The Agent Skills Directory

[SAFE]: The skill provides a set of structured rules and workflows for auditing web and mobile UI code. It includes detailed checklists for accessibility (WCAG 2.2), core UX heuristics, and platform-specific guidelines (Apple HIG and Material Design 3).
[SAFE]: A significant portion of the skill is dedicated to identifying and preventing 'Dark Patterns' (deceptive design). The instructions explicitly command the agent to refuse user requests to implement manipulative interfaces, which serves as a safety guardrail.
[COMMAND_EXECUTION]: The skill uses git diff --name-only HEAD~1 to identify relevant files for review. This is a standard and expected operation for a tool designed to integrate with developer workflows and pull request reviews.
[INDIRECT_PROMPT_INJECTION]: As a code auditing tool, the skill naturally ingests untrusted code provided by users or from repository history. This represents a known attack surface common to all LLM-based development tools. However, the skill implements a structured reporting format and specific auditing rules which help maintain focus on the intended task.

ux