Remote MCP Functions Dotnet

🚀 Creating an MCP Server on Azure with GitHub Copilot

Microsoft Azure provides a wide range of features and tools that enable developers to easily deploy and manage applications. This guide details how to use Azure Functions to create an MCP (Microsoft Cloud Platform) server and interact with it via GitHub Copilot.

📋 Prerequisites

1. Register a Microsoft Account: If you don't have one, you need to sign up for a Microsoft account first.
2. Install Azure CLI: Make sure the latest Azure CLI tool is installed on your computer and properly configured.
3. Install Visual Studio Code: As the primary development tool, VS Code is an excellent choice and supports various plugins to enhance functionality.

📦 Creating an Azure Function App

Step 1: Log in to Your Azure Account

Run the following command in the terminal or command prompt to log in to your Microsoft account:

az login

Follow the prompts to enter your credentials and complete the login.

Step 2: Create a Resource Group

Create a new resource group for your function app by executing the following command:

az group create --name myResourceGroup --location "East US"

This will create a resource group named myResourceGroup in the "East US" region.

Step 3: Create a Function App

Next, use the following command to create your Azure function app:

az functionapp create \
    --name myFunctionApp \
    --resource-group myResourceGroup \
    --location "East US" \
    --runtime dotnet-isolated \
    --functions-version 4

This command creates a function app named myFunctionApp using the .NET Isolated runtime and specifies version 4.

⚙️ Configuring the MCP Server

Step 1: Clone the Example Code Repository

To get started, you need to clone an example code repository that contains MCP server functionality. You can find a suitable project on GitHub or create a new one.

git clone https://github.com/yourusername/mcp-function.git

Step 2: Install Dependencies

After entering the project directory, install the required NuGet packages and other dependencies. For .NET projects, you can use dotnet restore to restore all dependencies:

cd mcp-function
dotnet restore

🚀 Deploying to Azure

Step 1: Build the Function App Package

Build your function app package for deployment to Azure.

dotnet build --configuration Release

This will generate a release folder named bin/Release/net6.0/isolated/runtimes/win-x64/publish.

Step 2: Deploy to the Azure Function App

Use the following command to deploy the built package to your function app:

az functionapp deploy --name myFunctionApp --resource-group myResourceGroup --path .\bin\Release\net6.0\isolated\runtimes\win-x64\publish

⚙️ Configuring MCP Tools

Step 1: Define MCP Tool Triggers

In your function, use the [McpToolTrigger] attribute to define the tool's name and description. For example:

[Function(nameof(SaveSnippet))]
[BlobOutput("snippets/{snippet}")]
public async Task<IActionResult> SaveSnippet(
    [McpToolTrigger("save-snippet", "Saves a snippet to storage")] ToolInvocationContext context,
    [McpToolProperty("snippet-name", "name of the snippet", "Name of the snippet")] string name,
    [McpToolProperty("snippet-content", "content of the snippet", "Content of the snippet")] string content)
{
    await blobService.SaveAsync(name, content);
    return new OkResult();
}

Step 2: Configure MCP Tool Properties

In your function, use the [McpToolProperty] attribute to define the tool's input parameters. For example:

[Function(nameof(GetSnippet))]
[BlobInput("snippets/{snippet}")]
public async Task<IActionResult> GetSnippet(
    [McpToolTrigger("get-snippet", "Retrieves a snippet from storage")] ToolInvocationContext context,
    [McpToolProperty("snippet-name", "name of the snippet", "Name of the snippet")] string name)
{
    var content = await blobService.GetAsync(name);
    return new ContentResult { Content = content };
}

⚙️ Configuring GitHub Copilot

Step 1: Install Necessary VS Code Plugins

Ensure you have installed the following plugins:

• VS Live Share: For real - time collaboration.
• Azure Functions: For interacting with Azure functions.

Step 2: Connect to Your Azure Account

In VS Code, use the "Connect to Azure Account" command in Azure to log in to your Microsoft account.

Step 3: Configure Copilot

In VS Code, enable GitHub Copilot and follow the prompts to set it up. After completion, you can get real - time help while coding.

💻 Usage Examples

Creating a New Tool

1. Create a New Function:

dotnet new functionapp -n MyMCPFunction --output .

2. Add MCP Tool Functionality: Edit the MyMCPFunction/Functions/Program.cs file and add the functionality you want. For example:

public class SaveSnippet : FunctionBase
{
    private readonly IBlobService blobService;

    public SaveSnippet(IBlobService blobService)
    {
        this.blobService = blobService;
    }

    [Function("save-snippet")]
    public async Task<IActionResult> Run(
        [McpToolTrigger] ToolInvocationContext context,
        [McpToolProperty("snippet-name", "name")] string name,
        [McpToolProperty("snippet-content", "content")] string content)
    {
        await blobService.SaveAsync(name, content);
        return new OkResult();
    }
}

Using the Tool

1. Invoke the Tool: In VS Code, use GitHub Copilot to enter the following command to invoke your MCP tool:

/invoke save-snippet --snippet-name "mysnippet" --snippet-content "This is my snippet content."

2. Check the Results: Confirm whether the tool executed successfully and check the output in the console or logs.

🔍 Verification and Troubleshooting

Check the Deployment Status

Use the following command to check the deployment status of your function app:

az functionapp show --name myFunctionApp --resource-group myResourceGroup

View Logs

If you encounter issues, you can view the function's application logs for debugging:

az functionapp log tail --name myFunctionApp --resource-group myResourceGroup

Verify Tool Invocation

Ensure the MCP tool works properly by invoking it and checking the results.

🗑️ Cleaning Up Resources

After completing all operations, it is recommended to clean up unnecessary resources to avoid unnecessary costs. You can use the following command to delete the resource group:

az group delete --name myResourceGroup

By following these steps, you have successfully created and configured an MCP server and interacted with it via GitHub Copilot.