review-hog-perspective-logic-correctness
Pass
Audited by Gen Agent Trust Hub on Oct 16, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted Pull Request code, which could contain instructions meant to deceive the agent. While the skill only performs search operations, the lack of clear boundary markers between code and instructions constitutes a minor injection surface. * Ingestion points: PR code chunks processed via ripgrep (SKILL.md). * Boundary markers: Absent in the provided instructions. * Capability inventory: File system searching via ripgrep (rg). * Sanitization: Not specified for the content of processed code.
Audit Metadata