Azure Functions Notes MCP Server

An Azure Functions-based MCP (Model Context Protocol) server implementation for managing notes and related functionality. This project demonstrates how to build a MCP server using Azure Functions, following best practices in a cloud native way.

Features

• MCP Protocol Support: Implements the Model Context Protocol for seamless integration with GitHub Copilot Chat and other MCP-aware clients
• Azure Functions: Built on .NET 8.0 using the latest Azure Functions runtime
• Infrastructure as Code: Complete infrastructure definition using Bicep templates
• Azure Integration: Integration with Azure services including Application Insights for monitoring
• Role-Based Access Control (RBAC): Access control for various Azure resources

MCP Tools

The server provides the following MCP tools for note management:

Tool Properties

• save_note:

  • title (required): The title or identifier for the note
  • category: The category of the note (e.g., meeting, task, idea, code-snippet, reminder)
  • tags: Comma-separated tags for organizing the note
  • content (required): The main content of the note

• get_note:

  • title (required): The title of the note to retrieve

• search_notes:

  • query: Search query for tags or category

• delete_note:

  • title (required): The title of the note to delete

Prerequisites

For local development:

• Azure Developer CLI (azd)
• .NET 8.0 SDK
• Azure Functions Core Tools
• An Azure subscription
• Visual Studio Code (recommended) with:
  • C# Dev Kit extension
  • Azure Functions extension
  • Bicep extension

Getting Started

Local Development

1. Clone the repository:

git clone <repository-url>
cd azure-functions-mcp-server

1. Install dependencies:

dotnet restore

1. Configure local settings:

• Copy local.settings.sample.json to local.settings.json

1. Run the project locally:

cd NotesMcp
func start

Using with VS Code and GitHub Copilot

To use the MCP tools in VS Code, you'll need to configure the MCP endpoint in your workspace. Create a .vscode/mcp.json file with the configuration based on your environment:

Local Development

When running the Function App locally:

{
    "servers": {
        "notes-mcp": {
            "type": "http",
            "url": "http://localhost:7071/runtime/webhooks/mcp"
        }
    }
}

Deployment

The project uses Azure Developer CLI (azd)

Deploy to Azure

1. Initialize the environment:

azd init

1. Login to Azure:

azd auth login

1. Provision and deploy:

azd up

This command will:

• Create required Azure resources
• Deploy the application code
• Set up configurations
• Output the Function App URL

To remove all resources:

azd down --purge

Usage after deployment

When using the deployed Function App in Azure the mcp.json from the client looks like this:

{
    "inputs": [
        {
            "type": "promptString",
            "id": "function-key",
            "description": "Azure Function App Key",
            "password": true
        }
    ],
    "servers": {
        "notes-mcp": {
            "type": "http",
            "url": "https://<replace with correct value>/runtime/webhooks/mcp", //See Function App URL
            "headers": {
              "x-functions-key": "${input:function-key}"
            }
        }
    }
}

Examples

1. Saving a note
   • Select the text you want to save in VS Code
   • Open GitHub Copilot Chat
   • Switch to Agent mode (click the robot icon)
   • Type a command like:

     Save this selected text as a note titled "git-commands" with category "code-snippet" and tags "git,commands,reference"
     

   This will invoke the save_note tool with your selected text as content.

1. Retrieving a note
   • In GitHub Copilot Chat (Agent mode), type:

     Get the note with title "git-commands"
     

   This will use the get_note tool to fetch and display the note.

1. Listing all notes

   • In GitHub Copilot Chat (Agent mode), type:

     List all my saved notes
     

   This will use the list_notes tool to show all your saved notes.

2. Searching notes

   • In GitHub Copilot Chat (Agent mode), type:

     Search for notes with tag "git"
     

   This will use the search_notes tool to find relevant notes.

1. Delete a note
   • In GitHub Copilot Chat (Agent mode), type:

     Delete my note called git-commands
     

   This will use the delete_note tool to delete the note.

1. Listing all notes (verify)
   • In GitHub Copilot Chat (Agent mode), type:

     List all my saved notes
     

   This will use the list_notes tool to show all your saved notes. Verify that after delete nothing is saved anymore.

Infrastructure

The project uses Bicep for Infrastructure as Code with the following components:

• main.bicep: Main infrastructure template
• app/api.bicep: API-related resources
• app/monitoring.bicep: Monitoring setup with Application Insights
• rbac/: Role-based access control definitions
  • appinsights-access.bicep
  • storage-access.bicep

Guidelines & best practices

Region availability

Azure Function App availability (including Flex Consumption) varies by region, verify availability here.

Choose the same region for related resources (Function App, Storage account) to reduce latency and egress costs. westeurope is a good default if you're in or near Europe.

Cost considerations

Estimate costs using the Azure Pricing Calculator. Primary cost drivers:

• Azure Functions – Consumption or Flex tiers (invocations, memory/time)
• Storage Account – capacity and transactions

Tip: enable cost alerts in your subscription and test workloads in a sandbox subscription before production.

Security

This project uses a User-Assigned Managed Identity for authentication between the Azure resources.

The Function App is assigned the following RBAC roles:

• Storage Blob Data Owner
• Application Insights Monitoring Metrics Publisher

When using in Production environment, the following is strongly recommended:

• Restrict inbound traffic with Private endpoints + VNet integration
• Use service endpoints + firewall rules where applicable

Contributing

1. Fork the repository
2. Create a feature branch
3. Commit your changes
4. Push to the branch
5. Create a Pull Request

License

MIT License