visual-design-system

Pass

Audited by Gen Agent Trust Hub on Jun 16, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill uses subprocess.run across several files (e.g., g2_charts/grammar_renderer.py, antv_infographic/scripts/antv_renderer.py) to execute local Node.js scripts. This acts as a bridge to leverage specialized JavaScript charting libraries from a Python environment.
  • [EXTERNAL_DOWNLOADS]: The infographic generation tool creates HTML files that load the AntV framework from the well-known unpkg.com CDN (https://unpkg.com/@antv/infographic@latest). This is a legitimate way to include standard visualization dependencies.
  • [PROMPT_INJECTION]: The skill ingests untrusted clinical and statistical data for visualization, representing an indirect prompt injection surface.
  • Ingestion points: External data is passed to chart builders in files like stat_card.py, forest_plot.py, and comparison.py.
  • Boundary markers: Data labels are interpolated directly into chart specifications without explicit boundary markers.
  • Capability inventory: The skill possesses command execution capabilities through its Python-to-Node subprocess bridge.
  • Sanitization: There is no evidence of input filtering for data strings passed to the rendering engine, though the risk is localized to the visualization output.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 16, 2026, 02:11 AM
Security Audit — agent-trust-hub — visual-design-system