device-frames
Pass
Audited by Gen Agent Trust Hub on Nov 18, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process external data such as screenshots, video recordings, and live web pages to place them inside device frames.
- Ingestion points: Found in
SKILL.md, where the agent is instructed to use a 'supplied screenshot, recording, page, or component' as the central content for the frame. - Boundary markers: There are no explicit delimiters or 'ignore instructions' warnings provided to the agent regarding the content of the media being framed.
- Capability inventory: The skill's capabilities are limited to generating visual outputs (CSS/SVG drawings, image renders, video composites) and live UI components.
- Sanitization: No sanitization or validation of the ingested media's content is specified, which is common for graphical processing tasks.
- [SAFE]: The core functionality of the skill involves providing static metadata (screen points, millimeter measurements, and bezel widths) for various devices. It uses standard vector-based layout logic and does not include any commands for network exfiltration, persistence, or credential access.
Audit Metadata