The Agent Skills Directory

[SAFE]: The skill provides educational guidance and defensive code patterns. Vulnerable code snippets are explicitly marked as bad examples for instruction purposes.
[COMMAND_EXECUTION]: Includes a local script scripts/security-audit.py that scans files for security patterns using Python standard libraries. It is intended for internal auditing and does not interact with external networks.
[CREDENTIALS_UNSAFE]: Examples in SKILL.md use placeholder secrets (e.g., 'sk-1234567890abcdef', 'p@ssw0rd') to demonstrate security risks; no actual credentials are included.
[DATA_EXFILTRATION]: No network operations or data transmission logic are present in any of the skill's components.

owasp