implementation
Pass
Audited by Gen Agent Trust Hub on Oct 20, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted design specifications and directions to generate frontend code, which creates a potential surface for indirect prompt injection if the design source contains adversarial instructions.
- Ingestion points: The skill ingests user-provided UI directions and the contents of a
DESIGN.mdfile to determine implementation details. - Boundary markers: There are no explicit boundary markers or instructions defined to prevent the agent from following commands that might be embedded within the design input.
- Capability inventory: The skill is capable of generating complete React and Next.js application code, including component structures, hooks, and style implementations.
- Sanitization: No sanitization or filtering logic is described for the input design descriptions before they are processed by the agent.
Audit Metadata