preview-svg

Pass

Audited by Gen Agent Trust Hub on Jul 21, 2026

Risk Level: SAFECOMMAND_EXECUTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill invokes the system's default browser (using open, xdg-open, or start) to display generated SVG previews. The file paths used for these commands are strictly sanitized to prevent path traversal or command injection.
  • [SAFE]: The skill safely handles potentially untrusted SVG data by implementing a multi-layered defense strategy:
  • Ingestion points: SVG content is ingested via standard input or local files in the run.sh script.
  • Boundary markers: While explicit delimiters are not added to the SVG content itself, the input is encapsulated within a controlled HTML generation process.
  • Capability inventory: The skill has the ability to write HTML and JavaScript files to the project-local .preview-skills/ directory and launch the system's default browser to display them.
  • Sanitization: Content is transported using Base64 encoding to prevent injection or script breakout during the HTML generation phase. The rendered preview environment is further secured with a restrictive Content Security Policy (CSP) that blocks unauthorized script execution, external network connections, and framing, effectively mitigating risks associated with malicious SVG payloads.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 21, 2026, 06:51 PM
Security Audit — agent-trust-hub — preview-svg