Continuity Protocol (MCP)

A comprehensive implementation of the Model Context Protocol (MCP) for maintaining continuity in large language model interactions.

About

The Continuity Protocol is a framework designed to enable persistent context and session management for AI assistants. It allows for:

• Seamless resumption of conversations
• Context awareness across multiple sessions
• Tool management with JSON-RPC 2.0
• Data persistence and recovery
• Context detection and management

Features

• Server Component: Implements the JSON-RPC 2.0 protocol for tool registration and execution
• Transport Layer: HTTP server for handling requests
• Tool Management: System for registering, describing and executing tools
• Context Tools: Store and retrieve contextual information
• Session Tools: Manage conversation sessions with versioning
• System Tools: Monitor system health and perform optimization

Getting Started

Installation

# Clone the repository
git clone https://github.com/yourusername/continuity-protocol.git
cd continuity-protocol

# Install dependencies
pip install -e .

Basic Usage

from continuity_protocol.server import MCPServer
from continuity_protocol.transport.http import start_server

# Create a server
server = MCPServer("My-MCP-Server")

# Register a tool
def add(a, b):
    return {"result": a + b}

server.register_tool("add", add, {
    "description": "Add two numbers",
    "parameters": {
        "type": "object",
        "properties": {
            "a": {"type": "number"},
            "b": {"type": "number"}
        },
        "required": ["a", "b"]
    }
})

# Register default tools (context, session, system)
server.register_default_tools()

# Start the server (default port: 8000)
start_server(server)

Client Example

import requests
import json

# Create a request
request = {
    "jsonrpc": "2.0",
    "id": "test-1",
    "method": "execute",
    "params": {
        "tool": "add",
        "parameters": {
            "a": 5,
            "b": 7
        }
    }
}

# Send the request
response = requests.post(
    "http://localhost:8000/mcp",
    json=request,
    headers={"Content-Type": "application/json"}
)

# Parse response
result = response.json()
print(result)  # {"jsonrpc": "2.0", "id": "test-1", "result": {"result": 12}}

Testing

Run tests using the provided test runner:

# Run all tests
./run_tests.sh --all

# Run specific test types
./run_tests.sh --unit
./run_tests.sh --integration

# Generate coverage report
./run_tests.sh --all --coverage

License

MIT

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.