json-render-generative-ui

Generative UI framework that renders AI-generated JSON specs into type-safe UI components across React, Vue, Svelte, Solid, React Native, video, PDF, and email.

# json-render Generative UI Framework

> Skill by [ara.so](https://ara.so) — Daily 2026 Skills collection.

json-render is a **Generative UI framework** that lets AI generate dynamic interfaces from natural language prompts, constrained to a predefined component catalog. AI outputs JSON; json-render renders it safely and predictably across any platform.

## Installation

```bash
# React (core)
npm install @json-render/core @json-render/react

# React + shadcn/ui (36 pre-built components)
npm install @json-render/shadcn

# React Native
npm install @json-render/core @json-render/react-native

# Vue
npm install @json-render/core @json-render/vue

# Svelte
npm install @json-render/core @json-render/svelte

# SolidJS
npm install @json-render/core @json-render/solid

# Video (Remotion)
npm install @json-render/core @json-render/remotion

# PDF
npm install @json-render/core @json-render/react-pdf

# Email
npm install @json-render/core @json-render/react-email @react-email/components @react-email/render

# 3D (React Three Fiber)
npm install @json-render/core @json-render/react-three-fiber @react-three/fiber @react-three/drei three

# OG Images / SVG / PNG
npm install @json-render/core @json-render/image

# State management adapters
npm install @json-render/zustand   # or redux, jotai, xstate

# MCP integration (Claude, ChatGPT, Cursor)
npm install @json-render/mcp

# YAML wire format
npm install @json-render/yaml
```

## Core Concepts

| Concept | Description |
|---|---|
| **Catalog** | Defines allowed components and actions (the guardrails for AI) |
| **Spec** | AI-generated JSON describing which components to render and with what props |
| **Registry** | Maps catalog component names to actual render implementations |
| **Renderer** | Platform-specific component that takes a spec + registry and renders UI |
| **Actions** | Named events AI can trigger (e.g. `export_report`, `refresh_data`) |

## Spec Format

The flat spec format uses a root key + elements map:

```typescript
const spec = {
  root: "card-1",
  elements: {
    "card-1": {
      type: "Card",
      props: { title: "Dashboard" },
      children: ["metric-1", "metric-2", "button-1"],
    },
    "metric-1": {
      type: "Metric",
      props: { label: "Revenue", value: "124000", format: "currency" },
      children: [],
    },
    "metric-2": {
      type: "Metric",
      props: { label: "Growth", value: "0.18", format: "percent" },
      children: [],
    },
    "button-1": {
      type: "Button",
      props: { label: "Export Report", action: "export_report" },
      children: [],
    },
  },
};
```

## Step 1: Define a Catalog

```typescript
import { defineCatalog } from "@json-render/core";
import { schema } from "@json-render/react/schema";
import { z } from "zod";

const catalog = defineCatalog(schema, {
  components: {
    Card: {
      props: z.object({ title: z.string() }),
      description: "A card container with a title",
    },
    Metric: {
      props: z.object({
        label: z.string(),
        value: z.string(),
        format: z.enum(["currency", "percent", "number"]).nullable(),
      }),
      description: "Displays a single metric value with optional formatting",
    },
    Button: {
      props: z.object({
        label: z.string(),
        action: z.string(),
      }),
      description: "Clickable button that triggers an action",
    },
    Stack: {
      props: z.object({
        direction: z.enum(["row", "column"]).default("column"),
        gap: z.number().optional(),
      }),
      description: "Layout container that stacks children",
    },
  },
  actions: {
    export_report: { description: "Export the current dashboard to PDF" },
    refresh_data: { description: "Refresh all metric data" },
    navigate: {
      description: "Navigate to a page",
      payload: z.object({ path: z.string() }),
    },
  },
});
```

## Step 2: Define a Registry (React)

```tsx
import { defineRegistry, Renderer } from "@json-render/react";

function format(value: string, fmt: string | null): string {
  if (fmt === "currency") return `$${Number(value).toLocaleString()}`;
  if (fmt === "percent") return `${(Number(value) * 100).toFixed(1)}%`;
  return value;
}

const { registry } = defineRegistry(catalog, {
  components: {
    Card: ({ props, children }) => (
      <div className="rounded-lg border p-4 shadow-sm">
        <h3 className="text-lg font-semibold mb-3">{props.title}</h3>
        {children}
      </div>
    ),

    Metric: ({ props }) => (
      <div className="flex flex-col">
        <span className="text-sm text-gray-500">{props.label}</span>
        <span className="text-2xl font-bold">
          {format(props.value, props.format)}
        </span>
      </div>
    ),

    Button: ({ props, emit }) => (
      <button
        className="px-4 py-2 bg-blue-600 text-white rounded"
        onClick={() => emit("press")}
      >
        {props.label}
      </button>
    ),

    Stack: ({ props, children }) => (
      <div
        style={{
          display: "flex",
          flexDirection: props.direction ?? "column",
          gap: props.gap ?? 8,
        }}
      >
        {children}
      </div>
    ),
  },
});
```

## Step 3: Render the Spec

```tsx
import { Renderer } from "@json-render/react";

function Dashboard({ spec, onAction }) {
  return (
    <Renderer
      spec={spec}
      registry={registry}
      onAction={(action, payload) => {
        console.log("Action triggered:", action, payload);
        onAction?.(action, payload);
      }}
    />
  );
}
```

## Generating Specs with AI (Vercel AI SDK)

```typescript
import { generateObject } from "ai";
import { openai } from "@ai-sdk/openai";
import { getCatalogSchema, getCatalogPrompt } from "@json-render/core";

async function generateDashboard(userPrompt: string) {
  const { object: spec } = await generateObject({
    model: openai("gpt-4o"),
    schema: getCatalogSchema(catalog),
    system: getCatalogPrompt(catalog),
    prompt: userPrompt,
  });

  return spec;
}

// Usage
const spec = await generateDashboard(
  "Create a sales dashboard showing revenue, conversion rate, and an export button"
);
```

## Streaming Specs

```tsx
import { streamObject } from "ai";
import { openai } from "@ai-sdk/openai";
import { getCatalogSchema, getCatalogPrompt, parseSpecStream } from "@json-render/core";
import { Renderer } from "@json-render/react";
import { useState, useEffect } from "react";

function StreamingDashboard({ prompt }: { prompt: string }) {
  const [spec, setSpec] = useState(null);

  useEffect(() => {
    async function stream() {
      const { partialObjectStream } = await streamObject({
        model: openai("gpt-4o"),
        schema: getCatalogSchema(catalog),
        system: getCatalogPrompt(catalog),
        prompt,
      });

      for await (const partial of partialObjectStream) {
        setSpec(partial); // Renderer handles partial specs gracefully
      }
    }
    stream();
  }, [prompt]);

  if (!spec) return <div>Generating UI...</div>;
  return <Renderer spec={spec} registry={registry} />;
}
```

## Using Pre-built shadcn/ui Components

```tsx
import { defineCatalog } from "@json-render/core";
import { schema } from "@json-render/react/schema";
import { defineRegistry, Renderer } from "@json-render/react";
import { shadcnComponentDefinitions } from "@json-render/shadcn/catalog";
import { shadcnComponents } from "@json-render/shadcn";

// Pick any of the 36 available shadcn components
const catalog = defineCatalog(schema, {
  components: {
    Card: shadcnComponentDefinitions.Card,
    Stack: shadcnComponentDefinitions.Stack,
    Heading: shadcnComponentDefinitions.Heading,
    Text: shadcnComponentDefinitions.Text,
    Button: shadcnComponentDefinitions.Button,
    Badge: shadcnComponentDefinitions.Badge,
    Table: shadcnComponentDefinitions.Table,
    Chart: shadcnComponentDefinitions.Chart,
    Input: shadcnComponentDefinitions.Input,
    Select: shadcnComponentDefinitions.Select,
  },
  action