Koi MCP

🚀 KOI-MCP

KOI-MCP is a framework integrating a distributed knowledge network based on KOI-Net with the Model Context Protocol (MCP). It aims to combine decentralized knowledge processing capabilities with advanced Large Language Model (LLM) tools to build a more intelligent and interconnected AI ecosystem.

🚀 Quick Start

KOI-MCP is a framework that combines the distributed knowledge processing capabilities of the KOI network with the standardized interfaces of LLM tools in the MCP protocol. This allows the knowledge network to integrate directly with various LLM tools, enabling more intelligent application scenarios.

✨ Features

Knowledge Processing Pipeline

Connects the distributed knowledge network with LLM tools through KOI's knowledge object processor.

@processor.register_handler(HandlerType.Bundle, rid_types=[AgentPersonality])
def personality_bundle_handler(proc: ProcessorInterface, kobj: KnowledgeObject):
    try:
        profile = PersonalityProfile.model_validate(kobj.contents)
        if mcp_adapter is not None:
            mcp_adapter.register_agent(profile)
        return kobj
    except ValidationError:
        return STOP_CHAIN

Personality Model

Defines and processes the personality traits of agents, allowing for dynamic updates.

{
  "traits": {
    "mood": "enthusiastic",
    "new_capability": {
      "description": "A new capability added at runtime",
      "is_callable": true
    }
  }
}

MCP Adapter

Bridges the KOI network with the MCP protocol, providing RESTful API interfaces.

# Run the MCP registration server
python -m koi_mcp.server.registry --port=9000

# Start the agent server
python -m koi_mcp.server.agent --name=helpful-agent --port=8101

📦 Installation

pip install koi-mcp

💻 Usage Examples

Create a New Knowledge Node

from koi_net.node import KoiNode

node = KoiNode(name="my-koi-node", contact_url="http://localhost:8000/koi-net")
node.start()

Define and Process Personality Traits

{
  "agent": {
    "name": "helpful-agent",
    "personality": {
      "traits": {
        "openness": 0.8,
        "conscientiousness": 0.9,
        "extraversion": 0.7,
        "agreeableness": 0.6,
        "neuroticism": 0.5
      }
    }
  }
}

MCP Interface Definition

📚 Documentation

What is KOI?

KOI is a distributed knowledge network protocol for organizing and sharing trusted knowledge on the blockchain. Through the KOI-Net library, developers can create decentralized knowledge nodes, process knowledge objects, and implement distributed collaboration.

Project Goals

The goal of KOI-MCP is to combine the distributed knowledge processing capabilities of the KOI network with the standardized interfaces of LLM tools in the MCP protocol. In this way, the knowledge network can directly integrate with various LLM tools to build more intelligent application scenarios.

Architecture Overview

graph LR
    A[Knowledge Node] --> B(KOI-NET)
    B --> C(Model Context Protocol Adapter)
    C --> D[LLM Tools]

Running Tests

# Initialize the database
python -m koi_mcp.utils.db init --db-path=./data/db.sqlite

# Start the services
python -m koi_mcp.server.registry --port=9000 &
python -m koi_mcp.server.agent --name=test-agent --port=8102 &

Project Structure

koi-mcp/
├── koi_mcp/
│   ├── adapters/          # MCP protocol adapters
│   ├── handlers/         # Processor modules
│   ├── models/           # Data model definitions
│   └── utils/           # Utility functions
└── README.md             # Project documentation

📄 License

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

Acknowledgments

Thanks to the following projects for their contributions:

• KOI-Net: Distributed knowledge network protocol.
• MCP: Model Context Protocol framework.

For more information, please visit our documentation page or join the developer community.