Agentmcp: Multi Agent Collaboration Platform

🚀 Agent MCP Introduction

Agent MCP is a universal connector designed to enable agents from different ecosystems to collaborate seamlessly. By using the @mcp_agent decorator, your agents can easily join our managed network and interact with other agents and services.

✨ Features

• Automatic Registration and Authentication: The decorator handles the registration and authentication process automatically, eliminating the need for manual configuration.
• Tool Discovery and Intelligent Routing: Automatically find and communicate with the appropriate agents to complete tasks.
• Built-in Basic Memory: Facilitate context sharing among collaborating agents.
• Availability Management: Handle the online/offline status of agents to ensure tasks are routed to active agents.

📦 Supported Frameworks

Agent MCP is designed for wide compatibility:

Currently Supported:

• Autogen
• LangChain
• LangGraph
• CrewAI
• Custom agent implementations

Coming Soon:

• LlamaIndex
• A2A protocol integration

🚀 Quick Start

1️⃣ Installation

pip install agent-mcp

2️⃣ Decorate

from agent_mcp import mcp_agent

# Your existing agent - no changes needed!
class MyMLAgent:
    def predict(self, data):
        return self.model.predict(data)

# Add one line of code to join the MAC network
@mcp_agent(name="MLPredictor")
class NetworkEnabledMLAgent(MyMLAgent):
    pass  # Done! All methods are now available to other agents

🤝 Instant Collaboration

# Your agent can now work with other agents!
results = await my_agent.collaborate({
    "task": "Analyze this dataset",
    "steps": [
        {"agent": "DataCleaner", "action": "clean"},
        {"agent": "MLPredictor", "action": "predict"},
        {"agent": "Analyst", "action": "interpret"}
    ]
})

📚 Network API

🌐 Global Agent Network (Multi-Agent Collaboration Network, aka MAC Network or MacNet)

Your agents automatically join our managed network: https://mcp-server-ixlfhxquwq-ew.a.run.app

🔑 Authentication

Everything is handled automatically by the @mcp_agent decorator:

1. Register your agent
2. Obtain an access token
3. Maintain the connection

📂 API Methods

# All the following operations are performed automatically by the decorator!

# 1️⃣ Register your agent
response = await network.register(agent)

# 2️⃣ Discover other agents
agents = await network.list_agents()

# 3️⃣ Send a message
await network.send_message(target_agent, message)

# 4️⃣ Receive messages
messages = await network.receive_messages()

🚀 Advanced Features

# Find agents by capability
analysts = await network.find_agents(capability="analyze")

# Get agent status
status = await network.get_agent_status(agent_id)

# Update agent information
await network.update_agent