ryux-visual-qa
Pass
Audited by Gen Agent Trust Hub on Oct 3, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to execute
npx playwright screenshotto capture visual representations of web pages. This is a standard and necessary operation for UI/UX testing and visual regression analysis. - [EXTERNAL_DOWNLOADS]: The use of
npx playwrightinvolves the potential download and execution of the Playwright runner and browser binaries. These are well-known tools from a reputable source, and their use is consistent with the skill's primary function. - [INDIRECT_PROMPT_INJECTION]: The skill processes external content by rendering and inspecting URLs provided during the QA process.
- Ingestion points: External web content accessed via
<url>inSKILL.md. - Boundary markers: None explicitly defined for the URL content, although the structured comparison table serves as a functional boundary.
- Capability inventory: Command execution via
npx playwright. - Sanitization: No specific sanitization of the rendered content is mentioned, but the human-in-the-loop nature of visual QA (comparing expected vs actual) mitigates the risk of the agent autonomously following instructions found within the target UI.
Audit Metadata