Python MCP Clients: Framework for LLM-Driven Tool Orchestration

mcp-clients is an open-source Python SDK designed for building and orchestrating LLM-powered tools using the Model Context Protocol (MCP). It allows seamless integration with OpenAI and Gemini, enabling natural language interactions with databases, file systems, APIs, and more. Developers can easily connect to multiple MCP servers through a unified interface and create intelligent agents that interact with real-world tools. It is free to use, modify, and distribute under the MIT license.

Features

• AI Models Integration: Built-in support for Google's Gemini and OpenAI models
• MCP Protocol Support: Seamless integration with MCP servers
• Tool Calling: Automatic tool discovery and execution
• Interactive Chat: Built-in chat interface with conversation history
• Customizable: Support for custom chat loops and system instructions
• Easy Setup: Simple configuration with environment variables
• Async/Await: Fully asynchronous for optimal performance

Installation

Install using pip:

pip install mcp-clients

Or install from uv:

uv add mcp-clients

Configuration

Environment Variables

Create a .env file in your project root:

# Required: Your API key
YOUR_API_KEY=your_api_key_here

# Required: Your MCP server path
MCP_SERVER=/path/to/your/mcp_server.py

Quick Start

Basic Usage with OpenAI

import asyncio
from dotenv import load_dotenv

from mcp_clients import OpenAI

load_dotenv()


async def main():
    client = await OpenAI.init(
        server_script_path="path_to_server_script",
    )
    try:
        await client.chat_loop()
    finally:
        await client.cleanup()


if __name__ == "__main__":
    asyncio.run(main())

Basic Usage with Gemini

import asyncio
from dotenv import load_dotenv

from mcp_clients import Gemini

load_dotenv()


async def main():
    client = await Gemini.init(
        server_script_path="path_to_server_script",
    )
    try:
        await client.chat_loop()
    finally:
        await client.cleanup()


if __name__ == "__main__":
    asyncio.run(main())

Custom Chat Loop

async def custom_chat_handler(client):
    """Custom chat loop with enhanced features"""
    print("Enhanced Chat Started!")
    print("Commands: 'help', 'history', 'clear', 'quit'")
    
    while True:
        try:
            query = input("\n💬 You: ").strip()
            
            if query.lower() == 'quit':
                break
            elif query.lower() == 'help':
                print("Available commands: help, history, clear, quit")
                continue
            elif query.lower() == 'history':
                print(f"Conversation has {len(client.history)} messages")
                continue
            elif query.lower() == 'clear':
                client.history = []
                print("Chat history cleared!")
                continue
                
            response = await client.process_query(query)
            print(f"🤖 Assistant: {response}")
            
        except KeyboardInterrupt:
            break
        except Exception as e:
            print(f"Error: {e}")

# Use the custom chat loop
async def main():
    client = await Gemini.init(
        server_script_path='weather_server.py',
        custom_chat_loop=custom_chat_handler
    )
    
    try:
        await client.chat_loop()
    finally:
        await client.cleanup()

API Reference

Gemini Class

The Gemini client class for interacting with Gemini AI through MCP servers.

Initialization

client = await Gemini.init(
    api_key=None,                    # Gemini API key (or use env var)
    server_script_path=None,         # Path to MCP server script
    model="gemini-2.5-flash",        # Gemini model to use
    system_instruction=None,         # Custom system instruction
    custom_chat_loop=None            # Custom chat loop function
)

OpenAI Class

The OpenAI client class for interacting with Gemini AI through MCP servers.

Initialization

client = await OpenAI.init(
    api_key=None,                    # Gemini API key (or use env var)
    server_script_path=None,         # Path to MCP server script
    model="gpt-4.1-nano",            # OpenAI model to use
    system_instruction=None,         # Custom system instruction
    custom_chat_loop=None            # Custom chat loop function
)

Methods

• process_query(query: str) -> str: Process a single query
• chat_loop(): Start interactive chat session
• cleanup(): Clean up resources (always call this!)
• connect_to_server(): Manually connect to MCP server

🔍 Troubleshooting

Common Issues

1. API Key Errors

   Error: Invalid API key
   

   • Ensure your API key is correct
   • Check that the API key is properly set in your environment

2. Server Connection Issues

   Error: Server script must be a .py or .js file
   

   • Verify your MCP server script path is correct
   • Ensure the file has the proper extension (.py or .js)

3. Import Errors

   ModuleNotFoundError: No module named 'mcp_clients'
   

   • Install the package: pip install mcp-clients or uv add mcp-clients
   • If installing from source: pip install -e .

Examples

Check out the examples/ directory for more usage examples:

• gemini_client: Simple chat with MCP tools using Gemini.
• openai_client: Simple chat with MCP tools using OpenAI.

Contributing

I welcome contributions! Here's how you can help:

Getting Started

1. Fork the repository

   git clone https://github.com/faizrazadec/mcp-clients.git
   cd mcp-clients
   

2. Set up development environment

   uv venv
   source .venv/bin/activate  # On Windows: venv\Scripts\activate
   uv sync
   

3. Create a feature branch

   git checkout -b feature/your-feature-name
   

Development Guidelines

• Code Style: Follow PEP 8 and use black for formatting
• Type Hints: Add type hints to all functions and methods
• Documentation: Add docstrings and update README if needed
• Testing: Write tests for new features (pytest)
• Commits: Use conventional commit messages

Types of Contributions

• Bug Fixes: Fix issues and improve stability
• New Features: Add new models, tools, or capabilities
• Documentation: Improve docs, examples, and tutorials
• Testing: Add tests and improve test coverage
• UI/UX: Improve user experience and interfaces

Submitting Changes

1. Run tests (when available)

   pytest
   

2. Format code

   cd mcp_clients/
   black .
   

3. Submit a pull request

   • Describe your changes clearly
   • Link any related issues
   • Include examples if applicable

Code of Conduct

Please be respectful and inclusive. We're building this together! 🌟

License

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

Acknowledgments

• Anthropic for the Model Context Protocol specification
• FastMCP for the excellent MCP server framework
• Contributors who help make this project better

---

Made with ❤️ by Muhammad Faiz Raza

If you find this project helpful, please consider giving it a ⭐ on GitHub!