home / mcp / nccn guidelines mcp server

NCCN Guidelines MCP Server

A Model Context Protocol (MCP) server that provides access to NCCN (National Comprehensive Cancer Network) clinical guidelines

Installation
Add the following to your MCP client configuration file.

Configuration

View docs
{
  "mcpServers": {
    "gscfwid-nccn_guidelines_mcp": {
      "command": "uv",
      "args": [
        "--directory",
        "<abslute_direction_of_NCCN_guidelines_MCP>",
        "run",
        "server.py"
      ],
      "env": {
        "NCCN_PASSWORD": "<your_nccn_password>",
        "NCCN_USERNAME": "<your_nccn_username>"
      }
    }
  }
}

You run a MCP server that reads NCCN clinical guidelines directly from PDFs, indexes them, and serves precise, content-driven answers to medical questions through an MCP client. This setup emphasizes accuracy by reading source documents and caching an up-to-date index for efficient queries.

How to use

You connect to the NCCN Guidelines MCP Server using an MCP client. The server exposes a small set of tools to retrieve the guidelines index, download PDFs, and extract content from specific pages. Typical workflows include checking the available index to see what guidelines exist, downloading a PDF for a given guideline, and extracting exact pages that contain the evidence you need. Use your client’s prompt to ask targeted medical questions; the server will analyze requests, read the relevant pages, and provide evidence-based answers based on the extracted content.

How to install

Prerequisites you need before starting: a computer with Python and the MCP runner available on your system. You also need an NCCN account for authentication when downloading PDFs.

Step 1. Install the MCP runner (uv) if you do not already have it.

Step 2. Prepare and run the server configuration. The following configuration starts the server in stdio mode using uv and points to the server script. Replace the placeholder with your actual NCCN credentials and directory path.

Additional content

Configuration notes: The server requires NCCN credentials to access and download guideline PDFs. You will need to provide these credentials to the server environment when you start it.

First-time setup: The server will generate a YAML index of NCCN guidelines on the initial run. This setup step may take 1–2 minutes. Wait for the index to complete before querying the server.

Response times: Because the server reads guidelines directly rather than using a retrieval-augmented generation approach, you should expect longer processing times while the PDFs are downloaded and pages are read for the first requests.

Available tools

get_index

Fetches the raw NCCN guidelines index YAML content to understand available guidelines and structure.

download_pdf

Downloads NCCN guideline PDFs using provided URL and optional credentials, saving them to a local file.

extract_content

Extracts content from specified PDF pages while preserving layout, returning the relevant text for answer generation.