Fetch MCP Server

This MCP server provides functionality to fetch web content in various formats, including HTML, JSON, plain text, and Markdown.

Available on NPM

Components

Tools

• fetch_html

  • Fetch a website and return the content as HTML
  • Input:
    • url (string, required): URL of the website to fetch
    • headers (object, optional): Custom headers to include in the request
    • max_length (number, optional): Maximum length to fetch (default 5000, can change via environment variable)
    • start_index (number, optional): Used together with max_length to retrieve contents piece by piece, 0 by default
  • Returns the raw HTML content of the webpage

• fetch_json

  • Fetch a JSON file from a URL
  • Input:
    • url (string, required): URL of the JSON to fetch
    • headers (object, optional): Custom headers to include in the request
    • max_length (number, optional): Maximum length to fetch (default 5000, can change via environment variable)
    • start_index (number, optional): Used together with max_length to retrieve contents piece by piece, 0 by default
  • Returns the parsed JSON content

• fetch_txt

  • Fetch a website and return the content as plain text (no HTML)
  • Input:
    • url (string, required): URL of the website to fetch
    • headers (object, optional): Custom headers to include in the request
    • max_length (number, optional): Maximum length to fetch (default 5000, can change via environment variable)
    • start_index (number, optional): Used together with max_length to retrieve contents piece by piece, 0 by default
  • Returns the text content of the webpage with HTML tags, scripts, and styles removed

• fetch_markdown

  • Fetch a website and return the content as Markdown
  • Input:
    • url (string, required): URL of the website to fetch
    • headers (object, optional): Custom headers to include in the request
    • max_length (number, optional): Maximum length to fetch (default 5000, can change via environment variable)
    • start_index (number, optional): Used together with max_length to retrieve contents piece by piece, 0 by default
  • Returns the content of the webpage converted to Markdown format

Resources

This server does not provide any persistent resources. It's designed to fetch and transform web content on demand.

Getting started

1. Clone the repository
2. Install dependencies: npm install
3. Build the server: npm run build

Usage

To use the server, you can run it directly:

npm start

This will start the Fetch MCP Server running on stdio.

Environment variables

• DEFAULT_LIMIT - sets the default size limit for the fetch (0 = no limit)

Usage with Desktop App

To integrate this server with a desktop app, add the following to your app's server configuration:

{
  "mcpServers": {
    "fetch": {
      "command": "npx",
      "args": [
        "mcp-fetch-server"
      ], 
      "env": {
        "DEFAULT_LIMIT": "50000" // optionally change default limit
      }
    }
  }
}

Features

• Fetches web content using modern fetch API
• Supports custom headers for requests
• Provides content in multiple formats: HTML, JSON, plain text, and Markdown
• Uses JSDOM for HTML parsing and text extraction
• Uses TurndownService for HTML to Markdown conversion

Development

• Run npm run dev to start the TypeScript compiler in watch mode
• Use npm test to run the test suite

License

This project is licensed under the MIT License.