumption. It outputs a file tree, file contents, and metadata including token count and extract size, so you know exactly how much context you are about to feed a model. The result is a flat dump that replaces the usual ritual of manually copying files or wrestling with raw tarballs.

The interesting bit

The cleverest entry point is a dead-simple URL hack: swap hub for ingest in any GitHub URL and you land on a rendered digest of that repo. Beyond the gimmick, the tool is packaged as a web service, CLI, Python library, and browser extension, which suggests the authors understand that LLM context preparation is a workflow problem, not just a formatting problem.

Key highlights

• Returns a clean three-part structure: summary, directory tree, and file contents
• Calculates token counts and total size before you paste into a model
• Respects .gitignore by default and supports private repositories via GitHub PAT
• Available as a web app, CLI tool, Python package, and browser extensions for Chrome, Firefox, and Edge
• Self-hostable via Docker with optional Prometheus metrics and Sentry error tracking

Verdict

If you regularly ask LLMs to review, refactor, or explain entire repositories, this saves you from writing yet another ad-hoc file concatenation script. If your LLM workflow never exceeds a single file or snippet, you probably do not need it.