MCP Server

Thunder Compute provides an MCP (Model Context Protocol) server that lets AI coding agents manage GPU instances on your behalf. Create, monitor, modify, and tear down instances without leaving your agent workflow.

Prerequisites

A Thunder Compute account
An AI agent that supports remote MCP servers (Claude Code, Cursor, etc.)

No local installation or API tokens required — authentication is handled via OAuth in your browser.

Setup

Claude Code
Cursor
Windsurf
Direct HTTP

Run this in Claude Code:

/mcp add --transport http https://api.thundercompute.com:8443/mcp

Then run /mcp to authenticate. A browser window will open for you to log in and authorize access.

Manual setup

Alternatively, add to ~/.claude.json (global) or .claude.json in your project root:

{
  "mcpServers": {
    "thunder-compute": {
      "url": "https://api.thundercompute.com:8443/mcp"
    }
  }
}

Add to .cursor/mcp.json in your project root:

{
  "mcpServers": {
    "thunder-compute": {
      "url": "https://api.thundercompute.com:8443/mcp"
    }
  }
}

Add to your MCP configuration:

{
  "mcpServers": {
    "thunder-compute": {
      "url": "https://api.thundercompute.com:8443/mcp"
    }
  }
}

For custom integrations, the MCP server uses Streamable HTTP transport at a single endpoint. Authentication is via OAuth 2.0 with standard MCP discovery.Endpoint: https://api.thundercompute.com:8443/mcp

curl -X POST https://api.thundercompute.com:8443/mcp \
  -H "Authorization: Bearer <access-token>" \
  -H "Content-Type: application/json" \
  -d '{
    "jsonrpc": "2.0",
    "method": "initialize",
    "params": {
      "protocolVersion": "2025-03-26",
      "capabilities": {},
      "clientInfo": { "name": "my-agent", "version": "1.0.0" }
    },
    "id": 1
  }'

Authentication

No API tokens or environment variables needed. When you first connect, a browser window opens for you to log in with your Thunder Compute account and authorize access. Tokens are refreshed automatically — you only authenticate once per session.

If your agent doesn’t support OAuth-based MCP authentication, you can use the local stdio transport with the tnr-mcp npm package instead. See Local (stdio) below.

Available Tools

Tool	Description
`list_instances`	List all GPU instances with status, IP, and configuration
`create_instance`	Create a new GPU instance (specify GPU type, template, mode, etc.)
`delete_instance`	Delete an instance (irreversible)
`modify_instance`	Change instance config (GPU type, vCPUs, disk, mode)
`get_specs`	Get available GPU specs (VRAM, vCPU options, storage ranges)
`get_pricing`	Get current per-hour GPU pricing
`list_templates`	List available OS templates (Ubuntu, PyTorch, etc.)
`list_snapshots`	List all instance snapshots
`create_snapshot`	Create a snapshot of an instance
`delete_snapshot`	Delete a snapshot (irreversible)
`list_ssh_keys`	List SSH keys in your organization
`add_ssh_key`	Add an SSH public key
`delete_ssh_key`	Delete an SSH key
`list_ports`	List all instances with their forwarded ports
`forward_port`	Forward HTTP ports on an instance
`remove_port`	Remove forwarded ports from an instance
`get_ssh_command`	Get the SSH command to connect to an instance
`get_scp_command`	Get the SCP command to copy files to/from an instance

Prompts

The MCP server includes built-in prompts that guide your agent through common multi-step workflows:

Prompt	Description
`create-dev-instance`	Set up a GPU development instance with sensible defaults
`deploy-model`	Deploy an ML model (supports Ollama, vLLM, and Transformers)
`check-costs`	Review running instances and estimated hourly costs
`snapshot-and-teardown`	Save instance state and clean up

Example Usage

Once configured, you can ask your AI agent things like:

“Spin up an A100 instance with PyTorch”
“What GPU types are available and how much do they cost?”
“List my running instances”
“Delete instance inst-abc123”
“Forward port 8080 on my instance”
“Create a snapshot of my instance before I make changes”
“Deploy Llama 3 on a GPU”

Local (stdio) Transport

If your agent doesn’t support remote MCP with OAuth, you can run the MCP server locally instead. This uses the tnr-mcp npm package, which runs on your machine and connects to the Thunder Compute API using an API token. Requirements: Node.js (npx must be available)

1. Get an API token

Generate a token from the Thunder Compute console, or run tnr login via the CLI.

2. Add the MCP server

Claude Code
Cursor
Windsurf

// ~/.claude.json or .claude.json in your project root
{
  "mcpServers": {
    "thunder-compute": {
      "command": "npx",
      "args": ["-y", "tnr-mcp"],
      "env": {
        "THUNDER_API_TOKEN": "your-token-here"
      }
    }
  }
}

// .cursor/mcp.json in your project root
{
  "mcpServers": {
    "thunder-compute": {
      "command": "npx",
      "args": ["-y", "tnr-mcp"],
      "env": {
        "THUNDER_API_TOKEN": "your-token-here"
      }
    }
  }
}

{
  "mcpServers": {
    "thunder-compute": {
      "command": "npx",
      "args": ["-y", "tnr-mcp"],
      "env": {
        "THUNDER_API_TOKEN": "your-token-here"
      }
    }
  }
}

Restart your agent after adding the config. If you’ve already run tnr login, you can omit THUNDER_API_TOKEN — the server picks up credentials from ~/.thunder/cli_config.json automatically.

Troubleshooting

Authentication fails or browser doesn’t open: Run /mcp in Claude Code to manually trigger authentication. Make sure you’re logged in to your Thunder Compute account in the browser. “Protected resource does not match” error: The URL in your MCP config must match the server’s configured resource URL exactly. Ensure you’re using https://api.thundercompute.com:8443/mcp. “token has invalid issuer” error: This is a server-side configuration issue. The MCP authentication client must be configured with the correct Stytch Connected Apps domain. Tools not appearing: Restart your AI agent after changing MCP configuration. Most agents only read MCP config on startup. Stdio server not starting: If using the local transport, verify npx is available by running which npx. If not, install Node.js from https://nodejs.org.

Getting Started

Operations

Guides

Reference

API Reference

Prerequisites

Setup

Authentication

Available Tools

Prompts

Example Usage

Local (stdio) Transport

1. Get an API token

2. Add the MCP server

Troubleshooting

Getting Started

Operations

Guides

Reference

API Reference

​Prerequisites

​Setup

​Authentication

​Available Tools

​Prompts

​Example Usage

​Local (stdio) Transport

​1. Get an API token

​2. Add the MCP server

​Troubleshooting

Prerequisites

Setup

Authentication

Available Tools

Prompts

Example Usage

Local (stdio) Transport

1. Get an API token

2. Add the MCP server

Troubleshooting