image-to-code
Pass
Audited by Gen Agent Trust Hub on Aug 20, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill instructs the agent to extract text from generated images and use it to guide implementation (Rules 9 and 21). This creates an indirect prompt injection surface where instructions embedded in the generated design could potentially override intended behavior during the implementation phase.
- Ingestion points: User-provided design prompts and text extracted from generated section images (SKILL.md).
- Boundary markers: Absent. No explicit instructions are provided to sanitize or ignore instructions found within extracted text.
- Capability inventory: Image generation and frontend code implementation.
- Sanitization: Absent. The skill lacks validation steps for text extracted from visual components.
Audit Metadata