Booner MCP

🚀 Booner_MCP

An AI Infrastructure-as-Code platform based on Model Context Protocol (MCP), leveraging Ollama for local server management and intelligent coding.

🚀 Quick Start

Booner_MCP is an AI Infrastructure-as-Code platform that enables AI agents to interact with local infrastructure. It deploys and manages various types of servers (web servers, game servers, databases) through the Model Context Protocol. The platform integrates a local Ollama deployment running on powerful hardware with the Mixtral model.

✨ Features

System Architecture

Machine Configuration

• Machine 1 (Booner_Ollama - 10.0.0.10):

  • Runs the Ollama LLM server on port 11434.
  • Provides AI capabilities for the MCP system.
  • Hardware: AMD 5700X3D processor, NVIDIA 4070 Ti Super graphics card, 64GB of RAM, Quadro P4000 graphics card.

• Machine 2 (Booner_MCP & agent - 10.0.0.4):

  • Runs the MCP core management API.
  • Runs the Next.js web interface.
  • Hardware: Ryzen 7 5700X3D processor, 32GB of RAM.
  • Serves as an agent worker node for intelligent coding.

• Machine 3 (booner-mcp-web & agent - 10.0.0.4):

  • Runs the Web GUI to interact with MCP and Ollama.
  • Serves as an agent worker node for intelligent coding.

• Machine 3 (OPN_IaC - 10.0.0.2):

  • Runs infrastructure-as-code tools.
  • Features a dedicated OPNSense network API.

• Deployment Targets (Machine N):

  • Runs game servers, web applications, etc.
  • Example hardware: Intel Xeon E5 - 2680 v4 processor, 16GB of RAM, NVIDIA RTX 3050 8GB graphics card.

• Storage: TrueNAS system with 8TB HDD and 2TB SSD (shared via NFS).

Software Stack

• Operating System: Ubuntu 24 (on all machines).
• LLM: Mixtral model running via Ollama.
• Primary Languages: Python, Go, NextJS.
• Containerization: Docker.

Project Structure

• agents/: AI agent definitions and orchestration code.
• servers/: MCP server implementations for different infrastructure tasks.
• api/: API server for communication with agents.
• config/: Configuration files for different environments and systems.

📦 Installation

Prerequisites

• Git with submodule support.
• Docker and Docker Compose.
• Node.js 18+ (for local development).
• Python 3.11+ (for local development).

Initial Setup

1. Clone the repository and recursively clone submodules:

   git clone --recurse-submodules https://github.com/vespo92/Booner_MCP.git
   cd Booner_MCP
   

2. Create an environment file:

   cp .env.example .env
   

3. Generate a secure AUTH_SECRET:

   # On Linux/macOS
   ./generate_auth_secret.sh
   
   # On Windows
   .\generate_auth_secret.ps1
   

4. Deploy using Docker Compose:

   docker-compose up -d
   

Accessing Services

• Web UI: http://10.0.0.1:3000
• API: http://10.0.0.1:8000
• Ollama: http://10.0.0.10:11434

💻 Usage Examples

Basic Usage

cd Booner_MCP
# Make code modifications and conduct development activities here