the-pdp-reviewer
Pass
Audited by Gen Agent Trust Hub on Aug 14, 2026
Risk Level: SAFE
Full Analysis
- [PROMPT_INJECTION]: The skill explicitly addresses the risk of indirect prompt injection by including a dedicated security reference (
references/agent-security.md). It instructs the agent to treat all fetched content as data rather than instructions and to report any steering attempts as findings. - [CREDENTIALS_UNSAFE]: Proactive credential protection is implemented through specific rules. The agent is instructed to detect sensitive strings like API keys or tokens in input data and flag them for rotation without reproducing or persisting them in any logs or output files.
- [DATA_EXFILTRATION]: No patterns of unauthorized data exfiltration were found. The skill adheres to data minimization principles, requiring the agent to persist only decisions and evidence rather than raw customer data, and to escape interpolated content to prevent injection attacks.
- [SAFE]: The skill's architecture is security-focused, utilizing structured auditing discipline and clear attribution. All external links point to the vendor's own domain (intempt.com), representing standard functional behavior for the service provided.
Audit Metadata