design-workflow
Pass
Audited by Gen Agent Trust Hub on Sep 14, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes external design bundles (ZIP/HTML) and Figma data, creating an attack surface for instructions embedded in design metadata.
- Ingestion points: Processing of external ZIP/HTML bundles in SKILL.md.
- Boundary markers: Uses DTCG validation and schema normalization, though explicit prompt delimiters for ingested content are not defined.
- Capability inventory: File extraction, writing to .moai/design/, and invoking dynamic skill generation tools.
- Sanitization: Implements a security scan rejecting executables, symlinks, and path traversal in ZIP files, and strips script tags from SVG metadata.
- [DYNAMIC_EXECUTION]: The skill invokes a meta-harness to dynamically generate the 'harness-figma-extractor' skill. Runtime generation of executable instructions is a specialized pattern that requires monitoring.
- [COMMAND_EXECUTION]: Performs automated file system operations including ZIP file extraction and directory checks, which are functional requirements for design artifact management.
Audit Metadata