fetch

Fetch MCP Server

fetch brings web page content into LLM workflows, converting HTML into Markdown or returning raw content. It serves people using MCP clients who need assistants to retrieve pages and read selected portions of longer results.

Server facts

fetch MCP Server

Tech

Runtime / language
node
Transports
stdio
Remote available
No
Authentication
Not specified

Activity

GitHub stars
Not recordedPart of the modelcontextprotocol/servers monorepo
Last commit
Data as of

Tags

  • html to markdown
  • llm tools
  • web fetching

Overview

Web page retrieval is the focus of fetch: its URL-based tool brings page content into an LLM workflow and converts HTML into Markdown. A raw-content option lets callers skip that conversion. Callers can set a character limit and choose the starting position within the extracted text, allowing an assistant to work through a long page in sections rather than rely on one truncated response. The project also provides a URL-taking prompt for requesting page extraction, alongside configuration examples for Claude.app and VS Code.

Request behavior differs according to how retrieval begins. Model-triggered tool requests respect robots.txt by default, while user-triggered prompt requests do not; configuration can disable that handling. User-agent customization and proxy configuration are also available. Deployment examples cover uvx, pip, and Docker, and the documentation describes an optional Node.js installation for a different HTML simplifier. Access boundaries deserve attention: the server can reach local and internal IP addresses, so its use may expose sensitive data if those destinations are accessible.

fetch setup notes

  • fetch records stdio only; confirm a network transport upstream before planning remote access.
  • fetch: Part of the modelcontextprotocol/servers monorepo
  • fetch has no recorded license; confirm reuse terms in its repository.

Related MCP servers