regards-art-direction
Pass
Audited by Gen Agent Trust Hub on Oct 14, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: Extensive review of the skill's scripts and instructions confirms no malicious behavior. The skill follows best practices for asset management and automated design testing.
- [EXTERNAL_DOWNLOADS]: The skill fetches design references and downloads media assets from established, well-known archives including Wikimedia Commons, The Metropolitan Museum of Art, the Art Institute of Chicago, NASA, and the Internet Archive. These operations are essential for the skill's function of finding and crediting honest imagery.
- [COMMAND_EXECUTION]: The skill utilizes
child_process.spawnto launch a headless browser (Chrome, Edge, or Chromium). This is used exclusively for theinspect.mjstool to render the project and perform layout measurements and screenshots for designer review. - [DYNAMIC_EXECUTION]: The skill performs execution of measurement logic within the headless browser context using
page.eval(). The JavaScript payloads (defined inprobe.mjsandframe.mjs) are static helpers used to verify typography, contrast, and motion rules. - [INDIRECT_PROMPT_INJECTION]: The skill has a data ingestion surface where it reads metadata (titles, descriptions) from external archives and the
regards.designindex. This information is processed by the agent to create design specimens. The skill provides clear instructions on how to handle and format this data, reducing the risk of unintended instruction following from external content.
Audit Metadata