What is Mcp Llms Micropython

mcp-LLMs-Micropython is a MicroPython library designed to provide a standardized interface for interacting with various Large Language Model (LLM) APIs, including OpenAI, Claude, and Gemini. It simplifies the process of switching between different LLM providers.

Use cases

Use cases include developing chatbots that can switch between different LLMs, creating educational applications that leverage multiple AI models, and building IoT devices that require language processing capabilities.

How to use

To use mcp-LLMs-Micropython, clone the repository, copy the files to your project or MicroPython device, ensure internet connectivity, and utilize the MCPFactory to create adapters for the desired LLM provider.

Key features

Key features include a unified interface for all LLM providers, a factory pattern for easy adapter creation, support for function calls across models, compatibility with MicroPython for resource-constrained environments, and robust error handling tailored to each API’s peculiarities.

Where to use

mcp-LLMs-Micropython can be used in various fields such as IoT applications, educational tools, chatbots, and any other projects that require interaction with LLM APIs in a resource-efficient manner.

MCP Micropython - Message Chat Protocol for LLMs

Description

MCP (Message Chat Protocol) is a MicroPython library that provides a standard and unified interface for interacting with different Large Language Model (LLM) APIs. It allows you to use OpenAI, Claude, and Gemini with a consistent interface, simplifying the process of switching between different providers.

Features

• Unified Interface: Same usage pattern for all LLM providers
• Factory Pattern: Easy creation of adapters through a factory
• Function Calling: Support for function calls across all models
• MicroPython Compatible: Specifically designed for resource-constrained environments
• Robust Error Handling: Adapted for the peculiarities of each API

Requirements

• MicroPython 1.17 or higher
• Working Internet connection
• API keys for the providers you wish to use (OpenAI, Claude, and/or Gemini)

Installation

1. Clone this repository:

git clone https://github.com/NickyDark1/mcp-LLMs-Micropython.git

2. Copy the files to your project directory or to your MicroPython device’s filesystem.

3. Make sure your device has internet connectivity configured.

Project Structure

mcp-protocol/
├── mcp_base.py            # Abstract base class for adapters
├── openai_mcp_adapter.py  # Adapter for OpenAI
├── claude_mcp_adapter.py  # Adapter for Claude (Anthropic)
├── gemini_mcp_adapter.py  # Adapter for Gemini (Google)
├── mcp_factory.py         # Factory for creating adapters
├── main_mcp.py            # Usage example
├── network_iot.py         # Utility for setting up internet connection
└── tools.py               # Example functions for function calling

Basic Usage

from mcp_factory import MCPFactory
from tools import functions_list_data

# Create an adapter for any provider using the Factory
adapter = MCPFactory.create_adapter(
    provider="openai",  # Can be "openai", "claude", or "gemini"
    api_key="your-api-key-here",
    modelo="gpt-3.5-turbo",  # Optional, each provider has defaults
    max_tokens=50,
    temperatura=0.7
)

# Add messages to the conversation history
adapter.agregar_mensaje("system", "You are an assistant that helps with math operations.")
adapter.agregar_mensaje("user", "I want to multiply 4 and 5")

# Make a query with function calling
respuesta = adapter.consultar(functions=functions_list_data, function_call="auto")

# Process the response
if respuesta["type"] == "function_call":
    # Handle function call
    funcion = respuesta["name"]
    argumentos = json.loads(respuesta["arguments"])
    print(f"Function call: {funcion}({argumentos})")
    
    # You can execute the function and add the result to the history
    resultado = ejecutar_funcion(funcion, argumentos)
    adapter.agregar_mensaje("assistant", f"The result is: {resultado}")
    
    # Continue the conversation
    respuesta_final = adapter.consultar()
    print(respuesta_final["content"])
else:
    # It's a normal text response
    print(respuesta["content"])

Examples

Complete Usage Example

Check the main_mcp.py file for a complete example of how to use the adapters with all three providers and process the responses.

Switching Between Providers

# Use OpenAI
openai_adapter = MCPFactory.create_adapter("openai", OPENAI_API_KEY)
openai_adapter.agregar_mensaje("user", "Hello, can you help me with math?")
respuesta = openai_adapter.consultar()

# Use Claude
claude_adapter = MCPFactory.create_adapter("claude", CLAUDE_API_KEY)
claude_adapter.agregar_mensaje("user", "Hello, can you help me with math?")
respuesta = claude_adapter.consultar()

# Use Gemini
gemini_adapter = MCPFactory.create_adapter("gemini", GEMINI_API_KEY)
gemini_adapter.agregar_mensaje("user", "Hello, can you help me with math?")
respuesta = gemini_adapter.consultar()

Specifying Specific Models

# OpenAI with GPT-4
adapter = MCPFactory.create_adapter("openai", OPENAI_API_KEY, "gpt-4")

# Claude with the latest model
adapter = MCPFactory.create_adapter("claude", CLAUDE_API_KEY, "claude-3-7-sonnet-20250219")

# Gemini with specific model
adapter = MCPFactory.create_adapter("gemini", GEMINI_API_KEY, "gemini-2.0-flash")

Customization

You can extend MCP to support other LLM providers by creating new adapters:

1. Create a new class that inherits from MCPAdapter
2. Implement the required methods (_realizar_peticion, _procesar_respuesta, _convertir_funciones)
3. Add your adapter to the factory in mcp_factory.py

Donations

If you find this project useful, please consider making a donation to support its development:

Ethereum (ETH): 0x691C9bbcA4Cc3335bdb21396817ed35b36f7cC0b

Your support is greatly appreciated and helps us continue improving the project.

Developed with ❤️ for the MicroPython community