mcp-fetch-node exposes a web-accessible content fetch tool that converts web pages into Markdown-friendly content for consumption by LLMs. The server runs as a web service (SSE endpoint) at /sse by default on port 8080 and provides a fetch tool that retrieves a URL and returns its extracted content. The fetch tool supports pagination through start_index and can limit length with max_length, making it suitable for chunked reading of long pages. You can also customize the User-Agent or disable robots.txt handling via command-line arguments when starting the server.

To use the tools, start the server and then call the /sse endpoint or integrate the fetch tool in your MCP workflow. The available tool described in the README is fetch, which takes a URL and returns the page content converted to Markdown; prompts can also reuse the same fetch capability by referencing the fetch tool description provided in the server's API.

Prerequisites:

• Node.js (recommended LTS version) and npm or pnpm installed on your system
• Optional: Docker if you prefer containerized running

Install and run from source:

1. Clone the repository git clone https://github.com/tgambet/mcp-fetch-node.git cd mcp-fetch-node

2. Install dependencies npm install

   or if you use pnpm

   pnpm install

3. Build (if the project requires a build step) npm run build

   or pnpm run build

4. Run the server (default port 8080) node server.js

5. Verify the server is running curl http://localhost:8080/sse

Running with npx (quick start): npx -y mcp-fetch-node

Running with Docker (image tgambet/mcp-fetch-node): docker run -it tgambet/mcp-fetch-node

Notes and tips:

• The server by default exposes an SSE endpoint at /sse on port 8080. You can override the port with the PORT environment variable when launching.
• Robots.txt handling can be customized per request by providing the --ignore-robots-txt flag or via environment variables.
• Customize the User-Agent string with --user-agent or by setting the USER_AGENT environment variable to apply to tool calls initiated by a model or a user prompt.
• If you plan to deploy behind a reverse proxy, ensure proper CORS/headers configuration if your client is not server-side.