ui-design-continuity
Pass
Audited by Gen Agent Trust Hub on Oct 8, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user feedback to generate UI design patches. It implements a structured 'Scoped Feedback Protocol' to normalize user intent into specific fields, which acts as a logical boundary against malicious instructions.
- Ingestion points: User feedback strings processed in
SKILL.mdandreferences/feedback-protocol.md. - Boundary markers: Use of an 'Output contract' and specific region maps (immutable vs. page-local) to limit the agent's scope.
- Capability inventory: The skill does not exhibit network, file-system write, or shell execution capabilities; it generates design briefs for other skills.
- Sanitization: Raw feedback is converted into a structured patch format to prevent direct execution of instructions hidden in user input.
- [SAFE]: No evidence of credential theft, obfuscation, or unauthorized remote code execution was found. The skill's instructions are consistent with its purpose of managing UI design maturity and continuity.
Audit Metadata