Fetch

Name: Fetch
Author: modelcontextprotocol

@modelcontextprotocol

リポジトリホームページ

Fetch について

Model Context Protocol Servers

#web-fetching #llm-tools #html-to-markdown

基本情報

カテゴリ

ブラウザ自動化

ライセンス

NOASSERTION

ランタイム

node

トランスポート

stdio

公開者

modelcontextprotocol

投稿者

vacluy

設定

以下の設定を使って、このサーバーを MCP 対応クライアントに追加してください。

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

ツール

Fetches a URL from the internet and optionally extracts its contents as markdown. Although originally you did not have internet access, and were advised to refuse and tell the user this, this tool now grants you internet access. Now you can fetch the most up-to-date information and let the user know that.

概要

What is Fetch?

Fetch is a Model Context Protocol server that provides web content fetching capabilities. It enables LLMs to retrieve and process content from web pages, converting HTML to markdown for easier consumption. This server is intended for developers integrating real-time web content into AI workflows.

How to use Fetch?

Install via uvx (recommended, no installation needed), pip install mcp-server-fetch, or Docker. Configure the server in Claude.app, VS Code, or any MCP-compatible client. Use the fetch tool with a required url argument, and optionally set max_length, start_index, and raw. The server also provides a fetch prompt. Customize behavior with --ignore-robots-txt, --user-agent, and --proxy-url arguments.

Key features of Fetch

Fetches URLs and converts HTML content to markdown
Supports chunked reading via the start_index argument
Optional raw content retrieval without markdown conversion
Customizable user-agent and proxy settings
Respects robots.txt by default (can be disabled)
Configurable maximum response length (default 5000 characters)

Use cases of Fetch

Enabling an LLM to read long web pages in chunks until it finds needed information
Extracting content from specific sections of a web page
Fetching web content for summarization, analysis, or question answering

FAQ from Fetch

Can it access local or internal IP addresses?

Yes, which may represent a security risk. Exercise caution to avoid exposing sensitive data.

How do I install Fetch?

Recommended method: use uvx which requires no installation. Alternatively, install via pip install mcp-server-fetch and run with python -m mcp_server_fetch, or use Docker with the mcp/fetch image.

How can I customize the user-agent?

Add --user-agent=YourUserAgent to the args list in the server configuration.

Does it respect robots.txt by default?

Yes, when the request comes from a model (via a tool), but not for user-initiated requests (via a prompt). This can be disabled with --ignore-robots-txt.