mcp-browser-automation

MCP.Pizza Chef: hrmeetsingh

The mcp-browser-automation server is a Model Context Protocol-based AI agent designed to automate web browsing tasks using Playwright. It integrates with the Claude Desktop client to receive user prompts and control a browser session programmatically. This server facilitates real-time, AI-driven interaction with web environments, enabling complex browsing workflows and automation directly from an LLM interface. It requires Playwright, Claude Desktop, and Node.js for setup and is ideal for developers looking to build AI-enhanced browsing agents or automate web navigation tasks securely and efficiently.

Use This MCP server To

Automate web navigation and interaction via AI prompts Control browser sessions programmatically with Playwright Integrate AI-driven browsing in desktop applications Enable real-time web automation from Claude Desktop Test web applications through scripted AI commands Simulate user browsing behavior for data extraction

README

MCP Browser Automation

This is demo project to practice Model Context Protocol based server implemenation for automating browsing with Playwright. It interacts with a Claude Desktop client to accept user prompts and use server to control browser.

Browser Automation Server MCP server

Pre-requisites

  • Playwright
  • Claude Desktop
  • Node.js

Building

  1. Clone the repository: git clone https://github.com/hrmeetsingh/mcp-browser-automation.git
  2. Install dependencies: npm install
  3. Verify the output executables are present in dist folder

Integration

  1. Create a configuration file in ~/Application\ Support/Claude/claude_desktop_config.json (This is for macOS)
  2. Copy the following to the file:
{
  "mcpServers": {
    "mcp-browser-automation": {
      "command": "node",
      "args": ["/path/to/mcp-browser-automation/dist/index.js"]
    }
  }
}
  1. Start Claude Desktop

Usage

  1. Open Claude Desktop
  2. Start a new conversation to open a browser and navigate to a URL

Example

  • Added MCP Server options Added MCP Server options

  • Navigating to a URL and doing actions with playwright Navigating to a URL and entering text

mcp-browser-automation FAQ

How do I install the mcp-browser-automation server?
Clone the GitHub repo, run npm install to get dependencies, and ensure Playwright, Claude Desktop, and Node.js are installed.
What platforms does mcp-browser-automation support?
It primarily supports macOS for integration with Claude Desktop but can be adapted for other platforms with Playwright and Node.js.
How does mcp-browser-automation interact with Claude Desktop?
It listens for user prompts from Claude Desktop and uses Playwright to automate browser actions accordingly.
Can I customize browser automation scripts?
Yes, you can modify the server code to tailor Playwright scripts for specific browsing tasks.
Is this server compatible with other LLM clients besides Claude Desktop?
While designed for Claude Desktop, the server can be adapted to work with other MCP-compatible LLM clients like OpenAI GPT-4 or Anthropic Claude.
What are the prerequisites for running this server?
You need Playwright, Node.js, and Claude Desktop installed and configured properly.
How do I configure the server for Claude Desktop?
Create a configuration file at ~/Application Support/Claude/claude_desktop_config.json with the appropriate MCP server details.
Does mcp-browser-automation support secure and scoped browser control?
Yes, it follows MCP principles to ensure safe, scoped, and observable interactions with the browser environment.