The Agent Skills Directory

[SAFE]: The skill provides a legitimate security audit methodology including reconnaissance, testing for common vulnerabilities (XSS, SQLi, broken auth), and reporting. No attempts to bypass agent safety or perform unauthorized actions were found.\n- [NO_CODE]: This skill consists exclusively of markdown documentation and JSON metadata; it does not include any Python or Node.js scripts, binaries, or remote code dependencies.\n- [INDIRECT_PROMPT_INJECTION]: The skill is designed to process data from external web targets, creating an attack surface for indirect prompt injection. However, since the skill provides no automated tools or executable capabilities, the risk is negligible. 1. Ingestion points: SKILL.md (Testing and API review sections). 2. Boundary markers: Absent. 3. Capability inventory: None (no code provided). 4. Sanitization: Absent.

security-audit