image-to-code
Pass
Audited by Gen Agent Trust Hub on Aug 3, 2026
Risk Level: SAFE
Full Analysis
- [PROMPT_INJECTION]: The skill uses strong directive language like 'CORE DIRECTIVE' and 'MANDATORY RULE' to ensure the AI follows a specific multi-step design workflow. These directives are legitimate task-oriented instructions and do not seek to override the AI's underlying safety filters or reveal system-level information.
- [INDIRECT_PROMPT_INJECTION]: The skill represents a surface for indirect prompt injection as it processes untrusted user input to generate website designs and code. 1. Ingestion points: User design requests (e.g., 'make me one hero section') as documented in the example interpretations within 'SKILL.md'. 2. Boundary markers: None identified; user input is not explicitly delimited from the skill's core instructions. 3. Capability inventory: The skill defines image generation and code implementation tasks; it does not request high-privilege tool access or network operations beyond its stated design purpose. 4. Sanitization: No specific sanitization or validation of user-provided input is mentioned.
Audit Metadata