Nightscout MCP Server

This project is a read-only MCP (Model Context Protocol) server for Nightscout, built with the @modelcontextprotocol/sdk. It allows AI agents and other tools to interact with your Nightscout data in a structured and secure way, enabling advanced analysis, insights, and visualizations.

Features

• Get Entries: Retrieve raw data from your Nightscout instance, such as blood glucose readings, and device statuses.
• Get Treatments: Retrieve treatment data from your Nightscout instance, such as insulin doses and carb intake.
• Extensible: Easily add new tools to expose more of the Nightscout API.
• Secure: Your Nightscout URL and token are kept secure on the server and are not exposed to the AI model.

API Support

• Nightscout API v1: Currently, this server only supports v1 of the Nightscout API.

Getting Started

Prerequisites

• Node.js and npm
• A Nightscout instance

Installation

1. Clone the repository:

   git clone https://github.com/ssuppe/nightscout-mcp.git
   cd nightscout-mcp
   

2. Install dependencies:

   npm install
   

3. Create a .env file in the root of the project and add the following environment variables:

   NIGHTSCOUT_URL=https://your-nightscout-site.com
   NIGHTSCOUT_TOKEN=your-nightscout-api-token
   

Running the Server

You can run the server in two modes:

• Development Mode: This will run the server using ts-node, which will automatically transpile and run the TypeScript code. This is the recommended way to run the server during development.

  npm run dev
  

• Production Mode: This will first build the project by compiling the TypeScript code to JavaScript, and then run the compiled code.

  npm run build
  npm start
  

The server will start and listen on http://localhost:3000/mcp. You will see a confirmation message in the console.

Query Quickstart

Try commands like:

• What is my most recent blood glucose reading?
• What is my average blood sugar since 8am?
• What treatments have I taken this morning?

Your mileage may vary. I intend to add more specific tooling for more predictable results.

TODO

• Add additional read-only capabilities directly from the API (starting with Profiles)
• Create helper tools (such as averages, percentiles, finding high and low times of day, etc)

Contributing

Contributions are welcome! Please see the Contributing Guidelines for more details on how to get involved.

License

This project is licensed under the MIT License. See the LICENSE file for details.