skill-security-auditor
Pass
Audited by Gen Agent Trust Hub on Sep 28, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill serves as a security auditor. Automated scan alerts for prompt injection (e.g., 'Ignore previous instructions') and remote code execution (e.g., 'python -c') are triggered by the skill's internal documentation and signature tables, which define malicious patterns for detection.\n- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted input in the form of target skill directories or remote repository URLs, which constitutes a potential attack surface.\n
- Ingestion points: The skill consumes the directory path or repository URL of a target skill to perform its audit (SKILL.md).\n
- Boundary markers: Explicit prompt boundary markers are not defined in the skill instructions to separate analysis logic from the content of the skill being audited.\n
- Capability inventory: The skill utilizes binary tools such as
node,grep, andsha256sum, and referenced workflows imply the use ofpythonandsemgrep(SKILL.md, references/workflows.md).\n - Sanitization: The skill incorporates its own sanitization logic via regex patterns, AST analysis, and metadata validation to identify and isolate malicious content in the inputs (references/tools.md).
Audit Metadata