Fetch

Reference MCP server that fetches a web page and returns it as Markdown, so a local model can read live content without a browser.

Install Source repository

Quick verdict

The simplest way to give an agent web reading: one tool, Markdown output that small local models handle well, no API key, and it is maintained in the MCP steering group's reference repository.

MaintainerModel Context Protocol reference servers (reference)
Security scopeRead-only
Transportsstdio
Install channelpypi · mcp-server-fetch
EvidenceSource Validated
LicenseMIT · Open source

The Model Context Protocol reference server for web content. Its single tool, fetch, retrieves a URL and converts the HTML to Markdown for the model, with paging (max_length, start_index) and a raw mode. It is a Python package run with uvx (not an npm package — an earlier version of this directory captioned it as one). It honours robots.txt for model-initiated requests by default and can be pointed at a proxy.

MCP security best practices

Install Fetch

uvx mcp-server-fetch

Config entry (mcpServers)

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

Runs with uvx (uv); pip install mcp-server-fetch + python -m mcp_server_fetch also works.

Credentials

No credentials required.

Verified configurations

Not run by us yet. This profile is Source Validated: the install channel, config entry, credentials and license were checked against the source and upstream metadata, but no configuration has been installed, discovered and executed in a recorded environment. When one is, it appears here with the eight scope dimensions below.

Best for

Grounding a local model in a specific page — documentation, a changelog, a public API reference — without a browser automation server.

Why use it

The simplest way to give an agent web reading: one tool, Markdown output that small local models handle well, no API key, and it is maintained in the MCP steering group's reference repository. Installing Node.js switches it to a more robust HTML simplifier.

Tradeoffs

Read-only, but the upstream README carries an explicit caution: the server can reach local and internal IP addresses, so treat it as network egress from your machine and do not run it where internal services should stay private. Requires uv/uvx (or Python 3 with pip). Robots.txt is respected only for model-initiated fetches unless you pass --ignore-robots-txt.

Upstream state

pypi package published · repository active. Lifecycle is derived from a committed upstream snapshot at build time, never authored by hand.

Editorial review

Reviewed byOpenSourcesAI EditorialLast updatedAugust 2026Sourcessrc/fetch/README.md — uvx/pip/Docker configs, the fetch tool, robots.txt and internal-IP caution; PyPI: mcp-server-fetch

Grade B — Source Validated. Re-review is due by 2027-02-16. Config entries, credential names and package status can change upstream; verify against the source repository before granting an agent access.

Sources

src/fetch/README.md — uvx/pip/Docker configs, the fetch tool, robots.txt and internal-IP caution (tier 1, checked 2026-08-16)
PyPI: mcp-server-fetch (tier 2, checked 2026-08-16)

FilesystemReference MCP server for local file operations — read, search, edit and move files inside the directories you explicitly allow.
Firecrawl MCPSearch, scrape, parse and crawl the live web through Firecrawl, with optional browser interaction and autonomous research.
Tavily MCPSearch, extract, map and crawl live web content through Tavily's official hosted MCP server.

Related guides

All MCP servers →