icon-draw
Pass
Audited by Gen Agent Trust Hub on Sep 6, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill adopts a strong security posture by validating all AI-generated SVG code against a strictly defined safe subset. The
check_svg.pyutility prevents the manifestation of malicious patterns by banning dangerous elements like and , as well as all external resource loading via url() or data: schemes. - [INDIRECT_PROMPT_INJECTION]: The skill processes user-defined concepts and design briefs to produce SVG candidates.
- Ingestion points: User concepts and icon-design/brief.md.
- Boundary markers: Not explicitly defined in interpolation logic.
- Capability inventory: File creation (SVG candidates) and subprocess execution of local Python utilities.
- Sanitization: The mandatory linting step ensures that user-provided text cannot result in the generation of active or malicious SVG content.
- [COMMAND_EXECUTION]: The skill executes local Python scripts (squircle.py and check_svg.py) to manage geometry and validation. These scripts are provided within the skill package, use only standard library modules, and do not pass unsanitized user input to shell commands or perform any network requests.
Audit Metadata