agri-review
Pass
Audited by Gen Agent Trust Hub on Jul 26, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill instructions define a legitimate scientific review workflow. It delegates execution to a related parent skill ('food-review') and references domain-specific knowledge files located within the project structure. All resources originate from the vendor's internal environment.
- [PROMPT_INJECTION]: The skill inherently processes untrusted data from manuscripts and cited sources, representing an indirect prompt injection surface. However, it includes safety-oriented instructions such as a 'privacy scan' and a strict rule to mark unretrieved sources rather than summarizing them, which serves as a mitigation. No adversarial patterns or safety bypasses were detected in the prompt logic.
- [DATA_EXFILTRATION]: No network operations, external data transfer mechanisms, or exfiltration patterns were found. The skill is designed to produce local Word documents (.docx) and place comments within manuscripts, which are standard file-system operations for its intended purpose.
- [CREDENTIALS_UNSAFE]: No hardcoded API keys, tokens, or sensitive credentials were found. The skill manages knowledge through provided reference files and standard bibliographic audits.
Audit Metadata