The Agent Skills Directory

[SAFE]: The skill analyzes local files within the agents/ and skills/ directories using standard utilities like grep, wc, and ls to verify structural compliance and content depth. These operations are limited to the repository scope and do not involve sensitive data access.
[SAFE]: Code quality verification is performed using python3 -m py_compile, which checks for syntax errors without executing the script. The subprocess.run call in scripts/validate.py uses a fixed list of arguments and does not invoke a shell, mitigating command injection risks.
[SAFE]: No network operations, credential exposures, or persistence mechanisms were found. The skill operates entirely in a local, read-only manner relative to the files it evaluates.

agent-evaluation