The Agent Skills Directory

[PROMPT_INJECTION]: The skill's documentation files and pattern catalog contain numerous references to prompt injection strings (e.g., 'ignore previous instructions', 'DAN mode'). These are documented attack patterns that the tool is designed to identify in other skills. In the context of a security tool, these are benign and serve as reference data for the detection logic.
[COMMAND_EXECUTION]: The scripts/audit.py script executes the git clone command via subprocess.run to fetch remote repositories for analysis. This is a standard and necessary function for the tool's stated purpose. The implementation follows safe practices by using a list of arguments and avoiding shell execution.
[EXTERNAL_DOWNLOADS]: The tool facilitates the download of external repositories via git clone for the purpose of auditing them. This behavior is part of the core functionality and is triggered by explicit user request.
[SAFE]: The skill demonstrates security consciousness by implementing file size limits, safe text reading, HTML escaping for reports, and proper temporary file management. The overall behavior is consistent with its role as a security utility.

skill-security-auditor