Search and docs · MCP Steering Group

Set up Fetch

Fetches a web page and converts it to Markdown so the model can read it. Part of the MCP project's reference set.

Publisher: MCP Steering Group Official
Run locally: uvx mcp-server-fetch
Transports: stdio
Authentication: None.
What it can touch: Sends outbound HTTP GET requests, including to internal network addresses.
Licence: MIT (existing) / Apache-2.0 (new contributions)
Source: github.com/modelcontextprotocol/servers/tree/main/src/fetch
Status: Active

Install

This server runs on your machine as a child process of the client. You need the runtime its command uses, such as Node.js for npx, uv for uvx or Docker, plus any variables listed under Environment.

Claude Code

shell

claude mcp add --scope project fetch -- uvx mcp-server-fetch

Cursor

.cursor/mcp.json

{
  "mcpServers": {
    "fetch": {
      "command": "uvx",
      "args": [
        "mcp-server-fetch"
      ]
    }
  }
}

Codex

~/.codex/config.toml

[mcp_servers.fetch]
command = "uvx"
args = ["mcp-server-fetch"]

Config for other clients →

Because it can reach internal network addresses, avoid running it on a machine that sees services you wouldn't expose to the model.

Other servers in search and docs

AWS

AWS Knowledge

AWS-hosted search over AWS documentation, blog posts, What's New entries and regional availability.

Microsoft

Microsoft Learn

Search and fetch official Microsoft and Azure documentation and code samples.

Brave

Brave Search

Brave's own server for web, news, image, video and local search through the Brave Search API.

Exa

Neural web search, page contents and code-context search for agents, hosted by Exa or run locally with an API key.

Tavily

Search, extract, crawl and map websites through Tavily's agent-oriented search API, hosted or run locally.

Firecrawl

Scrape, crawl, search and extract structured data from websites as Markdown or JSON.

Sources

github.com/modelcontextprotocol/servers