Export Engine

Pass

Audited by Gen Agent Trust Hub on Apr 30, 2026

Risk Level: SAFEPROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [PROMPT_INJECTION]: The functions composeSVG and generateFigmaJSON in SKILL.md directly interpolate layer data into output strings. This presents a surface for indirect prompt injection if the generated files are later read by an agent.
  • Ingestion points: layer.name and layer.svgContent in composeSVG; layer.name in generateFigmaJSON.
  • Boundary markers: None present in the interpolation templates.
  • Capability inventory: Generation of SVG, PSD, and JSON files for browser download.
  • Sanitization: Input data is not sanitized or escaped before being placed into XML/JSON structures.
  • [EXTERNAL_DOWNLOADS]: The skill uses the vtracer and ag-psd libraries for WASM-based image vectorization and PSD creation. These are standard third-party dependencies for the declared graphics export tasks.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 30, 2026, 06:24 AM
Security Audit — agent-trust-hub — Export Engine