Skynet MCP

🚀 SkyNet-MCP (This project is under development)

A hierarchical network of AI agents based on the Model Context Protocol (MCP). AI agents in this network can generate new agents to perform tasks, and each agent also includes all the tools of the initiating agent.

This agent network can handle various complex tasks, from research and reporting to coding.

If you like this idea and are proficient in TypeScript, feel free to contact me. I welcome your contributions!

🚀 Quick Start

SkyNet-MCP is an advanced architecture that implements the Model Context Protocol (MCP) to create a hierarchical network of AI agents. Each SkyNet-MCP instance runs as both an MCP server and an MCP client, allowing it to provide tools to higher-level agents while also being able to generate and manage sub-agents.

This architecture supports a recursive agent network, which can decompose complex tasks, execute work in parallel, and integrate specialized capabilities among multiple models and services.

✨ Features

• Dual-mode operation: Run as both an MCP server and a client simultaneously.
• Large language model integration: Connect to OpenAI or Anthropic models for intelligent decision-making.
• Tool discovery: Automatically discover and use tools on connected MCP servers.
• Hierarchical agent management: Create and manage sub-agents.
• Flexible configuration: Support a configuration system for multiple environments.
• FastMCP integration: Implemented based on the efficient FastMCP framework for a powerful MCP server implementation.
• Multiple transmission options: Support STDIO and SSE (Server-Sent Events) transmissions.
• Delayed execution: Run tasks asynchronously and check results later.

📦 MCP Tools

SkyNet-MCP provides the following MCP tools through its FastMCP implementation:

1. SkyNet Agent: An agent for generating, controlling, and communicating in the MCP network.

Configuration System

SkyNet-MCP uses a flexible configuration system that can load settings from the following sources:

• Default values
• Environment variables
• JSON configuration files for specific environments

Configuration files are stored in the config/ directory and are loaded based on the NODE_ENV environment variable.

💻 Usage Examples

Project Structure

sky-net-mcp/
├── src/                  # Source code directory
│   ├── server.js        # MCP server entry file
│   ├── client.js        # MCP client implementation
│   └── tools/           # Tool modules, including interfaces for interacting with OpenAI and Anthropic
├── config/              # Configuration file directory
│   ├── default.json     # Default configuration file
│   ├── development.json  # Development environment configuration file
│   └── production.json   # Production environment configuration file
├── Dockerfile          # Docker image build file
└── docker-compose.yml  # Docker container orchestration configuration file

Environment Variables

The key environment variables include:

• SERVER_PORT - The port the server listens on.
• SERVER_HOST - The host the server binds to.
• LOG_LEVEL - The logging level (debug, info, warn, error).
• PERSISTENCE_TYPE - The storage type (memory or redis).
• REDIS_URL - The Redis connection URL (when using Redis).
• OPENAI_API_KEY - The OpenAI API key (for large language model agents).
• ANTHROPIC_API_KEY - The Anthropic API key (for large language model agents).

📚 Documentation

CI/CD

SkyNet-MCP uses GitHub Actions for continuous integration and delivery.

Workflow:

1. Run on pushes to the main branch and pull requests.
2. Set up the Node.js environment.
3. Install dependencies.
4. Run code checks and formatting checks.
5. Build the project.
6. Execute all tests.
7. Build and validate the Docker image.

The CI/CD pipeline ensures code quality and stability by automatically testing changes before merging them.

📄 License

ISC