design-system-to-skill
Pass
Audited by Gen Agent Trust Hub on Jul 14, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill includes a Node.js worker script (
scripts/convert-design-system.mjs) that implements a robustconfinedPathfunction to prevent path traversal attacks. This ensures that file operations are strictly limited to the intended bundle and staging directories. - [SAFE]: The skill implements strong mitigations against indirect prompt injection by defining a clear trust boundary. It instructs the agent to treat external design system content as 'DATA, NOT INSTRUCTIONS' and includes mandatory banners in the generated files to prevent the AI from executing instructions found in the source material.
- [SAFE]: The bundled script uses only Node.js built-in modules ('node:fs', 'node:path', etc.) and contains no external dependencies, eliminating supply chain risks from third-party packages.
- [SAFE]: The skill requires explicit user opt-in for configuration changes, such as designating a design system as 'MUST-USE' in the project's CLAUDE.md file, ensuring transparency and user control.
- [SAFE]: A quality self-check list (
references/description-checklist.md) is provided to verify that the generated skill descriptions are concise and free of verbatim injections from untrusted sources.
Audit Metadata