The Agent Skills Directory

[COMMAND_EXECUTION]: The usage_finder.py script executes system search utilities (rg or grep) via subprocess.run to locate symbol references. Security is maintained through a strict regex-based validation of the symbol identifier (^[A-Za-z_][A-Za-z0-9_.]*$) to prevent command injection.
[SAFE]: The doc_review.py script implements a hardened path validation method (_validate_path) that uses Path.resolve() and relative path comparisons to protect against path traversal attacks during documentation maintenance.
[SAFE]: The skill incorporates a comprehensive 'Verification Trust Model' and a 'Zero Assumptions' principle. It explicitly instructs the AI agent to disregard information in docstrings or comments as unverified until confirmed by structural code analysis, effectively mitigating the risk of indirect prompt injection from malicious source code comments.
[SAFE]: File modification tools such as comment_rewriter.py include safety features like temporary backups and encoding error handling to prevent data loss or corruption during code cleanup operations.

deep-dive-analysis