Hyperbolic GPU MCP server

The Hyperbolic GPU MCP Server enables you to interact with Hyperbolic's GPU cloud directly from your AI assistant. This server allows you to view available GPUs, rent instances, connect via SSH, and run GPU-powered workloads through a simple interface.

Prerequisites

Before getting started, you'll need:

• Node.js 16 or higher
• npm or yarn
• A Hyperbolic API token
• (Optional) SSH private key for connecting to GPU instances

Setting Up Your Hyperbolic Account

Getting an Account and API Token

1. Register for a Hyperbolic account:

   • Visit https://app.hyperbolic.xyz/
   • Create an account or log in
   • Verify your email address

2. Deposit funds into your account:

   • Log in to your Hyperbolic application
   • Navigate to the "Billing" tab
   • Select how much you want to deposit (starting with $25 is recommended)
   • Click Pay Now and follow the payment instructions
   • Note that you need sufficient funds to rent GPU instances

3. Generate an API token:

   • In your Hyperbolic dashboard, go to "Settings"
   • Navigate to the API Key section
   • Copy the generated token and keep it secure

4. Add your SSH public key:

   • Generate an SSH key pair if you don't have one already
   • In your Hyperbolic application, go to "Settings"
   • Scroll to the SSH Public Key section
   • Paste your public key (usually from ~/.ssh/id_rsa.pub)
   • Save the key

Installation

1. Clone the repository:

   git clone <your-repo-url>
   cd hyperbolic-mcp
   

2. Install dependencies:

   npm install
   

3. Build the TypeScript files:

   npm run build
   

Using the Server

Running Locally

To start the server:

npm start

Connecting with Claude for Desktop

1. Add the server to your Claude for Desktop configuration:

{
  "mcpServers": {
    "hyperbolic-gpu": {
      "command": "node",
      "args": ["/path/to/hyperbolic-mcp-server/build/index.js"],
      "env": {
        "HYPERBOLIC_API_TOKEN": "your-hyperbolic-api-token",
        "SSH_PRIVATE_KEY_PATH": "/path/to/your/privatekey" 
      }
    }
  }
}

2. Restart Claude for Desktop.

3. Start a new conversation and interact with the server.

Note: You can provide environment variables either through the Claude Desktop config or by creating a .env file in the project root.

Available Tools

GPU Management Tools

list-available-gpus

Lists all available GPUs on the Hyperbolic network.

Example query: "Show me all available GPUs on Hyperbolic."

rent-gpu-instance

Rents a GPU instance from a specific cluster.

Parameters:

• cluster_name: Cluster name (e.g., "extrasmall-chamomile-duck")
• node_name: Node name (e.g., "prd-acl-msi-02.fen.intra")
• gpu_count: Number of GPUs to rent

Example query: "I want to rent 4 GPUs from the extrasmall-chamomile-duck cluster."

terminate-gpu-instance

Terminates a GPU instance you have rented.

Parameters:

• instance_id: ID of the instance to terminate

Example query: "Terminate my GPU instance with ID abc123."

list-user-instances

Lists all your active GPU instances.

Example query: "Show me all my active GPU instances."

get-cluster-details

Gets detailed information about a specific cluster.

Parameters:

• cluster_name: Name of the cluster

Example query: "Tell me more about the cluster called extrasmall-chamomile-duck."

SSH Tools

ssh-connect

Establishes an SSH connection to a remote server.

Parameters:

• host: Hostname or IP address
• username: SSH username
• password: (Optional) SSH password
• private_key_path: (Optional) Path to private key file
• port: (Optional) SSH port (default: 22)

Example query: "Connect to my GPU instance at 192.168.1.100 as user admin."

remote-shell

Executes a command on the connected server.

Parameters:

• command: Command to execute

Example query: "Run 'nvidia-smi' on the connected server."

ssh-status

Checks the current SSH connection status.

Example query: "What's the status of my SSH connection?"

ssh-disconnect

Closes the active SSH connection.

Example query: "Disconnect from the SSH server."

Troubleshooting

If you encounter issues:

1. Verify your API token is correct and not expired
2. Ensure you have sufficient credits on your Hyperbolic account
3. Check the server logs for error messages
4. Verify your network connection to the Hyperbolic API
5. If using SSH, check that your private key path is correct and the key has appropriate permissions

How to add this MCP server to Cursor

There are two ways to add an MCP server to Cursor. The most common way is to add the server globally in the ~/.cursor/mcp.json file so that it is available in all of your projects.

If you only need the server in a single project, you can add it to the project instead by creating or adding it to the .cursor/mcp.json file.

Adding an MCP server to Cursor globally

To add a global MCP server go to Cursor Settings > MCP and click "Add new global MCP server".

When you click that button the ~/.cursor/mcp.json file will be opened and you can add your server like this:

{
    "mcpServers": {
        "cursor-rules-mcp": {
            "command": "npx",
            "args": [
                "-y",
                "cursor-rules-mcp"
            ]
        }
    }
}

Adding an MCP server to a project

To add an MCP server to a project you can create a new .cursor/mcp.json file or add it to the existing one. This will look exactly the same as the global MCP server example above.

How to use the MCP server

Once the server is installed, you might need to head back to Settings > MCP and click the refresh button.

The Cursor agent will then be able to see the available tools the added MCP server has available and will call them when it needs to.

You can also explictly ask the agent to use the tool by mentioning the tool name and describing what the function does.