Browser automation powered by LLMs in JavaScript/TypeScript.
This library is currently under heavy development and NOT READY FOR PRODUCTION USE. The API is unstable and subject to major changes.
See ROADMAP.md for the current development status and priorities.
browser-use-node
is a powerful library that combines browser automation capabilities with Large Language Models (LLMs) to create intelligent browser interactions. Built on top of Playwright and LangChain, it provides a seamless way to automate browser tasks with AI assistance.
This is a JavaScript/TypeScript port of the original browser-use Python library.
This project is based on browser-use, originally created by:
We are grateful for their pioneering work in browser automation with LLMs.
- LLM-powered browser automation
- Multi-tab support
- Built on reliable technologies (Playwright, LangChain)
- TypeScript support
- Modern async/await API
⚠️ Note: These instructions are for development purposes only. The library is not yet ready for production use.
The package is available on npm for experimental purposes:
# ⚠️ Not recommended for production use
npm install browser-use-node
Note: While the package is available on npm as
browser-use-node
, it's currently intended for exploration only. The API is unstable and will undergo significant changes. For development and contributions, please use the development setup below.
- Bun (Required for development)
- A modern browser (Chrome/Firefox/Edge)
-
Clone the repository:
git clone https://github.com/browser-use/browser-use-node.git cd browser-use-node
-
Install dependencies:
bun install
-
Set up environment variables:
cp .env.example .env
Edit
.env
and add your API keys and configuration:OPENAI_API_KEY
- Required for LLM functionality- Other configuration options as needed
The examples/
directory contains sample scripts demonstrating various features:
bun run examples/amazon-search.ts
⚠️ These are development tests. The test suite is still being expanded.
# Run all tests
bun test
# Run specific test suite
bun test --grep "Browser"
# Run tests in watch mode
bun test --watch
When contributing new features or fixes:
- Add tests in the
tests/
directory - Follow the existing test patterns
- Include both unit and integration tests where applicable
- Use the provided test utilities in
tests/utils/
For development debugging:
- Use the
DEBUG
environment variable:DEBUG=browser-use* bun test
- Chrome DevTools are available when running browser tests
- VSCode launch configurations are provided in
.vscode/launch.json
Bun is used as the primary runtime because it provides:
- Faster dependency installation
- Quicker test execution
- Better TypeScript performance
- Improved overall development experience
- Native test runner with better performance
- Built-in TypeScript support
- Ability to build as a native binary or jsc bytecode which is a huge performance boost
# Building
bun run build # Build the project
bun run clean # Clean build artifacts
# Development
bun run type-check # Run TypeScript type checking
bun run format # Format code with Prettier
bun run lint # Run ESLint
bun run prepare # Run build before publishing
# Examples
bun run example:amazon # Run Amazon search example
bun run example:multi-tab # Run multi-tab example
bun run example:simple # Run simple example
# Testing
bun run test # Run all tests
bun run test:smoke # Run smoke tests
# Documentation
bun run docs:dev # Start documentation development server
bun run docs:build # Build documentation
bun run docs:serve # Serve built documentation
bun run docs:clear # Clear documentation cache
bun run docs:generate-api # Generate API documentation
bun run docs # Generate API docs, build and serve
The project generates these outputs in the dist
directory:
- Main bundle (target: bun)
- TypeScript declaration files
- Example builds
- Native binary (experimental)
-
Start with an example:
bun run example:simple
This helps verify your setup is working.
-
Before committing:
bun run type-check # Check types bun run format # Format code bun run lint # Check for issues bun run test # Run tests
The project supports various environment configurations:
# Required
OPENAI_API_KEY=your_key_here
# Optional
DEBUG=browser-use* # Enable debug logging
HEADLESS=false # Run browsers in headed mode
For the best development experience:
-
VSCode Extensions:
- ESLint
- Prettier
- TypeScript and JavaScript
- Playwright Test for VSCode
-
Recommended VSCode settings are provided in
.vscode/settings.json
-
Browser Debugging:
# Run with browser debugging enabled DEBUG=browser-use* HEADLESS=false bun run example:simple
-
Test Debugging:
# Run tests with debug output DEBUG=browser-use* bun test
- Use the
build:binary
option for maximum performance and maximum risks
The project uses Docusaurus for documentation and TypeDoc for API reference.
# Start development server
bun run docs:dev
# Build documentation
bun run docs:build
# Serve built documentation
bun run docs:serve
# Generate API documentation
bun run docs:generate-api
/docs/
- Main documentation site/docs/intro
- Getting started guide/docs/guides
- Usage guides and tutorials/docs/api
- Auto-generated API documentation/docs/examples
- Example usage and code snippets
- Documentation source files are in Markdown format
- API documentation is generated from TypeScript source code comments
- Examples should include both code and explanation
- Follow the documentation style guide
We welcome contributions! Please see our Contributing Guidelines for details on how to get started and our development process. By participating in this project, you agree to abide by our Code of Conduct.
We take security seriously. If you discover a security vulnerability, please follow our Security Policy for responsible disclosure.
- Node.js 16+
- Bun (for development)
- Required packages:
- Playwright ^1.49.1
- LangChain ^0.1.0
- @langchain/openai ^0.0.10
- Other dependencies as listed in package.json
- TypeScript ^5.3.3
- ESLint ^8.56.0
- Prettier ^3.1.1
- Various type definitions (@types/*)
Current Version: 0.1.11