Multi-Agent Communication System

A robust multi-agent system for trip planning using ZeroMQ (ZMQ) and the Multi-Agent Communication Protocol (MCP).

Features

• Modern Architecture: Built with ZeroMQ for efficient inter-process communication
• Protocol-Based Communication: Implements MCP (Multi-Agent Communication Protocol) for structured agent interactions
• Specialized Agents:
  • Planner Agent: Coordinates trip planning and manages requests
  • Travel Agent: Provides travel options and handles bookings
  • Hotel Agent: Manages hotel options and reservations
• Conversation Tracking: Maintains conversation history and context
• Error Handling: Robust error handling and logging
• Asynchronous Operations: Built with asyncio for non-blocking operations

Prerequisites

• Python 3.8+
• ZeroMQ (pyzmq)
• Apple Silicon compatible (M1/M2/M3)

Installation

1. Clone the repository:

git clone https://github.com/yourusername/multi-agent-communication.git
cd multi-agent-communication

2. Create and activate a virtual environment:

python -m venv venv
source venv/bin/activate  # On Windows: venv\Scripts\activate

3. Install dependencies:

pip install -r requirements.txt

Usage

1. Start the agents:

python main.py

2. Send a trip request:

import zmq
import json
from agents.mcp_message import MCPMessage, MCPPerformatives

# Create ZMQ context and socket
context = zmq.Context()
socket = context.socket(zmq.PAIR)
socket.connect("tcp://localhost:5555")  # Connect to planner agent

# Create trip request
request = MCPMessage(
    performative=MCPPerformatives.REQUEST,
    content=json.dumps({
        "trip_id": "TRIP-001",
        "destination": "Goa",
        "dates": {
            "check_in": "2024-04-01",
            "check_out": "2024-04-07"
        }
    }),
    sender="client",
    receiver="planner"
)

# Send request
socket.send_json(request.to_json())

Agent Communication Protocol (MCP)

The system uses a structured communication protocol with the following performatives:

• REQUEST: Initial request for action
• INFORM: Provide information
• QUERY: Ask for information
• RESPONSE: Response to query
• PROPOSE: Propose a solution
• ACCEPT: Accept a proposal
• REJECT: Reject a proposal
• FAILURE: Report failure
• CFP: Call for proposals
• CONFIRM: Confirm an action
• DISCONFIRM: Disconfirm an action

Message Format

{
  "protocol": "MCP-1.0",
  "performative": "REQUEST",
  "content": "message content",
  "sender": "agent_id",
  "receiver": "target_agent_id",
  "conversation_id": "uuid",
  "timestamp": "ISO-8601 timestamp"
}

Project Structure

multi-agent-communication/
├── agents/
│   ├── __init__.py
│   ├── base_agent.py      # Base agent implementation
│   ├── planner_agent.py   # Trip planning coordinator
│   ├── travel_agent.py    # Travel options provider
│   ├── hotel_agent.py     # Hotel options provider
│   └── mcp_message.py     # MCP message implementation
├── config.py              # Configuration settings
├── main.py               # Application entry point
├── requirements.txt      # Project dependencies
└── README.md            # Project documentation

Supported Destinations

Currently supported destinations with pre-configured options:

Goa

• Travel Options:
  • Air India Flight (₹5000, 2h)
  • Konkan Express Train (₹2000, 12h)
  • Luxury AC Bus (₹1500, 14h)
• Hotel Options:
  • Taj Exotica (Luxury, ₹15000/night)
  • Holiday Inn (Mid-range, ₹8000/night)
  • Beach Resort (Budget, ₹4000/night)

Mumbai

• Travel Options:
  • IndiGo Flight (₹4000, 1.5h)
  • Rajdhani Express Train (₹1800, 10h)
  • Deluxe Bus (₹1200, 12h)
• Hotel Options:
  • Taj Mahal Palace (Luxury, ₹20000/night)
  • Grand Hyatt (Luxury, ₹18000/night)
  • Comfort Inn (Mid-range, ₹6000/night)

Contributing

1. Fork the repository
2. Create a feature branch
3. Commit your changes
4. Push to the branch
5. Create a Pull Request

License

This project is licensed under the MIT License - see the LICENSE file for details.

Acknowledgments

• ZeroMQ for the communication framework
• Python asyncio for asynchronous operations
• The multi-agent systems community for inspiration