image-to-code
Pass
Audited by Gen Agent Trust Hub on Oct 1, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes user-provided images and screenshots, which serves as a vector for instructions hidden within visual data. 1. Ingestion points: Design reference images, screenshots, and mockups analyzed in Step 1. 2. Boundary markers: Absent; there are no specific delimiters or instructions for the agent to ignore text content within images. 3. Capability inventory: The skill can write component files and execute local Python and Node.js scripts. 4. Sanitization: Validation is performed by local tools such as contrast.py, lint_hardcodes.py, and npm run verify to ensure the output adheres to design system constraints.
- [COMMAND_EXECUTION]: The skill executes local scripts for workspace tasks including design system searches and rendering verification. It specifically calls python3 for design_systems.py and node for measure_render.mjs. The search functionality uses inferred terms from the reference design, which could be exploited for command injection if input is not sanitized.
Audit Metadata