The Agent Skills Directory

[SAFE]: The skill's instructions and capabilities are aligned with its intended purpose of visual testing. No evidence of malicious intent or deceptive practices was found.\n- [PROMPT_INJECTION]: Indirect Prompt Injection Surface: The skill processes untrusted external data by ingesting web content via content_get-as-html and a11y_take-aria-snapshot. While the skill does not explicitly define boundary markers for this untrusted input, its core capabilities (screenshot generation and UI consistency reporting) are low-privilege operations that do not expose sensitive system resources. This surface is inherent to browser-based automation skills and is considered safe within the context of UI verification.

visual-testing