mcp-master-puppeteer

MCP Master Puppeteer is an advanced MCP server that leverages Puppeteer to automate browser tasks while keeping responses concise and decision-focused. It exposes tools designed to minimize token usage by default, yet provide rich data when requested through optional flags. Core capabilities include navigating pages, taking optimized screenshots, extracting content in structured formats, collecting page metadata and SEO signals, and evaluating accessibility and performance metrics. You can use these tools to gather essential page information, render pages for verification, and obtain analysis suitable for agent-driven workflows without overwhelming you with unnecessary data.

To use the server, load it via the provided MCP command (for example using npx as shown in the config). The available tools are: puppeteer_navigate_analyze for visiting a URL and returning status, title, and optional content; puppeteer_screenshot_plus for generating image captures suitable for large language models; puppeteer_extract_content for extracting page content with configurable formats; and puppeteer_get_page_info for comprehensive page metadata, SEO, accessibility, and performance metrics. Each tool includes options to tailor outputs (contentFormat, includeMetadata, performance metrics, image breakpoints, etc.).

Prerequisites:

• Node.js (LTS) and npm installed on your system
• Basic familiarity with MCP commands

1. Install the MCP server package (via GitHub) using npx as in the README example:

npx -y github:flrngel/mcp-master-puppeteer#main

Note: If you prefer to clone the repo and install locally, you can:

git clone https://github.com/flrngel/mcp-master-puppeteer.git
cd mcp-master-puppeteer
npm install

2. Start the MCP server (example, depending on your MCP setup you may use a launcher):

# If the server exposes a CLI, e.g., node dist/server.js or npm run start
npm run start

3. Validate installation by calling one of the tools, e.g., navigate analyze, via your MCP client or CLI integration.

Prerequisites recap:

• Node.js and npm installed
• Access to a network-enabled environment for Puppeteer to load pages
• Basic MCP setup to register the server (as described in the README example)

Tips and considerations:

• The server emphasizes token-efficient responses. Use includeMetadata, includePerformance, or includeAnalysis flags only when you need deeper data.
• Adjust Puppeteer timeouts and navigation options to handle slow pages or dynamic content.
• When generating screenshots, you can resize payloads to stay within Claude's dimension limits. The tool supports multiple breakpoints and full-page captures.
• For content extraction, you can request different output formats (markdown, html, plain-text, or structured-json) depending on downstream needs.
• If you encounter resource or navigation errors, check network conditions, page load strategies (waitUntil), and ensure the target URL is accessible from your environment.
• Environment variables: NODE_OPTIONS can be tuned to increase memory if you process large pages or many concurrent tasks.