Fetch MCP Server

Overview

The MCP Fetch Server provides web content fetching capabilities, enabling LLMs to retrieve and process content from web pages, converting HTML to markdown for easier consumption. It supports chunked reading of web pages.

Key Features

Available Tools

Quick Reference

Detailed Usage

use_mcp_tool({
  server_name: "fetch",
  tool_name: "fetch",
  arguments: {
    url: "https://example.com",
    max_length: 5000,
    start_index: 0,
    raw: false
  }
});

Installation

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

Common Use Cases

1. Summarizing Web Pages

Fetch web content and have an LLM summarize it for quick insights:

// Fetch a news article and summarize it
use_mcp_tool({
  server_name: "fetch",
  tool_name: "fetch",
  arguments: {
    url: "https://example.com/news-article",
    max_length: 1000
  }
});

2. Extracting Specific Information

Extract data points from web pages, such as product details or contact information:

// Extract product price from an e-commerce page
use_mcp_tool({
  server_name: "fetch",
  tool_name: "fetch",
  arguments: {
    url: "https://example.com/product/123",
    // LLM would then parse the markdown to find the price
  }
});

3. Monitoring Websites

Periodically fetch content to detect changes or updates on a website:

// Monitor a competitor's pricing page
use_mcp_tool({
  server_name: "fetch",
  tool_name: "fetch",
  arguments: {
    url: "https://competitor.com/pricing"
  }
});

Connection String Format

Not applicable, as the Fetch MCP Server retrieves content from URLs rather than connecting to a database.

Sources

• https://github.com/modelcontextprotocol/servers/tree/main/src/fetch
• https://pypi.org/project/mcp-server-fetch/
• https://registry.npmjs.org/mcp-server-fetch/latest