ux-optimization
Pass
Audited by Gen Agent Trust Hub on Mar 11, 2026
Risk Level: SAFE
Full Analysis
- [PROMPT_INJECTION]: No attempts to override agent behavior or bypass safety guidelines were detected. The instructions are purely functional and focused on UX methodology.
- [DATA_EXFILTRATION]: The skill uses client-side
localStoragefor UI preference persistence and form draft saving, which is standard practice for the described patterns. No unauthorized exfiltration or access to sensitive file paths was found. - [EXTERNAL_DOWNLOADS]: All external references target reputable technology organizations, research institutions, and UX practitioners (e.g., Google, Microsoft, Shopify, Nielsen Norman Group). No unauthorized remote code execution patterns were found.
- [INDIRECT_PROMPT_INJECTION]: The skill provides capability for screenshot-based UX diagnosis. It proactively mitigates privacy risks by including mandatory instructions for users to anonymize screenshots and remove personally identifiable information (PII) before uploading.
- [REMOTE_CODE_EXECUTION]: No dynamic code loading or remote script execution (e.g.,
curl | bash) is present. JavaScript examples provided in the practices are static templates for developer implementation.
Audit Metadata