MCP Demo: Model Context Protocol Integration With Openai

🚀 MCP-Demo: Model Context Protocol Integration with OpenAI Demo

This is a comprehensive example demonstrating the Model Context Protocol (MCP) in a .NET application. It shows how to create MCP clients and servers and integrate them with OpenAI's Large Language Model (LLM) capabilities.

🚀 Quick Start

What is MCP?

Model Context Protocol (MCP) is an open protocol that standardizes how applications provide context to large language models. It enables secure integration between LLMs and various data sources and tools, and provides a standard way to extend LLM capabilities.

This demo showcases:

• An MCP server with custom tools
• A simple client for testing the tools
• Integration with OpenAI to use these tools in conversations

Solution Structure

MCP-Demo/
├── MCP.Server/        # MCP server with custom tools
├── MCP.Client/        # Simple test client
├── MCP.LlmIntegration/ # OpenAI integration with MCP tools
└── README.md          # This document

Features

MCP Server

It provides several function-based tools:

• Echo: Returns a greeting with the input message.
• Add: Adds two numerical values.
• GetDateTime: Returns the current date and time.

OpenAI Integration

• Connects to the OpenAI API using an OpenAI API key.
• Maintains conversation context.
• Extracts tool calls from AI responses.
• Executes tools via the MCP server.
• Returns tool results to the user.

Prerequisites

• .NET 9.0 SDK
• OpenAI API key
• A code editor such as Visual Studio, VS Code, or JetBrains Rider

Installation

1. Clone the repository

   git clone https://your-repository-url/MCP-Demo.git
   cd MCP-Demo
   

2. Build the solution

   dotnet build
   

3. Set the OpenAI API key

   cd MCP.LlmIntegration
   dotnet user-secrets set "OpenAI:ApiKey" "your-api-key-here"
   

Running the Demo

1. Run the LLM integration application

   cd MCP.LlmIntegration
   dotnet run
   

2. The application will:

   • Automatically start the MCP server.
   • Connect to the OpenAI API.
   • Provide an interactive chat interface.
   • Allow interaction with MCP tools through OpenAI.

💻 Usage Examples

Basic Calculation

You: What is 42 plus 17?

AI: I'll use the Add tool to calculate for you.
[TOOL_CALL:Add(a=42, b=17)]

Result: 42 + 17 = 59

Getting the Current Date and Time

You: What's the current time?

AI: I'll use the GetDateTime tool to provide the current time and date.
[TOOL_CALL:GetDateTime]

Response:
Date: October 5, 2023
Time: 14:30:00

🔧 Technical Details

Architecture

• MCP Server: Responsible for hosting and managing MCP API endpoints, handling tool requests, and returning results.
• OpenAI Integration: Uses OpenAI's .NET client library to interact with their API and coordinates tool calls.

Extending the Project

To extend this project:

1. Implement more custom tools and integrate them with the MCP server.
2. Extend existing tools or create new complex tools to meet broader needs.
3. Combine other AI models and services to enhance functionality.

Troubleshooting

If you encounter issues, refer to the following common solutions:

• API Errors: Check if the key is correctly configured and ensure a stable network connection.
• Tool Call Failures: Verify that the parameters match the tool requirements and ensure that the tools are correctly registered and configured.

📄 License

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

Acknowledgments

• Model Context Protocol for creating the protocol specification.
• OpenAI for providing the .NET client library.
• Microsoft for supporting the .NET framework and hosting libraries.