Enables Claude Code to benchmark llm-d from plain English, providing 35 tools and 5 workflow prompts for probing clusters, planning, deploying, running benchmarks, and analyzing results via the Model …