What problem does it solve? Agents and LLM backends often need to return rich, interactive interfaces rather than plain text, but hand-coding React components for every response is impractical. This Skill lets you render structured JSON widget definitions into forms, cards, layouts, and inputs inside React or Next.js applications. ## Core Features & Use Cases - Declarative JSON-to-UI Rendering: Convert widget JSON (rows, columns, boxes, text, buttons, inputs, selects, checkboxes, badges, icons, images) into rendered components via a single WidgetRenderer. - Form Handling & Actions: Capture submit and click actions with collected form data through a unified onAction callback. - Themed Layouts: Apply gradient backgrounds, spacing, radius, and typography variants for polished cards such as flight summaries or contact forms. - Use Case: An AI agent returns a JSON widget describing a booking confirmation card with an input and submit button; the frontend renders it directly and sends the user's form data back to the agent. ## Quick Start Install the widget renderer with npx shadcn@latest add https://ui.inference.sh/r/widgets.json, then render a JSON widget with the WidgetRenderer component and handle actions via onAction.