GPU MCP Server

Indexed by io.github.pmady

About

NVIDIA GPU metrics as MCP tools — utilization, memory, temperature, power. Supports MIG.

Use this server

Add it to your MCP client. The catalogue lists this config verbatim from its source — it does not run, download, or vouch for the server. Review the command before you run it.

Transportstdio (runs a local command)

Claude Code

claude mcp add gpu-mcp-server -- docker ghcr.io/pmady/gpu-mcp-server:v0.1.0

Cursor

Add to Cursor

If the button doesn't open Cursor, use the JSON below — Cursor accepts the same mcpServers config.

VS Code

code --add-mcp '{"name":"gpu-mcp-server","command":"docker","args":["ghcr.io/pmady/gpu-mcp-server:v0.1.0"]}'

JSON

{
  "mcpServers": {
    "gpu-mcp-server": {
      "command": "docker",
      "args": [
        "ghcr.io/pmady/gpu-mcp-server:v0.1.0"
      ]
    }
  }
}