responsive-preview
Pass
Audited by Gen Agent Trust Hub on Jul 25, 2026
Risk Level: SAFEDATA_EXFILTRATIONPROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [PROMPT_INJECTION]: The skill instructs the agent to read and evaluate the DOM and accessibility tree of a user-provided URL. Because this content originates from an untrusted source, it could contain hidden instructions designed to manipulate the agent's findings or subsequent behavior.
- Ingestion points: Browser DOM and accessibility tree of the target URL (SKILL.md, Step 3).
- Boundary markers: None; there are no instructions to the agent to disregard or delimit embedded content within the page being analyzed.
- Capability inventory: Browser resizing, screenshotting, DOM reading, and chat-based reporting.
- Sanitization: No explicit sanitization or filtering of the web page content is performed before the agent processes it.
- [DATA_EXFILTRATION]: The skill provides a link to an external service (
tinydesignshop.com) that appends the target URL as a query parameter. If the agent is used to preview local development servers or internal private paths, clicking this link would expose those internal URLs and directory structures to the external domain. - [EXTERNAL_DOWNLOADS]: The skill references an external URL for a live preview tool:
https://tinydesignshop.com/tools/responsive-preview/.
Audit Metadata