POC-MCP

ballombert/POC-MCP

3.2

If you are the rightful owner of POC-MCP and would like to certify it and/or have it hosted online, please leave a comment on the right or send an email to dayong@mcphub.com.

A proof of concept implementation of a Model Context Protocol (MCP) server using .NET 9.0 and the official Microsoft ModelContextProtocol SDK.

Tools
1
Resources
0
Prompts
0

POC-MCP - .NET Model Context Protocol Server

A proof of concept implementation of an MCP (Model Context Protocol) server using .NET 9.0 and the official Microsoft ModelContextProtocol SDK.

🚀 GitHub Copilot Ready! This project includes pre-configured MCP integration for VS Code. Open the project and ask Copilot: "Launch a Pomodoro timer with graphical interface" to see MCP in action!

Overview

This project demonstrates how to create and distribute MCP servers using C# and .NET. The server exposes tools that AI clients (like GitHub Copilot, Claude, etc.) can invoke through the standardized MCP protocol.

What is MCP?

Model Context Protocol (MCP) is an open standard that enables AI assistants to securely connect to external data sources and tools. It provides a standardized way for AI models to interact with various resources through a common interface.

Project Structure

POC-MCP/
├── .github/
│   └── copilot-instructions.md    # AI coding agent guidelines
├── docs/                          # Comprehensive documentation
│   ├── README.md                  # Documentation overview
│   ├── getting-started.md         # Setup and quick start guide
│   ├── architecture.md            # Technical architecture details
│   ├── api-reference.md           # MCP tools and API documentation
│   ├── deployment.md              # Build, versioning, and deployment guide
│   ├── adr/                       # Architecture Decision Records
│   │   ├── README.md              # ADR index and guidelines
│   │   ├── 001-dotnet-9-mcp-server.md
│   │   ├── 002-stdio-transport.md
│   │   ├── 003-attribute-based-tools.md
│   │   └── 004-self-contained-distribution.md
│   └── examples/                  # Code examples and patterns
├── scripts/
│   └── version.sh                 # Version management script
├── src/
│   └── SampleMcpServer/           # Main MCP server implementation
│       ├── .mcp/
│       │   └── server.json        # MCP server metadata
│       ├── Tools/
│       │   └── RandomNumberTools.cs # Example MCP tools
│       ├── Program.cs             # Application entry point
│       ├── SampleMcpServer.csproj # Project configuration
│       └── README.md              # Server-specific documentation
├── Directory.Build.props          # Centralized build properties and versioning
├── global.json                   # .NET SDK version specification
├── POC-MCP.sln                   # Visual Studio solution file
└── README.md                      # This file

Features

  • ✅ Cross-platform: Builds for Windows, macOS, and Linux (multiple architectures)
  • ✅ Self-contained: No .NET runtime required on target machines
  • ✅ NuGet distribution: Packaged and distributed via NuGet
  • ✅ stdio transport: Uses JSON-RPC over stdin/stdout for communication
  • ✅ Tool discovery: AI assistants can automatically discover available tools
  • ✅ IDE integration: Works with VS Code and Visual Studio
  • 🤖 GitHub Copilot Integration: Pre-configured MCP setup in .vscode/mcp.json
  • 🍅 Pomodoro Timer Tools: Console and graphical interface Pomodoro sessions
  • 🎲 Example Tools: Random number generation for testing

Quick Start

Prerequisites

  • .NET 9.0 SDK
  • Compatible IDE (VS Code or Visual Studio with MCP support)

Development

  1. Clone and navigate to the project:

    git clone https://github.com/ballombert/POC-MCP.git
cd POC-MCP

  2. Build the solution:

    dotnet build

  3. Run the MCP server:

    dotnet run --project src/SampleMcpServer

  4. Test with GitHub Copilot (VS Code):

    • The MCP server is pre-configured in .vscode/mcp.json
    • Open GitHub Copilot Chat in VS Code
    • Try these commands:
      • "Give me 3 random numbers"
      • "Launch a Pomodoro timer with graphical interface"
      • "Start a 30-minute work session in French"
    • The MCP server should be invoked automatically!

Building and Distribution

Building and Distribution

  1. Build for all platforms:

    dotnet pack -c Release

    Or use the version script:

    ./scripts/version.sh pack

  2. Manage versions:

    # Show current version
./scripts/version.sh show

# Bump version
./scripts/version.sh bump patch

# Set specific version
./scripts/version.sh set 1.0.0-rc1

  3. Publish to NuGet:

    dotnet nuget push bin/Release/*.nupkg --api-key <your-key> --source https://api.nuget.org/v3/index.json

🤖 GitHub Copilot Integration

This project comes with GitHub Copilot integration pre-configured! The MCP server is automatically available when you open this project in VS Code.

Available Commands

Ask GitHub Copilot any of these commands:

🍅 Pomodoro Timer
  • "Launch a Pomodoro timer with graphical interface"
  • "Start a 25-minute work session with 5-minute breaks"
  • "Open the Pomodoro timer in French"
  • "Stop my current Pomodoro session"
  • "What's the status of my Pomodoro timer?"
🎲 Utilities
  • "Give me 3 random numbers"
  • "Generate random numbers between 10 and 50"

How it Works

  1. Auto-Detection: VS Code automatically loads .vscode/mcp.json
  2. Background Server: The MCP server runs when Copilot needs it
  3. Tool Invocation: Copilot automatically chooses the right tool for your request
  4. Real Results: Get actual Pomodoro timers and random numbers!

See for detailed setup instructions.

Example Tools

The sample server includes multiple tool classes demonstrating:

🎲 RandomNumberTools

  • Tool method decoration with [McpServerTool]
  • Parameter documentation with [Description] attributes
  • Default parameter values
  • Return value handling

🍅 PomodoroTools

  • Console-based Pomodoro sessions
  • Graphical interface launching
  • Session management and status checking
  • Multi-language support (6 languages)
  • Process launching and error handling
[McpServerTool]
[Description("Generates a random number between the specified minimum and maximum values.")]
public int GetRandomNumber(
    [Description("Minimum value (inclusive)")] int min = 0,
    [Description("Maximum value (exclusive)")] int max = 100)
{
    return Random.Shared.Next(min, max);
}

Architecture

  • Host: Microsoft.Extensions.Hosting for application lifecycle
  • Transport: stdio-based JSON-RPC communication
  • SDK: Official ModelContextProtocol NuGet package
  • Logging: Configured to stderr (stdout reserved for protocol)
  • Packaging: Self-contained, single-file executable

Documentation

  • - Comprehensive guides and references
  • - Setup and quick start guide
  • - Technical architecture details
  • - MCP tools and methods documentation
  • - Build, versioning, and distribution
  • - Important architectural decisions and rationale
  • - Code examples and usage patterns
  • - Server-specific details
  • - For AI development agents
  • Official MCP Documentation - External resources
  • Protocol Specification - MCP protocol details

Contributing

This is a proof of concept project. When contributing:

  1. Follow the patterns established in the existing codebase
  2. Update documentation when adding new features
  3. Test with actual AI assistants before submitting
  4. Consider cross-platform compatibility

License

For detailed implementation guidance and development patterns, see .