Novita AI

Novita AI is an MCP server that connects MCP-compatible clients such as Claude Desktop and Cursor directly to the Novita AI platform. Built for cloud engineers, machine learning practitioners, and developers, this server provides programmatic management over Novita AI infrastructure and GPU computing resources. Instead of switching to an external web dashboard, users can issue natural language instructions to inspect and provision hardware for AI training and inference. In its current beta version, the server focuses on managing Novita GPU instance products. It enables clients to list available clusters and regions, browse GPU hardware offerings, and inspect or deploy instances. Developers can start, stop, restart, or delete running machines directly from conversation prompts. Additionally, the integration exposes controls for configuring reusable environment templates, managing container registry authentication credentials, and handling persistent network storage volumes. This centralizes remote compute orchestration into your primary development environment.

Category: AI & LLM Tooling

Tags: Generative AI, image generation, novita, text-to-image

Visit Novita AI

How to install and configure Novita AI

1. Obtain your API key from the Novita AI Key Management dashboard. 2. Install the package globally via npm using npm install -g @novitalabs/novita-mcp-server or run it dynamically with npx. 3. Open your Claude Desktop or Cursor configuration file (claude_desktop_config.json or your Cursor MCP settings). 4. Add @novitalabs/novita-mcp-server under the mcpServers object: json { "mcpServers": { "@novitalabs/novita-mcp-server": { "command": "npx", "args": ["-y", "@novitalabs/novita-mcp-server"], "env": { "NOVITA_API_KEY": "your_api_key_here" } } } } 5. Replace your_api_key_here with your actual Novita API key and restart your client.

What you can do with Novita AI

• Provisioning GPU instances: Launch new Novita GPU compute instances specifying product types, disk capacity, and CUDA container images via client prompts. - Inspecting cluster resources: Query all available Novita clusters, regions, and hardware products to evaluate available capacity for ML workloads. - Managing instance lifecycles: Start, stop, restart, or delete active Novita GPU instances directly inside your development workflow to optimize compute costs. - Configuring network storage: Create, update, list, and delete persistent network storage volumes attached to your training containers without leaving your editor. - Handling container registries: Add, inspect, and remove container registry authentication credentials required to pull custom private images onto GPU clusters.

Key facts

• https://github.com/novitalabs/novita-mcp-server
• AI & LLM Tooling, Design, Media & Creative
• Generative AI, image generation, novita, text-to-image

Part of MCP Servers

How do I install Novita AI MCP server?

You can install the server globally using npm install -g @novitalabs/novita-mcp-server, or install it through Smithery. Alternatively, run it directly via npx inside your MCP client configuration by passing -y and @novitalabs/novita-mcp-server with your API key.

What can Novita AI MCP server do?

The server enables management of Novita AI GPU instances. Supported actions include listing clusters and hardware products, creating and controlling instances, managing container templates, configuring registry authentications, and manipulating network storage volumes from an MCP client.

Which MCP clients work with Novita AI?

Novita AI works with any compatible Model Context Protocol client. The official documentation specifically highlights support and testing for Claude Desktop and Cursor.

What credentials are required to use Novita AI MCP server?

You need an active Novita AI account and an API key from the Novita AI Key Management portal. You must configure this key under the NOVITA_API_KEY environment variable in your client configuration file.