woo-guard
Pass
Audited by Gen Agent Trust Hub on Oct 29, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONMETADATA_POISONING
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted code for review, creating a surface for embedded instructions to potentially influence agent behavior. * Ingestion points: SKILL.md (Review and Guard-pass modes). * Boundary markers: Not explicitly defined for the untrusted data. * Capability inventory: Code analysis and report generation. * Sanitization: None defined.
- [METADATA_POISONING]: The skill frontmatter includes a future date (2026-07-13) in the date_added field, which is a deceptive metadata artifact.
Audit Metadata