Puppeteer

by Model Context Protocol

v2025.5.12

MIT

Repository Docs npm

Web & Search

browserautomationscrapingscreenshots

Summary

Browser automation for AI clients — navigate, click, type, screenshot, and run JavaScript in a real page.

Features

Navigate pages and interact with elements
Capture screenshots of pages or specific elements
Execute JavaScript in the page context

Installation

Set up this MCP server in your favorite AI agent — copy a ready-made configuration below.

Any MCP-compatible agent

Most agents (Claude, Cursor, Windsurf, VS Code, and more) read a standard mcpServers configuration.

Open your agent's MCP configuration file.
Merge the snippet below into it.
Restart the agent — the "Puppeteer" tools will be available.

{
  "mcpServers": {
    "puppeteer": {
      "command": "npx",
      "args": [
        "-y",
        "@modelcontextprotocol/server-puppeteer"
      ]
    }
  }
}

Claude Code

Add the server with a single CLI command.

claude mcp add puppeteer -- npx -y @modelcontextprotocol/server-puppeteer

Claude Desktop

Add the server to claude_desktop_config.json (Settings → Developer → Edit Config).

{
  "mcpServers": {
    "puppeteer": {
      "command": "npx",
      "args": [
        "-y",
        "@modelcontextprotocol/server-puppeteer"
      ]
    }
  }
}

Cursor

{
  "command": "npx",
  "args": [
    "-y",
    "@modelcontextprotocol/server-puppeteer"
  ]
}

Windsurf

{
  "command": "npx",
  "args": [
    "-y",
    "@modelcontextprotocol/server-puppeteer"
  ]
}

VS Code

{
  "command": "npx",
  "args": [
    "-y",
    "@modelcontextprotocol/server-puppeteer"
  ]
}

Description

The Puppeteer MCP server gives an assistant control of a real Chrome browser, which is how it reaches the large portion of the web that has no API.

What it can do

Navigate to any URL
Interact — click elements, fill and submit forms, select options, hover
Screenshot the full page or a specific element, which is how the assistant sees what it is doing
Execute JavaScript in the page context for extracting data or manipulating state
Read console output for debugging what a page is actually doing

Where it is useful

Testing web applications interactively, extracting data from sites without APIs, filling repetitive forms, and verifying that a deployed change looks right. For agent development, the screenshot capability is what closes the loop — the assistant can check its own work visually.

Notes

Pages that require login mean handing credentials to the browser session, so think about what that grants. For production-scale automation, hosted alternatives such as Browserbase handle proxies, fingerprinting, and concurrency that this server does not.

Part of the MCP reference collection, now in the archived servers repository.