grill-me
Pass
Audited by Gen Agent Trust Hub on Oct 3, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill functions as an interactive interview based on user-provided design documents and plans. This input acts as a potential surface for indirect prompt injection if malicious instructions are embedded within the design text. \n
- Ingestion points: User-provided plans or design documents processed by the interview logic. \n
- Boundary markers: Not explicitly present in the provided instruction files. \n
- Capability inventory: The skill instructs the agent to call the 'grilling' tool to perform the analysis. \n
- Sanitization: No evidence of input sanitization or filtering was found in the skill instructions. \n- [SAFE]: The skill acts as a benign entry point for design reviews, using a modular approach to call sub-skills. All external references point to the author's own repositories or established upstream sources. No persistence mechanisms, privilege escalation, or data exfiltration patterns were detected.
Audit Metadata