MCP-Hive

A desktop application for the Model Context Protocol (MCP) with an integrated Python backend packaged as an executable.

Overview

MCP-Hive is a desktop application that integrates with various AI model providers through the Model Context Protocol. The application consists of:

1. An Electron-based frontend for the user interface
2. A Python backend (packaged as an executable) for handling the AI model connections and processing

Building and Packaging

Prerequisites

• Node.js and npm
• Python 3.10+ with pip
• PyInstaller (pip install pyinstaller)
• Git

Installation Steps

1. Clone the repository:

   git clone https://github.com/yourusername/MCP-Hive.git
   cd MCP-Hive
   

2. Install Python dependencies for the backend:

   cd Hive
   pip install -r requirements.txt
   cd ..
   

3. Install Node.js dependencies for the frontend:

   cd MCP-Hive-Desktop
   npm install
   cd ..
   

Building the Executable Backend

The backend is packaged as an executable using PyInstaller:

cd Hive
python build_executable.py
cd ..

This script:

• Builds the Python backend as a standalone executable
• Copies the executable and necessary resources to the Electron resources directory

Building the Electron App with Packaged Backend

After building the backend, build the complete Electron app:

cd MCP-Hive-Desktop
npm run build

This command:

1. Runs the backend build script first (npm run build:backend)
2. Packages the Electron app with the executable backend (npm run build:app)

The packaged application will be available in the MCP-Hive-Desktop/dist directory.

Development

For development, you can run the app without packaging:

cd MCP-Hive-Desktop
npm start

This will start the Electron app and use the executable backend if available, or fall back to using the Python interpreter if needed.

Configuration

API Keys

Set up your AI provider API keys in the .env file in the Hive directory:

GOOGLE_API_KEY=your_google_api_key_here
GROQ_API_KEY=your_groq_api_key_here
DEFAULT_LLM_PROVIDER=gemini

MCP Servers

Configure MCP servers through the application's settings interface or by directly editing the Mcphive_config.json file.

License

MIT License