mcp-master 0.0.2

A Python package for master MCP servers.

mcp-master

A Python package for master MCP servers.

Python implementation of the Model Context Protocol (MCP)

Table of Contents

• MCP Master
  • Overview
  • Installation
  • Quickstart
    • Master MCP Servers
    • Master MCP Clients
  • License

Overview

The MCP Master package allows the creation of intermediaries between clients and MCP servers. This offloads tool-related reasoning and logic to the master server without requiring client-side modifications. This Python package provides full master MCP support, making it easy to:

• Build MCP master servers that connect clients to multiple MCP servers through the Streamable HTTP transport
• Choose tools via reasoning structures like LangGraph
• Easily scale the available selection of tools

Read about the Model Context Protocol here.

Installation

For projects using pip:

pip install mcp-master

Quickstart

Master MCP Servers

A simple master MCP server that connects to two of the included demo servers:

# master_server.py
from mcp_master import MasterMCPServer
import asyncio
from os import getenv

set_config({
    'judge_model_id': "us.anthropic.claude-3-5-haiku-20241022-v1:0",
    'judge_model_service_url': '', # Set this to where your judge LLM is hosted
    'OPENAI_API_KEY': getenv('OPENAI_API_KEY'), 
    'OPENAI_BASE_URL': getenv('OPENAI_BASE_URL'), # Set this to where your other LLMs will be hosted
})

# Create an MCP server on port 3000 with two test servers
# Ensure both test servers are running by starting them in the terminal before starting master_server.py
server = MasterMCPServer(3000, [
    ("http://localhost:8091/mcp", 'test_server_1'),
    ("http://localhost:8092/mcp", 'test_server_2')
])
asyncio.run(server.startup())

You can run this server in the terminal by running:

python master_server.py

Master MCP Clients

MCP clients utilizing the streamable-HTTP protocol can connect to master MCP servers without any modification. Details provided in the URL.

For a complete working example, see the MCP GitHub repository.

License

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