widgets-ui
Pass
Audited by Gen Agent Trust Hub on Sep 8, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill documentation includes a command to add UI components using
npx shadcn@latest add https://ui.inference.sh/r/widgets.json. This fetches component definitions and code from a remote vendor-controlled registry. - [INDIRECT_PROMPT_INJECTION]: The skill defines a mechanism for rendering UI based on structured agent responses (JSON). This creates an attack surface where an agent could be influenced to generate deceptive or malicious UI components (e.g., phishing forms or unauthorized buttons).
- Ingestion points: Data passed to the
WidgetRenderercomponent inSKILL.md. - Boundary markers: None explicitly defined; the skill expects structured JSON input.
- Capability inventory: Includes support for buttons with actions (
onClickAction), forms (asForm), and various input types. - Sanitization: Not specified in the documentation; integrity relies on the underlying React component implementation.
Audit Metadata