Fetch MCP Server
A minimal reference server that fetches a URL and converts its content to clean Markdown for an agent to read, the simplest way to give an agent basic web access.
Add to Claude Code
claude mcp add fetch -- uvx mcp-server-fetch
Source: https://github.com/modelcontextprotocol/servers/tree/main/src/fetch
About
The Fetch server does one thing: given a URL, it retrieves the page and converts the HTML into readable Markdown, with support for pagination through long pages via start-index and length parameters.
It's the lightest-weight way to give an agent web-reading ability when you don't need full browser automation or search, just the ability to pull and read a specific page's content.
Key features
- Fetches a URL and converts HTML to clean Markdown
- Supports paginating through long pages
- No API key or external service required
- Minimal, single-purpose tool surface
Use cases
- Let an agent read the content of a documentation page or article you link to
- Pull a specific web page into context for summarization
- Give a lightweight agent basic web-reading ability without full browser automation
Available tools
fetch
Fetches a URL and returns its content converted to Markdown.
Frequently asked questions
Does it render JavaScript?
No. It fetches the raw HTML response, so pages that require JavaScript to render their content won't show that content. Use a browser-automation server like Playwright MCP for those cases.
Do I need an API key?
No. It makes a plain HTTP request with no external service or credentials involved.