The Agent Skills Directory

[SAFE]: The skill provides a comprehensive framework for performance optimization, focusing on benchmarking, profiling, and controlled experimentation. The instructions promote best practices such as establishing baselines and verifying results with field data.
[COMMAND_EXECUTION]: The examples/membership_workload.py script uses subprocess.run to execute itself. This is used specifically to measure the end-to-end timing of process invocation as part of a synthetic benchmark exercise. The command is constructed using a fixed list with the current interpreter (sys.executable) and script path (__file__), preventing arbitrary command injection.
[EXTERNAL_DOWNLOADS]: The documentation includes links to established technical resources and documentation sites such as Cloudflare, OpenTelemetry, Go.dev, and Web.dev. These references are used for educational purposes and do not involve remote code execution or untrusted downloads.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to process external performance data, traces, and logs. While this constitutes an input surface, the methodology emphasizes disciplined data analysis and correctness checks, which serves as a natural mitigation against accidental obedience to instructions embedded in data.

performance-optimization