Where can I find the source code?

The source code is available on GitHub at https://github.com/philschmid/mcp-openai-gemini-llama-example.

Which LLMs are supported?

The example includes implementations for Meta Llama 3 and Google Gemini. It can also be adapted for use with OpenAI models.

What are the prerequisites to use this example?

You need Docker, a Hugging Face account with an access token (for Llama 3), and a Google API key (for Gemini).

What does this tool do?

This tool provides a basic example of connecting to an MCP server, loading tools, converting them into LLM-compatible functions, and interacting with an LLM using the `openai` or `google-genai` SDKs.

OpenAI Gemini Llama Example

Name: OpenAI Gemini Llama Example
Author: philschmid

byphilschmid

•

180

•

Demonstrates building an AI agent using the Model Context Protocol (MCP) with open LLMs like Meta Llama 3, OpenAI, or Google Gemini, and a SQLite database.

This repository provides a basic example of constructing an AI agent leveraging the Model Context Protocol (MCP) in conjunction with open LLMs, specifically Meta Llama 3, OpenAI, and Google Gemini, along with a SQLite database. It serves as a straightforward, educational illustration focused on connecting to an MCP server, loading and utilizing tools and resources, transforming tools into LLM-compatible function calls, and interacting with an LLM via the `openai` or `google-genai` SDK. The code sets up a simple CLI agent to interact with a SQLite database through an MCP server.

Key Features

01Connects to an MCP server

02Loads and uses tools and resources from the MCP server

03Converts tools into LLM-compatible function calls

04Interacts with an LLM using the `openai` SDK or `google-genai` SDK

05Includes examples for Llama 3 and Gemini models

06145 GitHub stars

Use Cases

01Building AI agents that interact with databases

02Demonstrating the use of MCP with various LLMs

03Educational purposes for understanding MCP implementation