gpu-mcp-server¶
An MCP server that exposes NVIDIA GPU metrics as tools. Any MCP-compatible AI agent (Claude, Goose, Cursor, Windsurf) can query real-time GPU utilization, memory, temperature, power, PCIe and NVLink throughput — no Prometheus or dcgm-exporter required.
Built on the official Go MCP SDK and NVIDIA go-nvml.
Tools¶
| Tool | Description |
|---|---|
list_gpus |
List all GPUs with utilization and memory info |
get_gpu_metrics |
Detailed metrics for a GPU by index or UUID |
get_gpu_processes |
PID-level GPU process attribution |
gpu_summary |
Aggregate stats across all devices |
All tools support MIG (Multi-Instance GPU) — MIG instances appear as separate devices with their parent GPU's shared metrics (temperature, power, PCIe).
Quick links¶
- Getting Started — build and run in under 5 minutes
- Architecture — how the server and Collector interface work
- Tools Reference — detailed input/output for each tool
- Agent Integration — copy-paste configs for Claude, Goose, Cursor, Windsurf
Project info¶
- License: Apache 2.0
- Language: Go
- AAIF alignment: MCP
- Related: keda-gpu-scaler — GPU autoscaling for Kubernetes