Puppeteer MCP Server

Example Use Cases

• 1

  Navigate websites, click and interact with page elements, fill forms, and automate web workflows in a real browser.

• 2

  Capture full-page or element-specific screenshots and expose them as screenshot:// resources for downstream analysis or reporting.

• 3

  Execute custom JavaScript in-page and collect browser console logs for debugging, validation, or data extraction.

Description

This MCP server runs Puppeteer to let LLMs and agents control a real browser—navigate pages, click and hover elements, fill forms, select options, take screenshots, and execute arbitrary JavaScript. It also captures browser console logs and exposes screenshots as named resources accessible via the MCP protocol. The project is deployable via Docker or NPX and is distributed under the MIT License. It is useful for automated web testing, content extraction, visual capture, and in-browser debugging workflows.