The Agent Skills Directory

[SAFE]: The skill is entirely instructional and does not contain any executable scripts, automated tool invocations, or network operations. Analysis of the markdown content reveals no prompt injection attempts or malicious behavior.
[COMMAND_EXECUTION]: The instructions suggest the use of common system performance and profiling tools (such as uptime, top, perf, and strace) for manual or agent-assisted performance analysis. These are standard tools used for the skill's intended purpose of benchmarking and do not constitute a security risk in this context.
[EXTERNAL_DOWNLOADS]: A reference to a legitimate external repository (github.com/cursor/plugins) is provided in the documentation for licensing and attribution purposes, which is a standard and safe practice.

benchmark-checklist