The Agent Skills Directory

[SAFE]: The skill consists entirely of instructional markdown content. It establishes a 'Benchmark Checklist' for users and agents to follow when measuring software performance.
[SAFE]: No scripts, binaries, or automated command executions are present. The references to commands like uptime, nproc, top, pidstat, node --cpu-prof, py-spy, perf, and strace -c are provided as suggestions for a human or agent to manually use during a benchmarking process to gather evidence.
[SAFE]: Metadata fields in SKILL.md and configuration in agents/openai.yaml are benign and match the stated purpose of the skill.
[SAFE]: The UPSTREAM.md file provides transparent documentation of the skill's origin and local modifications, and references a well-known public repository (GitHub cursor/plugins) for its source material.

benchmark-checklist