widgets-ui
Warn
Audited by Gen Agent Trust Hub on Feb 17, 2026
Risk Level: MEDIUMEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONPROMPT_INJECTION
Full Analysis
- EXTERNAL_DOWNLOADS (MEDIUM): The skill documentation recommends running
npx shadcn@latest add https://ui.inference.sh/r/widgets.json. This fetches configuration and component code from an unverified remote domain (inference.sh), which is not in the trusted source list. - REMOTE_CODE_EXECUTION (MEDIUM): Use of
npx skills add inference-sh/skills@agent-uisuggests the execution of a custom CLI tool to pull and integrate remote packages into the local environment. - PROMPT_INJECTION (LOW): [Indirect Prompt Injection Surface]
- Ingestion points: The
WidgetRenderercomponent accepts awidgetobject derived from "structured agent responses" (SKILL.md). - Boundary markers: None identified. There are no instructions to validate the structure or content of the agent-provided JSON before rendering.
- Capability inventory: The renderer supports interactive elements including buttons with actions and forms with data collection, which could be exploited if the agent is manipulated into generating malicious UI schemas.
- Sanitization: Not present in documentation.
Audit Metadata