autofigure

Pass

Audited by Gen Agent Trust Hub on Apr 28, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill downloads background removal models from HuggingFace, a well-known and trusted repository for AI models, during its initial execution.
  • [PROMPT_INJECTION]: The processing of user-supplied figures via a multimodal LLM to generate structured XML creates a surface for indirect prompt injection where malicious instructions embedded in images could attempt to influence the output.
  • Ingestion points: User-provided image files processed by the tool workflow.
  • Boundary markers: No explicit delimiter or instruction-ignoring warnings are defined for the image processing step.
  • Capability inventory: The skill has the capability to write files to the local output directory and make network requests to the fal.ai API.
  • Sanitization: The workflow includes automatic XML validation and repair logic to ensure the integrity of the generated DrawIO templates.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 28, 2026, 03:54 PM