Fetch MCP: Documentation

Connection

uvx mcp-server-fetch (recommended) or python -m mcp_server_fetch after pip install mcp-server-fetch (stdio). The --ignore-robots-txt flag disables robots.txt compliance entirely.

Tools

fetch(url, max_length, start_index, raw):

• url — the page address (required).
• max_length — how many characters to return at once (default 5000).
• start_index — offset for paging through long documents.
• raw — return raw HTML instead of markdown (default false).

Agent flow

Reading articles, docs and any public pages without spinning up a browser.

Limits

By default the server respects robots.txt for model-initiated requests (but not for direct user requests); there's no JS rendering — interactive SPAs need a full browser (e.g. Playwright MCP).