The Networking Protocol for AI Agents

Discover. Communicate. Collaborate.

Mycelium is the open-source networking protocol designed to connect AI agents, enabling them to discover, communicate, and collaborate with each other. It addresses the problem of isolated AI agents by providing a standardized network, akin to a telephone network for AI.

• Discover other agents using natural language.
• Communicate through a standard protocol.
• Collaborate in multi-agent chains.
• Build trust through successful interactions.

⚡ Quick Start

Install the package using pip:

pip install mycelium-agents

Create your first agent in just a few lines of Python:

• Define an agent with a name and description.
• Use decorators to define event handlers for agent communication.
• Serve the agent to make it discoverable and callable on the network.

Discover and use other agents on the network:

• Initialize a Network object.
• Use network.discover() to find agents based on natural language queries.
• Use network.request() to send requests to discovered agents.

🎨 Spatial Dashboard

A visual control center for your Mycelium network, featuring a glassmorphism design and built with vanilla JavaScript. It allows you to visualize agents, their connections, and network statistics in real-time.

• Visualize agents as orbiting pills around a central registry.
• See live connections represented by filament lines.
• Monitor real-time network statistics.
• Search for agents using natural language (Ctrl+K).
• Send requests directly from the dashboard.
• Inspect agent capabilities and logs.

🌍 Real-World Agents

Includes 5 production-ready agents that leverage live APIs:

• RealWeather: Provides live weather data for any city using OpenWeatherMap.
• CryptoTracker: Offers live Bitcoin and crypto prices via CoinGecko.
• RealTranslator: Translates text into over 50 languages using MyMemory.
• WikiBrain: Accesses Wikipedia knowledge and search capabilities.
• CurrencyMaster: Provides live exchange rates for over 150 currencies using ExchangeRate API.

Demonstrates multi-agent chain capabilities, such as fetching crypto prices and translating them.

📚 Documentation

Comprehensive documentation is available, covering:

• Getting Started guides.
• The full Protocol Specification.
• Architecture details.
• API Reference.
• Frequently Asked Questions (FAQ).

🤝 Contributing

Contributions are welcome! Follow the guidelines in CONTRIBUTING.md. Good first issues include adding new example agents, improving documentation, writing tests, or adding language translations.

📜 License

This project is licensed under the MIT License, encouraging open use, building, and modification.