prototype
Pass
Audited by Gen Agent Trust Hub on Oct 9, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied descriptions to generate executable UI code variants. This creates a surface where malicious instructions in the description could influence the generated code. However, the skill operates in an isolated prototype environment by design.
- Ingestion points: User descriptions provided during invocation (SKILL.md, Phase 1).
- Boundary markers: No specific delimiters or safety instructions are defined for the input descriptions.
- Capability inventory: The skill writes prototype files (SKILL.md, Phase 4) and integrates code into the project upon user selection (SKILL.md, Phase 6).
- Sanitization: No explicit sanitization or validation of the input description is performed before code generation.
- [DYNAMIC_EXECUTION]: The provided reference implementation in PICKER.md uses innerHTML to dynamically mount and switch between UI variants in the prototype harness. This is a common pattern for local UI previewing but involves the execution of generated markup strings.
Audit Metadata