branding
Pass
Audited by Gen Agent Trust Hub on Oct 3, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill fetches font files from the official Google Fonts repository on GitHub via curl.
- [COMMAND_EXECUTION]: Employs command-line utilities including Inkscape for SVG manipulation, ImageMagick for raster image processing, and the open command to display generated HTML content in a web browser.
- [INDIRECT_PROMPT_INJECTION]: The skill ingests user-provided project metadata that is subsequently interpolated into HTML preview pages and brand guides. It mitigates potential risks by specifically instructing the agent to use safe DOM manipulation methods such as textContent instead of innerHTML. Ingestion points occur during the discovery phase (SKILL.md, Phase 1), with capabilities including local file writing and execution of the browser via the open command. Sanitization is guided by explicit implementation instructions.
Audit Metadata