The Agent Skills Directory

[SAFE]: The skill instructions in SKILL.md and the language-specific references provide legitimate, high-quality advice for performance engineering, focusing on profiling, benchmarking, and measured code changes.
[SAFE]: The Python utility scripts/compare_benchmarks.py is a standard data processing script for comparing CSV-formatted benchmark results. It uses the csv and decimal libraries to perform numerical comparisons without any dangerous system calls or network activity.
[SAFE]: The evaluation framework in evals/evals.json contains automated test assertions that use standard language toolchains (Go, Cargo, Javac). These are used to verify the skill's effectiveness in a development environment and do not contain malicious payloads.
[SAFE]: Non-ASCII characters found in the evaluation tests (e.g., 'ключ' and 'żółw') are legitimate test cases for verifying how encoders and indexers handle multi-byte characters and are not homoglyph attacks.
[SAFE]: The mock client in evals/files/fetch/client.py simulates network latency using time.sleep but does not perform actual network operations, posing no data exfiltration risk.

optimize-code-performance