Fetch URL MCP Server

Strips web pages down to clean Markdown by fetching the URL and extracting just the main content, no navigation or scripts. Automatically rewrites GitHub, GitLab, Bitbucket, and Gist URLs to pull raw content instead of rendered pages. Exposes a single fetch_url tool that returns the converted Markdown plus metadata like title and content size. For large or slow pages, it can run as an async task with progress updates. Runs over stdio by default, but supports HTTP mode with auth and rate limiting if you need a proper endpoint. Useful when you want to feed web content to an LLM without the noise.