gitingest: Turn Git Repositories into LLM-Ready Text

Overview

Gitingest turns a Git repository or local directory into a text digest that is easier to pass to a large language model than a collection of individual files. The digest includes repository structure, file contents, and statistics such as size and estimated token count.

It is useful when you want to give an LLM a codebase as context for questions, analysis, or development tasks. You can use the hosted service, install the Python package, or self-host the web application.

Key Features

• Ingest repositories from GitHub URLs, including links to subdirectories, or from local directories.
• Produce an LLM-oriented text digest with a file tree and source contents.
• Report extract size and token count.
• Provide command-line, synchronous Python, and asynchronous Python interfaces.
• Support private GitHub repositories with a personal access token.
• Allow submodules and Git-ignored files to be included through options.
• Offer a hosted web app and browser extensions, as well as Docker-based self-hosting.

Use Cases

• Developers can provide an LLM with a repository snapshot when asking for explanations or codebase-level help.
• Teams can prepare a compact text representation of a project for prompt-based review or analysis.
• Python users can incorporate repository ingestion into notebooks or other scripts.
• Users who need to keep processing under their control can run the web application themselves with Docker.

Project Facts

• Language: Python
• License: MIT
• Stars: 15.8k
• Forks: 1.2k
• Topics: ai, code, developer-tool, ingestion
• Archived: false

Getting Started

Install the package and ingest a repository from the command line:

pip install gitingest
gitingest https://github.com/coderamp-labs/gitingest

For Python usage, private repositories, and self-hosting instructions, see the repository README.

Considerations

• Requires Python 3.8 or newer for package use.
• Accessing private GitHub repositories requires a GitHub personal access token.
• Files matched by .gitignore are skipped by default; opt in if they should appear in the digest.
• A digest includes repository file contents, so review what you share with an external LLM or service.
• Token count is an estimate intended to help size prompt context, not a guarantee that a particular model will accept the digest.