review-hog-resolution-criteria-small
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 review threads which could contain malicious instructions designed to subvert the agent's logic or inject unauthorized code changes.
- Ingestion points: External pull request comments and conversation threads are ingested as the primary data source for decision-making (SKILL.md).
- Boundary markers: Absent. The instructions lack specifications for delimiters or specific safety prompts to separate untrusted user input from the agent's core instructions.
- Capability inventory: The agent has the capability to implement code changes and commit them to the repository ('fixed' outcome).
- Sanitization: Absent. No explicit sanitization or escaping of the ingested comment text is performed. The skill includes a mitigation requiring the agent to 'independently confirm the problem' for unknown authors, but the technical threat surface remains.
Audit Metadata