MCP With Semantic Kernel

🚀 Integrating Model Context Protocol (MCP) Tools with Semantic Kernel: A Step-by-Step Guide

This repository demonstrates how to integrate Model Context Protocol (MCP) tools with Microsoft Semantic Kernel, enabling seamless interaction between AI models and external data sources or tools. By following this guide, you'll learn how to connect to an MCP server, convert MCP tools into Semantic Kernel functions, and leverage large language models (LLMs) for function calls—all within a reusable and extensible framework.

🚀 Quick Start

What is the Model Context Protocol (MCP)?

Model Context Protocol (MCP) is an open standard protocol designed to standardize how applications provide context for AI models. It acts as a universal connector, allowing LLLMs to interact with various data sources (e.g., APIs, databases, or services) in a consistent manner. Think of MCP as a bridge that enhances AI interoperability, flexibility, and context understanding.

In this project, we use MCP to expose tools that the Semantic Kernel can utilize, enabling AI-driven workflows and integration with real-world applications (such as automation, data retrieval, or system integration).

Why Combine Semantic Kernel with MCP?

Microsoft Semantic Kernel is a powerful SDK that simplifies the process of building AI agents and orchestrating complex workflows. By integrating MCP tools, you can:

• Extend the capabilities of the Semantic Kernel using an MCP server.
• Enable LLLMs to dynamically call functions based on user prompts.
• Promote interoperability between AI models and non-Semantic Kernel applications.
• Simplify development by using a standardized protocol for tool integration.

This repository provides a practical example of how to combine these technologies, along with sample code to help you get started quickly.

Prerequisites

Before diving into the code, ensure you have the following:

• .NET SDK (recommended version 8.0 or higher).
• A valid OpenAI API key (or other LLM provider compatible with the Semantic Kernel).
• The ModelContextProtocol NuGet package.
• A basic familiarity with C# and Semantic Kernel concepts.
• (Optional) An MCP server for testing, such as the sample "Everything" MCP server for demonstration purposes.

Step-by-Step Guide

This section guides you through the process of integrating MCP tools with the Semantic Kernel, as implemented in this repository.

Step 1: Set Up the Project

1. Clone this repository:

git clone https://github.com/LiteObject/mcp-with-semantic-kernel.git
cd mcp-with-semantic-kernel

2. Restore dependencies:

dotnet restore

3. Configure your OpenAI API key (or other LLM credentials) using environment variables or user secrets:

dotnet user-secrets set "OpenAI:ApiKey" "your-api-key"
dotnet user-secrets set "OpenAI:ChatModelId" "gpt-4o"

Step 2: Connect to the MCP Server

The project includes code to connect to an MCP server using the ModelContextProtocol package. The MCP client retrieves available tools from the server, which can then be used by the Semantic Kernel.

Example code (see mcp-with-semantic-kernel/src/Program.cs):

using ModelContextProtocol;

var mcpConfig = new McpServerConfig
{
    Id = "everything",
    Name = "Everything",
    TransportType = TransportTypes.Sse,
    Location = "http://localhost:8931"
};

var mcpClient = await McpClientFactory.CreateAsync(mcpConfig);
var tools = await mcpClient.ListToolsAsync();

This code snippet connects to an MCP server (e.g., the "Everything" demo server) and retrieves its available tools.