Browser Driver MCP Server

A browser automation tool based on Puppeteer that follows the Model Context Protocol (MCP).

Overview

This package provides a simple MCP server that enables browser automation through Puppeteer. It allows AI models or other MCP clients to control a browser, navigate to websites, take screenshots, fill forms, and more.

Key Features

• Visible Browser: Always runs in non-headless mode so you can see what's happening
• Simple API: Provides basic Puppeteer actions through a clean MCP interface
• Screenshots: Capture full page or element screenshots
• Form Manipulation: Fill forms, click buttons, select options

Installation

npm install -g @brianbuildz/browser-driver

Usage

As a Command Line Tool

Run the MCP server:

browser-driver

This will start the MCP server and open a browser window. The server communicates using the MCP protocol via standard input/output.

In an MCP Client

You can use this tool with any MCP client. Here's a basic example:

// Example MCP client communication
const mcp = require("some-mcp-client-library");

async function runBrowserAutomation() {
  // Connect to the browser-driver
  const browser = await mcp.connect("browser-driver");

  // Navigate to a URL
  await browser.call("puppeteer_navigate", { url: "https://www.google.com" });

  // Take a screenshot
  await browser.call("puppeteer_screenshot", { name: "google-home" });

  // Fill in a form field
  await browser.call("puppeteer_fill", {
    selector: 'textarea[name="q"]',
    value: "Puppeteer automation",
  });

  // Click a button
  await browser.call("puppeteer_click", { selector: 'input[name="btnK"]' });
}

Testing with the Included Test Script

We've included a simple test script to demonstrate the functionality:

node test-puppeteer.js

This will run a sequence of browser automation commands to navigate to Google, search for "Puppeteer browser automation", and take screenshots of the process.

Available Tools

| Tool Name | Description | Parameters | | ---------------------- | --------------------------------- | ------------------------------------------------------------------------------------------------------- | | puppeteer_navigate | Navigate to a URL | url (string) | | puppeteer_screenshot | Take a screenshot | name (string), selector (string, optional), width (number, optional), height (number, optional) | | puppeteer_click | Click an element | selector (string) | | puppeteer_fill | Fill an input field | selector (string), value (string) | | puppeteer_select | Select an option in a dropdown | selector (string), value (string) | | puppeteer_hover | Hover over an element | selector (string) | | puppeteer_evaluate | Execute JavaScript in the browser | script (string) |

Available Resources

| Resource URI | Description | | --------------------- | ------------------------------------ | | console://logs | Browser console logs | | screenshot://{name} | Screenshots taken during the session |

License

MIT