📦 bundle4llm

A specialized JavaScript bundler that creates single-file bundles optimized for feeding into Large Language Models (LLMs). Inspired by GitIngest.

Why bundle4llm?

LLMs like Claude and GPT process code more effectively when it's presented as a cohesive, self-contained unit without imports, exports, or external dependencies. This tool automatically:

• Resolves and includes all module dependencies
• Removes import/export statements
• Preserves logical code organization
• Estimates token consumption for different LLM models
• Creates a single file that's ready for LLM processing

GitIngest was my go-to for this, but I wanted something integrated into my workflow that I could run on the command line and integrate with any project I'm working on.

The ideal use case is when you have a self-contained project that you want to feed into an LLM. This bundle is exceptionally useful for running on the Claude REPL to help you find issues in your code.

Features

• 🧩 Recursive Module Resolution: Automatically includes all dependencies
• 📋 Export-Driven Ordering: Uses index.js files to determine correct module order
• 🔄 Directory Structure Preservation: Maintains logical organization with comments
• 🧹 Code Sanitization: Removes imports/exports and optionally strips comments
• 📏 Token Estimation: Provides accurate token counts for various LLM models
• 🛠️ Configurable: Multiple presets and options for customization

Installation

# Install globally
npm install -g bundle4llm

# Or use with npx
npx bundle4llm

Usage

bundle4llm --src ./src --out ./dist --file bundle.js --model claude --preset es2020 --strip-comments -v

Options

| Option | Description | Default | |--------|-------------|---------| | --src | Source directory | ./src | | --out | Output directory | ./dist | | --file | Output filename | llm-bundle.js | | --preset | Code sanitization preset (es2020, es2015, minimal) | es2020 | | --strip-comments | Remove comments from the output | false | | --model | LLM model for token estimation (claude, gpt3, gpt4) | claude | | -v, --verbose | Enable verbose logging | false |

Example

Suppose you have a project with multiple JavaScript files and subdirectories:

src/
├── index.js
├── utils/
│   ├── index.js
│   ├── helpers.js
│   └── math.js
└── components/
    ├── index.js
    ├── base.js
    └── advanced.js

Running:

bundle4llm --src ./src --out ./dist --file llm-code.js --model claude -v

Will produce a single file ./dist/llm-code.js with all project files included, imports/exports removed, and token estimation. The bundle should be a valid JavaScript file that can be run directly either on the browser on a JS REPL.

Project Integration

Adding to your project

To use bundle4llm as part of your development workflow, install it as a devDependency:

# Using npm
npm install --save-dev bundle4llm

# Using yarn
yarn add --dev bundle4llm

# Using pnpm
pnpm add -D bundle4llm

Adding to your scripts

Once installed, you can add it to your package.json scripts:

"scripts": {
  "build:llm": "npx bundle4llm --src=./src --out=./llm-docs --file=myproject-llm.js -v"
}

Then simply run:

npm run build:llm

This will generate an LLM-optimized bundle of your code that you can then feed to models like Claude or GPT for analysis, documentation, or refactoring assistance.

CI/CD Integration

You can also integrate bundle4llm into your CI/CD pipelines to automatically generate LLM-ready code snapshots. This is particularly useful when you want to:

• Generate documentation with AI assistance
• Perform automated code reviews with LLMs
• Create code snapshots for debugging sessions with AI

Add the build:llm script to your CI workflow to generate these files during your build process.

Limitations

• Currently only supports JavaScript (ES2020)
• Does not handle circular dependencies
• Does not include external dependencies so it's best suited for self-contained projects.
• Best suited for projects using ES modules

License

APACHE

Contributing

Contributions welcome! Open an issue or submit a PR.

1. Fork the repository
2. Create your feature branch: git checkout -b feature/amazing-feature
3. Commit your changes: git commit -m 'Add some amazing feature'
4. Push to the branch: git push origin feature/amazing-feature
5. Open a Pull Request