The Agent Skills Directory

[SAFE]: The skill provides structured instructions for UX/UI tasks without any detected malicious patterns. It includes clear scope definitions and safety invariants, such as prioritizing human manual verification over automated accessibility scanners.
[DATA_EXFILTRATION]: No exfiltration patterns were identified. The skill explicitly requires security best practices for host integration, such as pinning the host origin and validating event.origin during postMessage communication to prevent data leakage.
[COMMAND_EXECUTION]: The skill utilizes browser tools for JavaScript execution and console reading strictly for UI verification purposes, such as asserting getComputedStyle to validate transitions. These actions are scoped to the development and audit environment.
[REMOTE_CODE_EXECUTION]: No suspicious remote code execution or external downloads were found. The skill.json references well-known and trusted documentation sources from organizations like W3C, Apple, and Google.
[PROMPT_INJECTION]: The instructions focus on routing and heuristics for interface design and do not contain patterns typical of prompt injection or safety filter bypass attempts.

ux