runtime-diagnostics

Quick Review

The skill provides a clear high-level description and workflow for running system diagnostics, but lacks critical detail for autonomous invocation. The description mentions what diagnostics are collected (CPU, memory, GPU, NPU, Docker, K8s) but doesn't specify output format, failure modes, or how results are interpreted. Task knowledge is minimal—only a single command is provided with brief notes about retries and timeouts, but no detail on what the script actually does, what metrics are collected, or how to act on results. Structure is adequate for a simple skill with one main script. Novelty is low: running a Python diagnostics script is straightforward for a CLI agent and doesn't represent significant complexity reduction; most system diagnostic commands (top, nvidia-smi, docker ps, kubectl get nodes) are already simple CLI operations. The skill would benefit from more concrete examples of output interpretation, error handling procedures, and clearer guidance on when specific flags or elevated permissions are needed.