design-critic-skill
Pass
Audited by Gen Agent Trust Hub on Jul 29, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [SAFE]: No security issues were detected. The skill instructions focus on design critique and follow privacy-oriented workflows.\n- [PROMPT_INJECTION]: Evaluated the attack surface for indirect prompt injection. The skill processes external visual and code data (screenshots, Figma frames, and code context) to generate design critiques. This risk is mitigated by explicit privacy guidelines and the requirement for user approval before using external cloud services.\n
- Ingestion points: Screenshots, Figma frames, and local code source context as defined in SKILL.md.\n
- Boundary markers: No explicit boundary markers or "ignore instructions" warnings are defined to isolate user-provided data from the agent's instructions.\n
- Capability inventory: The skill is limited to generating design reports and fix-briefs; it does not possess tools for unauthorized file-system writes or outbound network communication.\n
- Sanitization: The skill relies on model-level safety guardrails without specifying internal sanitization or filtering logic.
Audit Metadata