frontend-design
Pass
Audited by Gen Agent Trust Hub on Aug 15, 2026
Risk Level: SAFEPROMPT_INJECTIONNO_CODE
Full Analysis
- [PROMPT_INJECTION]: The skill explicitly directs the agent to prioritize user-provided design briefs, stating 'the brief's own words always win,' which could allow embedded malicious instructions to override the agent's safety or behavior constraints.
- [PROMPT_INJECTION]: The skill presents a surface for indirect prompt injection as it processes untrusted data (design briefs) to perform code generation tasks. Ingestion points: User-supplied design briefs (referenced in SKILL.md). Boundary markers: Not specified. Capability inventory: Generation of HTML and CSS code. Sanitization: No sanitization or filtering of input data is defined.
Audit Metadata