azure-machine-learning-mcp

# Azure Machine Learning MCP Server A comprehensive Model Context Protocol (MCP) server that provides access to Azure Machine Learning services through the Azure SDK for Go. This server enables you to manage Azure ML workspaces, compute resources, quotas, and more through standardized MCP tools. ## 🚀 Quick Start **Want to use this in VS Code right away?** Check out our [VS Code Setup Guide](VSCODE_SETUP.md) for step-by-step instructions to get started in minutes! ## Features This MCP server provides the following Azure ML management capabilities: ### Workspace Management - **list_workspaces_by_subscription**: List all Azure ML workspaces in a subscription - **get_workspace**: Get detailed information about a specific workspace - **create_workspace**: Create a new Azure ML workspace ### Compute Resource Management - **list_compute**: List all compute resources in a workspace - **get_compute**: Get detailed information about a specific compute resource - **start_compute**: Start a compute resource - **stop_compute**: Stop a compute resource ### Resource Monitoring - **list_quotas**: List resource quotas for a specific Azure region - **list_usage**: List current resource usage for a specific Azure region - **list_vm_sizes**: List available virtual machine sizes for compute ### Network & Security - **list_private_endpoints**: List private endpoint connections for a workspace - **list_workspace_connections**: List workspace connections - **list_workspace_features**: List available features for a workspace ## Prerequisites 1. **Azure Subscription**: You need an active Azure subscription 2. **Azure Authentication**: The server will automatically prompt for authentication when you first use it. You can use any of these methods: - **Interactive Browser Login**: The server will open your browser for authentication (default) - **Device Code Flow**: Used automatically in headless environments - **Azure CLI**: If you're already logged in with `az login` - **Service Principal**: Set environment variables `AZURE_CLIENT_ID`, `AZURE_CLIENT_SECRET`, `AZURE_TENANT_ID` - **Managed Identity**: When running on Azure resources 3. **Go 1.18+**: Required to build and run the server ## Installation ### VS Code Setup (Recommended) For the easiest setup experience in Visual Studio Code, follow our dedicated [VS Code Setup Guide](VSCODE_SETUP.md). This guide covers: - Docker-based setup with pre-built images - Local binary setup from source - Authentication configuration - Troubleshooting common issues - Usage examples with GitHub Copilot ### Manual Installation If you prefer to set up manually or use with other MCP clients: 1. Clone this repository: ```bash git clone <repository-url> cd aml-cmp ``` 2. Install dependencies: ```bash go mod tidy ``` 3. Build the server: ```bash go build ./cmd/mcp-server ``` ## Usage Examples ### 1. List Workspaces ``` Use the list_workspaces_by_subscription tool with: - subscription_id: "your-azure-subscription-id" ``` ### 2. Create a Workspace ``` Use the create_workspace tool with: - subscription_id: "your-azure-subscription-id" - resource_group_name: "my-resource-group" - workspace_name: "my-ml-workspace" - location: "eastus" - description: "My ML workspace for experiments" - friendly_name: "ML Workspace" ``` ### 3. Manage Compute Resources ``` # List compute resources Use list_compute with workspace details # Start a compute instance Use start_compute with: - subscription_id, resource_group_name, workspace_name, compute_name # Stop a compute instance Use stop_compute with: - subscription_id, resource_group_name, workspace_name, compute_name ``` ### 4. Monitor Resources ``` # Check quotas Use list_quotas with: - subscription_id: "your-subscription-id" - location: "eastus" # Check current usage Use list_usage with same parameters ``` ### 5. List Available VM Sizes ``` Use list_vm_sizes with: - subscription_id: "your-subscription-id" - location: "eastus" ``` ## Tool Reference ### Workspace Tools #### `list_workspaces_by_subscription` Lists all Azure ML workspaces in a subscription. **Parameters:** - `subscription_id` (required): Azure subscription ID **Returns:** List of workspaces with names, locations, and resource groups. #### `get_workspace` Gets detailed information about a specific workspace. **Parameters:** - `subscription_id` (required): Azure subscription ID - `resource_group_name` (required): Resource group name - `workspace_name` (required): Workspace name **Returns:** Detailed workspace information including properties, URLs, and configuration. #### `create_workspace` Creates a new Azure ML workspace. **Parameters:** - `subscription_id` (required): Azure subscription ID - `resource_group_name` (required): Resource group name - `workspace_name` (required): Workspace name - `location` (required): Azure region (e.g., "eastus", "westus2") - `description` (optional): Workspace description - `friendly_name` (optional): Friendly display name **Returns:** Confirmation of workspace creation with workspace ID. ### Compute Tools #### `list_compute` Lists all compute resources in a workspace. **Parameters:** - `subscription_id` (required): Azure subscription ID - `resource_group_name` (required): Resource group name - `workspace_name` (required): Workspace name **Returns:** List of compute resources with types, states, and locations. #### `get_compute` Gets detailed information about a specific compute resource. **Parameters:** - `subscription_id` (required): Azure subscription ID - `resource_group_name` (required): Resource group name - `workspace_name` (required): Workspace name - `compute_name` (required): Compute resource name **Returns:** Detailed compute information including state, creation time, and configuration. #### `start_compute` / `stop_compute` Start or stop a compute resource. **Parameters:** - `subscription_id` (required): Azure subscription ID - `resource_group_name` (required): Resource group name - `workspace_name` (required): Workspace name - `compute_name` (required): Compute resource name **Returns:** Confirmation of operation completion. ### Monitoring Tools #### `list_quotas` Lists resource quotas for a specific region. **Parameters:** - `subscription_id` (required): Azure subscription ID - `location` (required): Azure region **Returns:** List of quotas with limits, units, and resource types. #### `list_usage` Lists current resource usage for a specific region. **Parameters:** - `subscription_id` (required): Azure subscription ID - `location` (required): Azure region **Returns:** Current usage with limits and available capacity. #### `list_vm_sizes` Lists available virtual machine sizes for compute. **Parameters:** - `subscription_id` (required): Azure subscription ID - `location` (required): Azure region **Returns:** Available VM sizes with vCPU and memory specifications. ### Network & Security Tools #### `list_private_endpoints` Lists private endpoint connections for a workspace. **Parameters:** - `subscription_id` (required): Azure subscription ID - `resource_group_name` (required): Resource group name - `workspace_name` (required): Workspace name **Returns:** Private endpoint connections with status and configuration. #### `list_workspace_connections` Lists workspace connections (data stores, linked services). **Parameters:** - `subscription_id` (required): Azure subscription ID - `resource_group_name` (required): Resource group name - `workspace_name` (required): Workspace name **Returns:** Workspace connections with types and authentication methods. #### `list_workspace_features` Lists available features for a workspace. **Parameters:** - `subscription_id` (required): Azure subscription ID - `resource_group_name` (required): Resource group name - `workspace_name` (required): Workspace name **Returns:** Available workspace features and capabilities. ## Error Handling The server provides detailed error messages for common scenarios: - **Authentication Errors**: Issues with Azure credentials or permissions - **Resource Not Found**: When specified resources don't exist - **Permission Errors**: When the authenticated user lacks required permissions - **API Errors**: Azure API-specific errors with detailed messages ## Development ### Project Structure ``` . ├── cmd/mcp-server/ # Main application entry point │ └── main.go ├── internal/ # Internal packages (not importable externally) │ ├── azure/ # Azure client management │ │ ├── clients.go # Azure service clients │ │ └── tests/ # Tests for Azure client functionality │ │ └── clients_test.go │ ├── helpers/ # Utility functions │ │ ├── azure.go # Azure-specific helpers │ │ └── tests/ # Tests for helper functions │ │ └── azure_test.go │ ├── server/ # MCP server implementation │ │ ├── server.go # Server setup and configuration │ │ └── tests/ # Server tests │ │ └── server_test.go │ └── tools/ # MCP tool implementations │ ├── workspace.go # Workspace management tools │ ├── compute.go # Compute resource tools │ ├── monitoring.go # Monitoring tools (quotas, usage, VM sizes) │ ├── network.go # Network and security tools │ └── tests/ # Tests for tool implementations │ ├── workspace_test.go │ ├── compute_test.go │ ├── monitoring_test.go │ └── network_test.go ├── .github/workflows/ # GitHub Actions CI/CD ├── Makefile # Build and development commands ├── go.mod # Go module definition ├── go.sum # Go module checksums └── README.md # This file ``` ### Building ```bash # Build the server make build # Or manually go build ./cmd/mcp-server ``` ### Testing ```bash # Run all tests make test # Run tests with verbose output make test-verbose # Run tests with coverage make test-coverage # Run specific test suites make test-helpers # Test utility functions make test-azure # Test Azure client creation make test-tools # Test MCP tool implementations make test-server # Test server setup ``` ### Development Commands ```bash # Install dependencies make deps # Format code make format # Run linter make lint # Run all checks (format, lint, test) make check # Clean build artifacts make clean # Build for multiple platforms make build-all # Run the server locally make run ``` ### Running Locally ```bash # Using make make run # Or directly go run ./cmd/mcp-server ``` ### Adding New Tools To add new Azure ML functionality: 1. Create a new tool definition using `mcp.NewTool()` 2. Implement the handler function with Azure SDK calls 3. Add the tool to the server using `s.AddTool()` 4. Update this README with documentation ## Contributing 1. Fork the repository 2. Create a feature branch 3. Make your changes 4. Add tests if applicable 5. Update documentation 6. Submit a pull request ## License This project is licensed under the MIT License. See LICENSE file for details. ## Support For issues and questions: - Check the Azure ML documentation: https://docs.microsoft.com/azure/machine-learning/ - Review Azure SDK for Go documentation: https://pkg.go.dev/github.com/Azure/azure-sdk-for-go - Open an issue in this repository ## Azure ML SDK Coverage This MCP server covers the following Azure ML SDK clients: - ✅ WorkspacesClient - ✅ ComputeClient - ✅ QuotasClient - ✅ UsagesClient - ✅ VirtualMachineSizesClient - ✅ PrivateEndpointConnectionsClient - ✅ WorkspaceConnectionsClient - ✅ WorkspaceFeaturesClient - ⏳ WorkspaceSKUsClient (planned) - ⏳ PrivateLinkResourcesClient (planned) - ⏳ OperationsClient (planned) Future versions will expand coverage to include additional Azure ML services and capabilities.