Last released Oct 31, 2026
Keeps a vLLM container from overheating an NVIDIA DGX Spark, and lets AI agents read their hardware temperature over MCP