MyChatApp

A powerful Windows Forms application that enables chat interactions with multiple Large Language Model (LLM) providers and integrates with Model Context Protocol (MCP) servers, similar to Claude Desktop's functionality.

Features

• Multi-Provider Support: Connect to various LLM providers including:

  • OpenAI and OpenAI-compatible APIs
  • Ollama (local models)
  • X.ai (Grok models)
  • Any OpenAI api compatible model

• MCP Server Integration: Leverage Model Context Protocol servers to extend AI capabilities with external tools and data sources. Only Stdio tranport is supported.

• Function Calling: Full support for tool/function calling with automatic tool discovery and execution

• Configurable Settings: Easy JSON-based configuration for providers and MCP servers (Claude desktop mcp configuration works)

• Modern UI: Clean Windows Forms interface built on .NET 9.0

Version

• 1.0.0 - Initial release with core features
• 1.1.0 - Added support for multiple LLM providers, MCP server integration, and function calling
• 1.2.0 - Chat history management
• 1.3.0 - Added function invocation filter and logging

Prerequisites

• Windows OS
• .NET 9.0 Runtime
• Visual Studio 2022 or later (for development)

Installation

1. Clone the repository:

   git clone <repository-url>
   cd MyChatApp
   

2. Build the solution:

   dotnet build
   

3. Configure your settings (see Configuration section below)

4. Run the application:

   dotnet run --project MyChatApp
   

Configuration

LLM Providers

Configure your LLM providers in appSettings.json:

{
  "McpConfigFilePath": "E:\\ws\\claude-desktop\\mcp.json",
  "LLMProviders": [
    {
      "Name": "OpenAI-GPT-4",
      "Type": "OpenAI",
      "Model": "gpt-4",
      "BaseUrl": "https://api.openai.com/v1",
      "ApiKey": "your-api-key-here"
    },
    {
      "Name": "Local-Ollama",
      "Type": "Ollama",
      "Model": "llama2:latest",
      "BaseUrl": "http://localhost:11434",
      "ApiKey": ""
    }
  ]
}

MCP Servers

Configure MCP servers in your mcp.json file:

{
  "McpServers": {
    "filesystem": {
      "Command": "npx",
      "Args": ["-y", "@modelcontextprotocol/server-filesystem", "/path/to/directory"],
      "Env": {}
    },
    "git": {
      "Command": "uvx",
      "Args": ["mcp-server-git", "--repository", "/path/to/repo"],
      "Env": {}
    }
  }
}

Project Structure

MyChatApp/
├── Program.cs              # Application entry point
├── ChatForm.cs             # Main UI form
├── AIChatProviders.cs      # LLM provider management
├── ToolRepository.cs       # MCP server and tool management
├── MyChatAppSettings.cs    # Configuration models
├── appSettings.json        # LLM provider configuration
└── mcp.json               # MCP server configuration

Key Components

AIChatProviders

Manages multiple LLM providers using Microsoft Semantic Kernel framework. Supports:

• Dynamic provider loading
• Function calling with auto-discovery
• Provider-specific prompt execution settings

ToolRepository

Handles MCP server lifecycle and tool management:

• Loads MCP server configurations
• Establishes stdio transport connections
• Discovers and exposes available tools
• Manages tool execution contexts

ChatForm

Main Windows Forms interface providing:

• Provider selection
• Chat history management
• Tool toggle functionality
• Real-time status updates

Dependencies

• Microsoft.SemanticKernel: Core AI orchestration framework
• ModelContextProtocol: MCP client implementation
• Markdig: Markdown processing for chat rendering
• Microsoft.Extensions.Configuration: Configuration management
• Microsoft.Web.WebView2: Enhanced web content rendering

Usage

1. Launch the application and select your preferred LLM provider from the dropdown
2. Enable/disable tools using the tools toggle to control MCP server integration
3. Start chatting with your chosen AI model
4. Leverage MCP tools for enhanced capabilities like file system access, git operations, or custom integrations

Development

Building from Source

# Clone and build
git clone <repository-url>
cd MyChatApp
dotnet restore
dotnet build

# Run in development mode
dotnet run --project MyChatApp

Adding New Providers

1. Add provider configuration to appSettings.json
2. Update AIChatProviders.cs to handle the new provider type
3. Implement provider-specific settings if needed

Adding MCP Servers

1. Install the MCP server (npm, pip, etc.)
2. Add server configuration to mcp.json
3. Restart the application to load the new server

Acknowledgments

• Built with Microsoft Semantic Kernel
• Leverages Model Context Protocol for extensibility
• Inspired by Claude Desktop's MCP integration