The Agent Skills Directory

[DYNAMIC_EXECUTION]: The scripts/lib/browser.mjs file resolves and imports the playwright module using computed paths that include the current working directory and user cache. This is a functional requirement for the skill's browser-based auditing but introduces a potential risk of loading unintended local modules.
[INDIRECT_PROMPT_INJECTION]: The skill ingests and parses HTML from external URLs (Ingestion points: audit-markup.mjs, observe-runtime.mjs). Maliciously crafted HTML could attempt to influence the agent's analysis. The skill includes task-specific reporting labels as boundary markers and instructs the agent to cite official sources for legal claims. However, the custom HTML parser lacks safety-specific sanitization for prompt construction.
[COMMAND_EXECUTION]: The skill's operation requires the agent to execute Node.js scripts that perform network requests and control a headless browser. While necessary for accessibility testing, these represent a high capability tier that should be monitored.

wcag-readiness