live-panel

Pass

Audited by Gen Agent Trust Hub on Oct 22, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill includes a helper script scripts/fetch_fonts.sh that downloads a specific font file (Xiaolai-Regular.ttf) from a public GitHub repository (github.com/lxgw/kose-font). This is used to provide handwriting-style text in the 'excalidraw' diagram theme.
  • [COMMAND_EXECUTION]: The skill executes local system commands to perform rendering and validation:
  • scripts/livepanel.py launches Chrome/Chromium in headless mode using a DevTools pipe for high-fidelity screenshots.
  • scripts/render.py and scripts/check_frames.py execute FFmpeg to encode images into MP4 files and extract frames for comparison.
  • scripts/verify_view.py executes internal scripts for automated testing.
  • All subprocess calls use list-based arguments, which is a safe practice to prevent command injection.
  • [INDIRECT_PROMPT_INJECTION]: The skill exposes an ingestion surface by processing JSON configuration files (config.json). This configuration is interpolated into an HTML template for rendering. The skill mitigates potential script injection risks by escaping the </ tag sequence when stringifying the JSON into the HTML page context.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 22, 2026, 02:33 PM