vk-cloud-mcp-server

# VK Cloud MCP Server The Model Context Protocol (MCP) server for VK Cloud is a solution that provides the ability to manage cloud infrastructure using LLM through a standardized interface. It allows AI assistants and other tools to interact with VK Cloud resources. IMPORTANT INFORMATION UNOFFICIAL PROJECT: This is an unofficial implementation of the Model Context Protocol (MCP) server for VK Cloud. This project is not affiliated with VK Cloud, is not supported or endorsed by them. It was developed independently as a weekend experimental project to explore the capabilities of MCP. Questions, suggestions, and contributions to the project are welcome. Please contact me via Telegram: @volinski. ## Capabilities This MCP server provides the following operations for VK Cloud: ### Virtual Machine (VM) Operations - View the list of all VMs in the project - Start a stopped VM - Stop a running VM - Restart a VM (soft or hard reboot) - Create a new VM with various configuration options - View the list of VM types (flavors) with filtering options - View the list of VM images with advanced filtering capabilities ### Network Operations - View the list of all networks in the project - View the list of all security groups in the project ### Storage Operations - View the list of all disks (volumes) with filtering options - Create bootable volumes from images ### Access Operations - View the list of all SSH keys in the project ## Requirements - Python 3.8 or higher - An active VK Cloud account with API access ## Installation #### Create virtual environment ```bash python -m venv vk-cloud-mcp-venv ``` #### Activate the virtual environment On Windows: ```bash vk-cloud-mcp-venv\Scripts\activate ``` On macOS/Linux: ```bash source vk-cloud-mcp-venv/bin/activate ``` #### Install dependencies ```bash pip install -r requirements.txt ``` 1. Create a configuration file (`config.yaml`) with your VK Cloud credentials. All data to fill in can be found in the "Project Settings" -> "API Access" section https://cloud.vk.com/docs/ru/tools-for-using-services/api/rest-api/enable-api ```yaml vkcloud: auth_url: "https://infra.mail.ru:35357/v3/" username: "your-username" password: "your-password" project_id: "your-project-id" user_domain_name: "users" project_domain_name: "Default" destructive_operations: false # Set to true to allow creating/starting/stopping VMs compute_endpoint: "https://infra.mail.ru:8774/v2.1" image_endpoint: "https://infra.mail.ru:9292/v2" block_storage_endpoint: "https://public.infra.mail.ru:8776/v3" ``` 2. Example configuration file for setting up the MCP server for Windsurf `mcp_config.json` https://docs.windsurf.com/windsurf/mcp#adding-a-new-server https://docs.windsurf.com/windsurf/mcp#mcp-config-json ```json { "mcpServers": { "vkcloud": { "serverUrl": "http://localhost:8000/sse" } } } ``` ## Running the Server Start the server with the command: ```bash python main.py ``` This will start the MCP server at http://localhost:8000/sse by default. ## Security Considerations - By default, destructive operations (creating, starting, stopping, restarting VMs) are disabled. Set `destructive_operations: true` in the configuration file. ## Available MCP Tools The server provides the following MCP tools: - `vkcloud_vm_list`: List of all VMs in the project - `vkcloud_network_list`: List of all networks - `vkcloud_security_group_list`: List of all security groups - `vkcloud_vm_start`: Start VM (requires enabling destructive operations) - `vkcloud_vm_stop`: Stop VM (requires enabling destructive operations) - `vkcloud_vm_restart`: Restart VM (requires enabling destructive operations) - `vkcloud_flavor_list`: List of VM types with filtering options - `vkcloud_image_list`: List of VM images with advanced filtering capabilities - `vkcloud_disk_list`: List of all disks (volumes) with filtering options - `vkcloud_keypair_list`: List of all SSH keys - `vkcloud_create_bootable_volume`: Create a bootable volume (requires enabling destructive operations) - `vkcloud_vm_create`: Create a new VM (requires enabling destructive operations) For detailed information on the parameters and usage of each tool, please refer to the documentation in the code.