This skill implements a strict operational standard for managing local inference runtimes. It enforces a canonical directory structure, requires provenance tracking via manifests, and mandates smoke testing before any runtime is promoted to 'current' status.
Use this skill when installing new LLM inference engines, updating existing toolchains, or preparing a reproducible environment for benchmarking model performance.
manifest.json (tracking versions, hardware, and source) and profiles (model-specific serve flags).Suitable for technical agents managing HPC or AI research workstations, particularly those integrating with benchmark toolchains.
This skill has not been reviewed by our automated audit pipeline yet.