Remote MCP Servers using .NET SDK - Integrating with custom data and APIs

A comprehensive example demonstrating how to build Model Context Protocol (MCP) servers using the .NET SDK, showcasing integration with custom data sources and APIs. This project implements a weather forecast service as an example of how to expose your own APIs through MCP tools.

Table of Contents

• Overview
• What is MCP?
• Features
• Architecture
• Prerequisites
• Project Structure
• Development
• Quick Start
• Available MCP Tools
• Usage
• Contributing
• License

Overview

This project demonstrates how to create remote MCP servers that can be consumed by AI applications or clients (like Claude Desktop, VS Code Copilot, etc.) to extend their capabilities with custom tools and data sources. The example implementation includes a weather forecast service that showcases the integration pattern.

MCP servers really shine when they’re connected to existing APIs or services, allowing clients to query real, live data. There’s an expanding ecosystem of MCP servers that can already be used by clients, including tools we rely on daily like Git, GitHub, local filesystem, etc.

With that in mind, let’s level up our MCP server: plug it into an API, pass in some query params, and start returning responses that are actually driven by real data.

What is MCP?

Model Context Protocol (MCP) is an open protocol that standardizes how applications provide context to Large Language Models (LLMs). It enables AI assistants to:

• Access external data sources
• Execute custom tools and functions
• Integrate with your own APIs and services
• Maintain context across multiple interactions

Features

• HTTP Transport Support - Expose MCP servers via HTTP endpoints for remote access
• Stdio Transport Support - Support for standard input/output communication
• Custom Tool Integration - Easy integration of your own APIs and services
• Health Monitoring - Built-in health check endpoints
• Stateless Operation - Designed for scalable, stateless deployments
• .NET 10.0 - Built on the latest .NET framework
• Structured Logging - Comprehensive logging for debugging and monitoring

Architecture

The project follows a clean architecture pattern:

┌─────────────────┐
│   MCP Client    │ (Claude Desktop, Cursor IDE, Continue.dev, etc.)
└────────┬────────┘
         │ HTTP/SSE
         ▼
┌─────────────────┐
│  ASP.NET Core   │
│   Web Host      │
└────────┬────────┘
         │
         ▼
┌─────────────────┐
│  MCP Protocol   │
│     Layer       │
└────────┬────────┘
         │
         ▼
┌─────────────────┐
│   MCP Tools     │ (Weather, Ping, etc.)
└────────┬────────┘
         │
         ▼
┌─────────────────┐
│  Your APIs &    │
│  Data Sources   │
└─────────────────┘

Prerequisites

• .NET 10.0 SDK or later
• Visual Studio 2022 / VS Code / any compatible editor
• An MCP-compatible developer IDE:
  • Claude Desktop
  • VS Code with MCP extension
  • Kiro or other MCP clients

Project Structure

remote-MCP-servers-using-dotnet-sdk-integrating-with-custom-data-or-APIs/
├── src/
│   └── McpServer/
│       ├── McpServer/
│       │   ├── Program.cs              # Application entry point
│       │   ├── appsettings.json        # Configuration settings
│       │   ├── McpServerTools.cs       # MCP tool implementations
│       │   └── Services/               # Business logic services
│       │       └── WeatherService.cs   # Weather API integration
│       └── McpServer.sln
├── docs/                               # Additional documentation
├── README.md   

... [View full README on GitHub](https://github.com/azurecorner/remote-MCP-servers-using-dotnet-sdk-integrating-with-custom-data-or-APIs#readme)