Mathematica MCP

🚀 Mathematica MCP Server

This project offers a Model Context Protocol (MCP) server for interacting with a Wolfram Mathematica kernel. It enables Large Language Models (LLMs) to execute Wolfram Language code in a secure, session - based environment.

🚀 Quick Start

This project provides a way for LLMs to interact with a Wolfram Mathematica kernel through the MCP server. It also includes a tool for generating animal - based identifiers for more accurate transcription.

✨ Features

• The animalid folder contains a tool to generate unique, animal - based identifiers, which are more likely to be accurately transcribed by LLMs compared to UUIDs.
• It offers several tools:
  • create_mathematica_session: Initializes a new Wolfram Language session and returns a unique session ID.
  • execute_mathematica_code: Executes Wolfram Language code within a specified session.
  • close_mathematica_session: Terminates a session and releases its resources.

📦 Installation

Prerequisites

• Python 3.10 or higher.
• uv Python package manager. (Installation guide)
• A local installation of the Wolfram Engine or Mathematica. The wolframclient library requires this to function.

Installation Steps

1. Set the Security Key: This server uses a secret key to generate secure session IDs. You must set this as an environment variable.

   export ANIMALID_SECRET_KEY='your-super-secret-and-long-key-here'
   

   ⚠️ Important Note

   Do not use a weak key or hardcode it in the script.

💻 Usage Examples

Basic Usage

To use this server, you need to connect it to an MCP - compatible client, such as Claude for Desktop.

1. Configure the MCP Client: Open your client's MCP configuration file (e.g., claude_desktop_config.json for Claude for Desktop) and add the following server configuration. Important: Replace /path/to/your/project/my_mcp with the absolute path to this project's directory on your system.

   {
     "mcpServers": {
       "mathematica": {
         "command": "uv",
         "args": [
           "--directory",
           "/path/to/your/project/my_mcp",
           "run",
           "wolfram_mathematica.py"
         ],
          "env": {
               "ANIMALID_SECRET_KEY": "default-secret-key-for-dev"
             }
       }
     }
   }
   

   You may need to use the full path to the uv executable in the command field if it's not in your system's PATH. You can find it by running which uv (macOS/Linux) or where uv (Windows).

📚 Documentation

Screenshot