AI-ChatAgent-With-MCP

ChatAgent with Model Context Protocol (MCP)

A .NET-based chat agent implementation that uses OpenAI's models and Model Context Protocol (MCP) for extensible tool integration. This project demonstrates how to create an AI-powered chat agent with custom tool capabilities.

Project Structure

The solution consists of two main projects:

1. ChatAgent: The main application that implements the chat interface using OpenAI and MCP client.
2. MCP.Server: A Model Context Protocol server that provides tools for the chat agent.

Prerequisites

• .NET 9.0
• Visual Studio 2022 or later
• OpenAI API access

Configuration

1. Configure your AI settings in appsettings.json:

{
    "AI": {
        "ModelId": "YOUR-MODEL",
        "APIKey": "YOUR-KEY",
        "Endpoint": "YOUR-ENDPOINT"
    },
    "MCPServers": "PATH-TO-MCP-SERVER"
}

Replace the placeholders with your actual values:

• ModelId: Your OpenAI model ID (e.g., "gpt-4", "gpt-3.5-turbo")
• APIKey: Your OpenAI API key
• Endpoint: Your OpenAI API endpoint (usually "https://api.openai.com")
• MCPServers: Path to the MCP.Server executable (semicolon-separated for multiple servers)

Features

• Interactive chat interface with AI-powered responses
• Integration with Model Context Protocol (MCP) for extensible tools
• Custom tool implementation example (UserTool) that fetches user data
• Support for multiple MCP servers
• Environment-specific configuration support
• Streaming chat responses for better user experience

Technologies Used

• Microsoft.SemanticKernel for AI integration
• Model Context Protocol (MCP) for tool integration
• OpenAI API for chat completion
• .NET 9.0 features
• Microsoft.Extensions.Configuration for configuration management

Getting Started

1. Clone the repository
2. Build the solution
3. Configure your appsettings.json with appropriate AI credentials
4. Run the MCP.Server project
5. Run the ChatAgent project

Usage

1. Start the application
2. Type your messages in the console
3. Use 'EXIT' to end the chat session
4. The agent will respond using both its AI capabilities and available MCP tools

Tool Integration

The project includes a sample tool (UserTool) that demonstrates how to integrate external functionality:

[McpServerToolType]
public class UserTool
{
    [McpServerTool, Description("Tüm kullanıcıları, bilgileriyle birlikte getirir.")]
    public async Task<List<object>> GetUsersAsync()
    {
        // Implementation
    }
}

You can add more tools by creating new classes with the [McpServerToolType] attribute and methods with [McpServerTool] attribute.

Development

• Use appsettings.Development.json for development-specific settings
• The project supports environment-specific configurations through DOTNET_ENVIRONMENT variable
• Tool development should follow the MCP protocol specifications

Contributing

Feel free to contribute to this project by:

1. Forking the repository
2. Creating a feature branch
3. Submitting a pull request