@magarcia/gitingest NPM

Git Ingest

Turn any Git repository into a simple text digest of its codebase.

This is useful for feeding a codebase into any LLM.

Inspired by gitingest.com

Features

Extracts text content from all files in a git repository
Generates a tree structure of the repository
Respects .gitignore rules
Supports additional ignore patterns via --ignore flag
Automatically detects and skips binary files
Provides clear file separation in output
Supports multiple repository input formats:
- Local directory
- Git URLs (HTTPS and SSH)
- GitHub web URLs (with branch support)
- GitLab web URLs (with branch support)
Optional clipboard support
Built-in help with examples (--help)
Cross-platform support (Windows, macOS, Linux)
Written in TypeScript with full type safety

Requirements

Node.js >= 12.0.0
Git installed and available in PATH

Installation

npm install -g @magarcia/gitingest

Usage

Once installed, the tool can be executed as git ingest.

Basic Usage

# Show help and available options
git ingest --help

# Process current directory
git ingest

# Process specific local directory
git ingest /path/to/repo

# Copy output to clipboard instead of printing
git ingest -c
git ingest --copy

# Ignore specific files or patterns
git ingest --ignore "*.log"
git ingest --ignore "temp/*" --ignore "docs/*.md"

Remote Repositories

The tool supports various repository URL formats:

# GitHub URLs
git ingest https://github.com/username/repo
git ingest https://github.com/username/repo/tree/branch

# GitLab URLs
git ingest https://gitlab.com/username/repo
git ingest https://gitlab.com/username/repo/-/tree/branch

# Direct Git URLs
git ingest git@github.com:username/repo.git
git ingest https://github.com/username/repo.git

# With ignore patterns
git ingest https://github.com/username/repo --ignore "*.test.js" --ignore "dist/*"

When using a remote repository URL: 1. The repository is cloned to a temporary directory 2. Content is extracted 3. Temporary directory is automatically cleaned up

Ignore Patterns

The tool supports ignoring files and directories using glob patterns:

# Ignore all log files
git ingest --ignore "*.log"

# Ignore multiple patterns
git ingest --ignore "temp/*" --ignore "docs/*.md"

# Ignore test files in any directory
git ingest --ignore "**/*.test.js"

Ignore patterns:

Can be specified multiple times with --ignore
Use the same glob syntax as .gitignore
Are applied in addition to existing .gitignore rules
Affect both content extraction and tree structure

Output Format

The tool generates two main sections: 1. Repository Tree Structure: A visual representation of the repository's file structure 2. Repository Content: The actual content of text files

Example output:

Repository Tree Structure:
├── src
│   ├── index.ts
│   └── utils.ts
├── package.json
└── README.md

Repository Content:
File: src/index.ts
[content...]

File: src/utils.ts
[content...]

API

The package also exports functions that you can use in your own code:

import { extractRepositoryContent } from '@magarcia/gitingest';

// Extract content from a local repository
const { content, tree } = extractRepositoryContent('/path/to/repo');
console.log('Tree structure:', tree);
console.log('Content:', content);

Functions

`extractRepositoryContent(repoPath: string): { content: string, tree: string }`

Extracts text content and tree structure from a local git repository.

repoPath: Path to the local git repository
Returns an object with:
- content: Concatenated text content from all files
- tree: Visual tree structure of the repository