概要
What is Fetcher MCP?
Fetcher MCP is an MCP server for fetching web page content using the Playwright headless browser. It is designed for users who need to extract content from dynamic websites that require JavaScript execution, and supports intelligent main content extraction via the Readability algorithm.
How to use Fetcher MCP?
Run directly with npx -y fetcher-mcp, after installing the required browser with npx playwright install chromium. Configure the server in Claude Desktop JSON, or deploy via Docker. Use --transport=http to enable HTTP and SSE endpoints. The server provides tools like fetch_url, fetch_urls, and browser_install.
Key features of Fetcher MCP
- JavaScript execution via Playwright headless browser
- Intelligent content extraction removing ads and navigation
- Supports both HTML and Markdown output formats
- Parallel batch fetching with
fetch_urlstool - Automatic blocking of unnecessary resources (images, styles, etc.)
- Robust error handling and configurable parameters
Use cases of Fetcher MCP
- Fetch content from modern JavaScript-heavy websites
- Scrape multiple URLs concurrently for batch data collection
- Integrate with AI assistants to provide web content in a readable format
- Handle websites with anti-bot mechanisms by waiting for full page load
FAQ from Fetcher MCP
What is the advantage over traditional web scrapers?
Fetcher MCP uses Playwright to execute JavaScript, enabling it to handle dynamic web content and modern web applications that traditional scrapers cannot.
What are the runtime requirements?
Requires Node.js and Playwright's Chromium browser. Can be run via npx or Docker.
How does content extraction work?
The built-in Readability algorithm automatically extracts the main content from web pages, removing ads, navigation, and other non‑essential elements.
Can I fetch multiple URLs at once?
Yes, the fetch_urls tool enables concurrent fetching of multiple URLs with parallel processing for improved efficiency.
What transports are supported?
The server supports both Streamable HTTP endpoint (/mcp) and SSE endpoint (/sse) when started with --transport=http.
