rag-security-reviewer
Pass
Audited by Gen Agent Trust Hub on Aug 6, 2026
Risk Level: SAFEPROMPT_INJECTIONNO_CODE
Full Analysis
- [PROMPT_INJECTION]: The skill includes example injection payloads like "Ignore previous instructions" within documentation blocks. These are clearly marked as test cases for developers to evaluate their RAG systems and do not represent attempts to hijack the agent's own behavior.
- [NO_CODE]: The skill contains only informational text and markdown tables. It does not define tools, execute shell commands, or include any script files.
- [SAFE]: The content aligns with security best practices, advocating for robust access controls, PII redaction, and audit logging. No exfiltration, obfuscation, or persistence patterns were found.
Audit Metadata