prototype
Pass
Audited by Gen Agent Trust Hub on Oct 2, 2026
Risk Level: SAFECOMMAND_EXECUTIONDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill requires the agent to manage files by creating prototype routes and files, and subsequently deleting them after a variant is selected or integrated (Phase 4 and Phase 6).
- [DYNAMIC_EXECUTION]: The picker harness renders agent-generated code variants dynamically using innerHTML as specified in the reference wiring in PICKER.md.
- [INDIRECT_PROMPT_INJECTION]: The skill ingests user-supplied UI descriptions to generate and execute code, establishing an injection surface.
- Ingestion points: User-provided descriptions in the initial invocation and Phase 1.
- Boundary markers: No specific delimiters or instructions to ignore embedded commands are present for the user descriptions.
- Capability inventory: File system write/delete operations and dynamic DOM manipulation (innerHTML).
- Sanitization: No explicit sanitization, validation, or escaping of the generated content is described.
Audit Metadata