MCP-Puppeteer-Linux

MCP-Puppeteer-Linux is an MCP server that connects LLM clients to a local Puppeteer-driven browser tailored for Linux operating systems. Designed for developers, automated testers, and AI agents operating on desktop Linux setups, it enables models to navigate web pages, capture element-specific or full-page screenshots, and execute JavaScript inside a live browser environment. This particular fork addresses common Linux GUI hurdles by integrating automatic display server detection for both X11 and Wayland environments, setting required environment variables dynamically across desktop environments like GNOME and KDE. Beyond basic navigation, the server allows AI agents to interact with interactive DOM elements through clicking, hovering, form field inputs, and dropdown selections. It also surfaces two structured resources: live browser console logs and rendered screenshot images in PNG format. These capabilities allow assistants to observe runtime web behaviors, inspect layout states, and gather real-time data directly on Linux machines.

Category: Browser & Web Automation

Tags: javascript, linux, puppeteer, screenshots, web scraping

Visit MCP-Puppeteer-Linux

How to install and configure MCP-Puppeteer-Linux

Follow these steps to set up and configure MCP-Puppeteer-Linux: 1. Install automatically for Claude Desktop using Smithery: bash npx -y @smithery/cli install @PhialsBasement/mcp-puppeteer-linux --client claude 2. Alternatively, perform a manual installation by cloning the repository and installing its dependencies: bash git clone https://github.com/PhialsBasement/MCP-Puppeteer-Linux.git cd MCP-Puppeteer-Linux npm install 3. Add the server to your Claude Desktop configuration file under mcpServers: json { "mcpServers": { "puppeteer": { "command": "npx", "args": ["ts-node", "/path/to/index.ts"] } } } 4. Restart your MCP client to load the tools and resources.

What you can do with MCP-Puppeteer-Linux

• Automating form submissions and dropdown selections across multi-step web workflows running on Linux desktop environments. - Capturing full-page or selector-specific PNG screenshots of web applications to visually inspect rendering issues. - Executing custom JavaScript expressions directly in the browser console to extract dynamic DOM content. - Monitoring web application runtime errors and debugging front-end states by reading live browser console logs. - Navigating complex single-page applications under native Wayland or X11 display servers without manual display configuration.

Key facts

• Open Source
• https://github.com/PhialsBasement/MCP-Puppeteer-Linux
• Browser & Web Automation, Design, Media & Creative
• javascript, linux, puppeteer, screenshots, web scraping

Part of MCP Servers

What is MCP-Puppeteer-Linux?

MCP-Puppeteer-Linux is an open source Model Context Protocol server that enables LLMs to control a Puppeteer web browser on Linux. It includes built-in detection for X11 and Wayland display servers, allowing agents to navigate pages, click buttons, fill out forms, execute JavaScript, and retrieve console logs or screenshots directly within desktop environments.

How do I install MCP-Puppeteer-Linux?

You can install it automatically for Claude Desktop using the Smithery CLI with the command npx -y @smithery/cli install @PhialsBasement/mcp-puppeteer-linux --client claude. For manual setups, clone the git repository, run npm install, and configure your client to run ts-node index.ts.

Which display servers are supported by MCP-Puppeteer-Linux?

The server automatically detects and configures both Wayland and traditional X11 display servers. It configures relevant environment variables such as WAYLAND_DISPLAY, QT_QPA_PLATFORM, GDK_BACKEND, and DISPLAY to maintain compatibility across desktop environments like GNOME and KDE, along with XWayland support.

Is MCP-Puppeteer-Linux open source?

Yes, MCP-Puppeteer-Linux is open source software released under the MIT License. This permits you to freely use, modify, adapt, and distribute the server code subject to the terms and conditions outlined in the repository license.