figma-pixel-implement
Pass
Audited by Gen Agent Trust Hub on Oct 19, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests data from external Figma design nodes—including text content, layer names, and annotations—to drive code generation and documentation creation. This represents an indirect prompt injection surface where adversarial content in a design file could influence the agent's output.
- Ingestion points: Figma design-context payload (metadata, structure, and content) fetched via environment tools (SKILL.md, Workflow Step 2).
- Boundary markers: No specific delimiters or safety instructions are defined to separate untrusted design strings from the agent's instruction context.
- Capability inventory: File system write operations for creating components, Storybook stories, and Markdown documentation (SKILL.md, Workflow Steps 6-9).
- Sanitization: The instructions do not specify sanitization or validation routines for text strings extracted from the external design files.
- [EXTERNAL_DOWNLOADS]: The skill fetches design assets, including SVG and raster images, from Figma to be incorporated into the target repository (SKILL.md, Workflow Step 8).
- Source: Downloads originate from Figma, an established and well-known design service.
- Execution: Assets are stored as static files in the project's local asset directory and are not executed.
Audit Metadata