Playwright Universal MCP

🚀 Playwright Universal MCP

A universal Model Context Protocol (MCP) server for browser automation, designed to run reliably in containerized environments with restricted permissions. It offers browser control capabilities for AI assistants like Claude and GPT-4.

✨ Features

• 🌐 Multi-browser Support: Choose from Chromium, Chrome, Microsoft Edge, Firefox, or WebKit.
• 🐳 Container-friendly: Suitable for restricted environments such as Docker.
• 👁️ Headless/Headed Mode: Use headless mode on the server and headed mode for debugging.
• 🛠️ Comprehensive Browser Control: Navigate, click, type, take screenshots, and more.
• 📄 Multi-page Support: Create and manage multiple browser tabs.

📦 Installation

Option 1: Install via pipx (Recommended)

# Install the MCP server globally
pipx install playwright-universal-mcp

# Install the required browser
playwright install chromium
# Optional: Install other browsers
playwright install firefox webkit msedge chrome

Option 2: Install in a Virtual Environment

# Create and activate a virtual environment
python -m venv playwright-mcp-venv
source playwright-mcp-venv/bin/activate

# Install the package
pip install playwright-universal-mcp

# Install the browser
playwright install chromium

Option 3: Install from Source

# Clone the repository
git clone https://github.com/yourusername/playwright-universal-mcp.git
cd playwright-universal-mcp

# Install the package
pip install -e .

# Install the browser
playwright install chromium

💻 Usage Examples

Command-line Options

playwright-universal-mcp --help

Display help information:

Usage: playwright-universal-mcp [options]
Common options:
  --version   Show the version number and exit
  -h, --help  Show this help message and exit

📚 Documentation

Claude Desktop Configuration

Add the following content to Claude's settings.json to specify the MCP server address:

{
  "mcp": {
    "serverUrl": "http://localhost:1234"
  }
}

Configuration for Other Applications

Configure according to the documentation of the specific application.

Containerization

# Use the base image
FROM python:3.8

# Install dependencies
RUN pip install playwright-universal-mcp

# Install the required browser
RUN playwright install chromium

# Start the server
ENTRYPOINT ["playwright-universal-mcp", "--headless"]

Browser Toolkit

The MCP server provides the following functions:

• navigate: Navigate to a specified URL.
• click: Click an element by selector or text.
• type: Type text into an input field.
• get_text: Get the text content of an element.
• get_page_content: Get the HTML content of the current page.
• take_screenshot: Take a screenshot of the current page.
• new_page: Create a new browser tab.
• switch_page: Switch to a specified tab.
• get_pages: List all available tabs.
• wait_for_selector: Wait for an element to appear and be visible.
• get_browser_info: Get browser session information.

📄 License

MIT License

Acknowledgments

This project is based on:

• Playwright for browser automation capabilities.
• Model Context Protocol (MCP) for connection standards.
• MCP Python SDK for MCP implementation.