Repomix - Pack repositories into AI-friendly files

Language: TypeScript (93.1%), Vue (4.7%), JavaScript (1.1%)
License: MIT
Stars: ~25,600
Owner: yamadashy
Repo: https://github.com/yamadashy/repomix
Version: 1.14.0
Node.js: >=22.0.0

Core Dependencies:
- commander - CLI framework
- @clack/prompts - Interactive CLI prompts
- gpt-tokenizer - Token counting for LLMs
- @modelcontextprotocol/sdk - MCP server integration
- globby - File pattern matching
- iconv-lite - Character encoding conversion
- jschardet - Character encoding detection
- valibot & zod - Schema validation
- secretlint - Security scanning

Dev Tools:
- vitest - Testing framework
- biome - Fast linter and formatter
- typescript - Type checking

# Run without installation
npx repomix@latest

# Pack a specific directory
npx repomix@latest path/to/directory

# Process a remote repository
npx repomix@latest --remote yamadashy/repomix

# NPM
npm install -g repomix

# Yarn
yarn global add repomix

# Bun
bun add -g repomix

# Verify installation
repomix --version

# Install
brew install repomix

# Verify
repomix --version

# Run in current directory
docker run -v .:/app ghcr.io/yamadashy/repomix

# Create an alias for convenience
alias repomix='docker run -v .:/app ghcr.io/yamadashy/repomix'

# Use the alias
repomix --version

# Pack current directory (creates repomix-output.xml)
repomix

# Pack specific directory
repomix path/to/directory

# Pack and copy to clipboard
repomix --copy

# Pack to specific output file
repomix --output my-codebase.xml

# XML format (default, best for Claude)
repomix --style xml

# Markdown format (human-readable)
repomix --style markdown

# Plain text format
repomix --style plain

# JSON format (programmatic use)
repomix --style json

# Include specific patterns
repomix --include "src/**/*.ts,**/*.md"

# Exclude specific patterns
repomix --ignore "**/*.log,tmp/,*.test.ts"

# Combine include and exclude
repomix --include "src/**/*.ts" --ignore "**/*.test.ts"

# Create .repomixignore file for project-specific exclusions
echo "dist/" >> .repomixignore
echo "node_modules/" >> .repomixignore

# Process a public GitHub repository
repomix --remote yamadashy/repomix

# Process specific branch
repomix --remote yamadashy/repomix --remote-branch main

# Process with remote config (use cautiously)
repomix --remote yamadashy/repomix --remote-trust-config

# Enable compression
repomix --compress

# Combine with other options
repomix --compress --style markdown --output compressed-output.md

# Token counts are included in output by default
repomix

# The output file includes:
# - Total token count
# - Per-file token counts
# - File and character statistics

# Security check is enabled by default
repomix

# Disable security check (not recommended)
repomix --no-security-check

# Security warnings appear in the output
# Review and remove sensitive data before sharing with AI

{
  "output": {
    "filePath": "repomix-output.xml",
    "style": "xml",
    "compress": false,
    "splitOutput": null,
    "removeComments": false,
    "removeEmptyLines": false,
    "topFilesLength": 5,
    "showLineNumbers": false,
    "copyToClipboard": false,
    "git": {
      "includeLogs": false,
      "logCount": 50,
      "includeDiff": false
    }
  },
  "ignore": {
    "useGitignore": true,
    "useDefaultPatterns": true,
    "customPatterns": [
      "*.log",
      "tmp/**",
      "**/*.test.ts"
    ]
  },
  "security": {
    "enableSecurityCheck": true
  },
  "include": ["src/**/*", "**/*.md"],
  "maxFileSize": 52428800
}

# Split by size (e.g., 1mb, 500kb, 2mb)
repomix --split-output 1mb

# Creates multiple files:
# - repomix-output-part1.xml
# - repomix-output-part2.xml
# - etc.

# Include last 50 commit logs (default)
repomix --git-logs

# Include specific number of logs
repomix --git-logs 100

# Include diffs
repomix --git-diff

# Combine logs and diffs
repomix --git-logs --git-diff

# Install the MCP server
npx @repomix/mcp-server

# Configure in Claude Desktop config (~/Library/Application Support/Claude/claude_desktop_config.json):
# {
#   "mcpServers": {
#     "repomix": {
#       "command": "npx",
#       "args": ["-y", "@repomix/mcp-server"]
#     }
#   }
# }

# Available MCP tools:
# - pack_repository: Pack local repositories
# - pack_remote_repository: Pack GitHub repos
# - search_repository: Search through packed outputs
# - read_file: Read files with security scanning

name: Pack Repository
on: [push]

jobs:
  pack:
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@v4
      
      - name: Setup Node.js
        uses: actions/setup-node@v4
        with:
          node-version: '22'
      
      - name: Pack repository
        run: |
          npx repomix@latest --output packed-repo.xml
      
      - name: Upload artifact
        uses: actions/upload-artifact@v4
        with:
          name: packed-repository
          path: packed-repo.xml

import { pack } from 'repomix';

// Pack a repository
const result = await pack({
  input: './my-project',
  output: {
    filePath: 'output.xml',
    style: 'xml',
    compress: false,
  },
  ignore: {
    customPatterns: ['*.log', 'tmp/**'],
  },
});

console.log(`Packed ${result.fileCount} files`);
console.log(`Total tokens: ${result.tokenCount}`);

Chrome Extension:
https://chrome.google.com/webstore (search for "Repomix")

Firefox Add-on:
https://addons.mozilla.org (search for "Repomix")

Usage:
1. Navigate to any GitHub repository
2. Click the Repomix extension icon
3. Choose format and options
4. Download or copy the packed output

Installation:
1. Open VS Code
2. Go to Extensions (Cmd+Shift+X / Ctrl+Shift+X)
3. Search for "Repomix Runner"
4. Click Install

Usage:
1. Right-click on a folder in Explorer
2. Select "Pack with Repomix"
3. Choose options in the command palette
4. Output appears in workspace

Website: https://repomix.com

Features:
- Paste repository URL
- Choose output format
- Configure options via UI
- Download or copy result
- No installation required

Limitations:
- Public repositories only
- May have size limits
- No custom configuration files

Usage: repomix [options] [directory]

Options:
  -v, --version                Display version number
  -o, --output <file>          Output file path
  --style <type>               Output style: xml, markdown, plain, json
  --compress                   Enable Tree-sitter compression (~70% reduction)
  --copy                       Copy output to clipboard
  --split-output <size>        Split output by size (e.g., 1mb, 500kb)
  --include <patterns>         Include file patterns (comma-separated)
  --ignore <patterns>          Ignore file patterns (comma-separated)
  --no-security-check          Disable security scanning
  --remote <repo>              Process remote GitHub repository
  --remote-branch <branch>     Specify remote branch
  --remote-trust-config        Trust remote repository config
  --git-logs [count]           Include commit logs
  --git-diff                   Include git diffs
  --remove-comments            Remove code comments
  --remove-empty-lines         Remove empty lines
  --show-line-numbers          Show line numbers in output
  -c, --config <file>          Config file path
  --init                       Initialize config file
  -h, --help                   Display help

1. Code Review: Feed codebase to Claude/ChatGPT for comprehensive review
   repomix --compress --style markdown

2. Documentation Generation: Generate docs from entire codebase
   repomix --include "src/**/*.ts" --style markdown

3. Refactoring Planning: Analyze architecture before major changes
   repomix --compress --git-logs 100

4. Bug Analysis: Share relevant code with AI for debugging
   repomix --include "src/**/*.ts,**/*.test.ts"

5. Onboarding: Create codebase summaries for new team members
   repomix --style markdown --remove-comments

6. Security Audit: Scan entire repository for issues
   repomix --git-logs --git-diff

7. Migration Planning: Analyze legacy code before migration
   repomix --compress --include "src/**/*"

8. API Documentation: Generate API docs from source
   repomix --include "**/*.ts" --ignore "**/*.test.ts"

9. Code Search: MCP integration for AI-powered search
   (Use MCP server with Claude Desktop)

10. CI/CD Analysis: Automated repository packing in pipelines
    (Use GitHub Actions integration)

1. Start with compression for large repositories
   - Reduces tokens by ~70%
   - Preserves structure for AI understanding

2. Use .repomixignore for consistent exclusions
   - Version control your ignore patterns
   - Keep output focused on relevant code

3. Choose the right format for your LLM
   - XML: Best for Claude (default)
   - Markdown: Best for human readability
   - JSON: Best for programmatic processing

4. Monitor token counts
   - Stay within LLM context limits
   - Use --split-output for very large repos

5. Enable security scanning
   - Always review output before sharing
   - Remove sensitive data from repositories

6. Use specific include patterns
   - Focus on relevant code
   - Reduce noise in AI analysis

7. Leverage git history for context
   - Include recent commits for change context
   - Add diffs for specific analysis

8. Automate with GitHub Actions
   - Keep packed outputs up to date
   - Integrate with documentation workflows

9. Use MCP for interactive AI sessions
   - Enable Claude to pack and search dynamically
   - Better for iterative analysis

10. Test with different formats
    - Some LLMs work better with specific formats
    - Markdown is more token-efficient for some models

Issue: Output too large for LLM context
Solution: Use --compress or --split-output 1mb

Issue: Sensitive data in output
Solution: Review security warnings, add to .repomixignore

Issue: Node.js version error
Solution: Upgrade to Node.js 22 or later

Issue: Remote repository fails
Solution: Check repository is public, verify URL format

Issue: Unexpected files included
Solution: Check .gitignore, add patterns to .repomixignore

Issue: Compression removes needed code
Solution: Don't use --compress for implementation review

Issue: Character encoding errors
Solution: Repomix auto-detects, but verify source file encoding

Issue: Token count inaccurate
Solution: Token counts are estimates, verify with your LLM

Issue: Config file not loaded
Solution: Ensure repomix.config.json in project root

Issue: Performance slow on large repos
Solution: Use --include to limit scope, exclude node_modules

Configuration:
  repomix.config.json          - Project configuration (JSON)
  repomix.config.ts            - Project configuration (TypeScript)
  repomix.config.js            - Project configuration (JavaScript)
  .repomixignore               - Project-specific ignore patterns

Output:
  repomix-output.xml           - Default XML output
  repomix-output.md            - Markdown output
  repomix-output.txt           - Plain text output
  repomix-output.json          - JSON output
  repomix-output-part*.xml     - Split output files

Git Integration:
  .gitignore                   - Automatically respected
  .ignore                      - Automatically respected