puppeteer-mcp-claude

Overview

puppeteer-mcp-claude is a comprehensive MCP server that enables Claude Code and other MCP-aware clients to control browsers through the Puppeteer library. It provides tools for navigating pages, clicking elements, typing text, executing JavaScript, capturing screenshots, and managing cookies. The server features automatic browser launching with sensible defaults, structured content responses using TOON format, and various modes for stealth browsing, request interception, and integration with existing Chrome instances. The implementation includes detailed examples and CLI utilities for both development and production use cases.

Try asking AI

After installing, here are 5 things you can ask your AI assistant:

• puppeteer_launch

  Launch a browser or connect to an existing Chrome via browserWSEndpoint

  Try:Launch a browser in stealth modeTry:Connect to my running Chrome at localhost:9222
• puppeteer_new_page

  Open a new browser tab

  Try:Open a new tab on GitHubTry:Create a new page for shopping
• puppeteer_navigate

  Navigate to a specific URL

  Try:Go to news.ycombinator.comTry:Navigate to example.com
• puppeteer_click

  Click on a CSS selector element

  Try:Click the login buttonTry:Click the submit form button
• puppeteer_type

  Type text into an input element

  Try:Type my search queryTry:Fill in the username field
• puppeteer_get_text

  Get text content from an element

  Try:Get the title of the articleTry:Extract all product names
• puppeteer_screenshot

  Take a screenshot of the current page

  Try:Take a screenshot of this pageTry:Capture a full page screenshot
• puppeteer_evaluate

  Run JavaScript in the page context

  Try:Get all links on the pageTry:Count how many images are loaded
• puppeteer_wait_for_selector

  Wait for an element to appear on the page

  Try:Wait for the results to loadTry:Wait for the download button to appear
• puppeteer_close_page

  Close the current browser tab

  Try:Close this tabTry:Finish with this page
• puppeteer_set_cookies

  Set cookies for the current page

  Try:Save my login cookiesTry:Set session cookie for authentication
• puppeteer_set_request_interception

  Block resources or inject request headers

  Try:Block images for faster loadingTry:Add authentication headers to requests

Comparable tools

Installation

curl -fsSL https://raw.githubusercontent.com/jaenster/puppeteer-mcp-claude/main/install.sh | bash

iwr -useb https://raw.githubusercontent.com/jaenster/puppeteer-mcp-claude/main/install.ps1 | iex

npm install -g puppeteer-mcp-claude
claude mcp add puppeteer-mcp-claude -- npx -y puppeteer-mcp-claude serve

{
  "mcpServers": {
    "puppeteer": {
      "command": "npx",
      "args": ["-y", "puppeteer-mcp-claude", "serve"]
    }
  }
}

FAQ

How do I connect to an existing Chrome browser?

Use the `puppeteer_launch` tool with a `browserWSEndpoint` parameter. For example, `puppeteer_launch { browserWSEndpoint: "ws://localhost:9222" }` connects to Chrome running on port 9222.

Can I block certain resources to speed up scraping?

Yes, use the `puppeteer_set_request_interception` tool with the `blockResources` parameter. You can block images, media, fonts, stylesheets, and more to speed up page loading during scraping.

Compare puppeteer-mcp-claude with