The Agent Skills Directory

[SAFE]: The skill's primary purpose is defensive. It provides a framework for auditing AI agent skills by scanning for known attack patterns (C2 activity, persistence, credential access, etc.).
[SAFE]: The instructions implement a structured process for security analysis, including pattern matching, file tree inspection, and AI-driven semantic evaluation to reduce false positives.
[SAFE]: The identified patterns in SKILL.md (e.g., searches for curl, base64, subprocess) are used as detection signatures for the scanner, not as part of the skill's own execution logic.
[SAFE]: The skill follows least-privilege principles by explicitly excluding sensitive directories like .git, node_modules, and .venv from its scan.
[SAFE]: No obfuscation, data exfiltration, or persistence mechanisms were found within the skill's own code or instructions.

scan-skill