MCP UI Kit

Build interactive React UIs for MCP tools.

The Model Context Protocol (MCP) enables AI assistants to interact with external tools. mcp-ui-kit simplifies building rich UI components that render directly in AI chat:

• Bundling on-demand — Write React components, bundled automatically when the tool is called
• Zero config — No webpack/vite setup, just createUI() and point to your .tsx file
• Props from server — Pass data from your MCP tool directly to React via useProps()
• Two-way communication — Components can sendPrompt() to the AI or callTool() to invoke other MCP tools

MCP UI Inspector — Built on @mcp-ui/server

Installation

npm install mcp-ui-kit @modelcontextprotocol/sdk

Server Usage

import { McpServer } from '@modelcontextprotocol/sdk/server/mcp.js';
import { createUI } from 'mcp-ui-kit/server';

const server = new McpServer({ name: 'my-server', version: '1.0.0' });

const dashboardUI = createUI('my-dashboard', import.meta.resolve('./Dashboard.tsx'));

server.registerTool(
  'dashboard',
  { description: 'Interactive dashboard', _meta: dashboardUI.meta },
  async () => ({
    content: [
      { type: 'text', text: 'Dashboard loaded' },
      await dashboardUI.component({ props: { title: 'Hello' }, frameSize: ['700px', '500px'] })
    ]
  })
);

With Input Schema

import { z } from 'zod';
import { createUI } from 'mcp-ui-kit/server';

const stockUI = createUI('stocks', import.meta.resolve('./StockDashboard.tsx'));

server.registerTool(
  'stock_portfolio',
  {
    description: 'View stock portfolio',
    _meta: stockUI.meta,
    inputSchema: {
      symbols: z.array(z.string()).default(['AAPL', 'GOOGL']),
      timeframe: z.enum(['1D', '1W', '1M', '1Y']).default('1M'),
    },
  },
  async ({ symbols, timeframe }) => ({
    content: [
      { type: 'text', text: `Showing ${symbols.join(', ')} for ${timeframe}` },
      await stockUI.component({ props: { symbols, timeframe } })
    ]
  })
);

Client Usage

The file passed to createUI() must render a component to the #root element:

// StockDashboard.tsx
import { createRoot } from 'react-dom/client';
import { sendPrompt, callTool, useProps } from 'mcp-ui-kit/ui';

function StockDashboard() {
  const { symbols, timeframe } = useProps({ symbols: ['AAPL'], timeframe: '1M' });

  return (
    <div>
      <h1>Stock Portfolio ({timeframe})</h1>
      {symbols.map(symbol => (
        <div key={symbol}>
          <span>{symbol}</span>
          <button onClick={() => sendPrompt(`Analyze ${symbol} over ${timeframe}`)}>
            Analyze
          </button>
          <button onClick={() => callTool('get_stock_price', { symbol })}>
            Refresh
          </button>
        </div>
      ))}
    </div>
  );
}

// Render to DOM
createRoot(document.getElementById('root')!).render(<StockDashboard />);

API

Server (mcp-ui-kit/server)

createUI(name, componentPath) — Creates a UI component

• componentPath: Path to a .tsx file that renders to #root
• Returns { meta, component(opts?) }
• opts.props: Data passed to your React component via useProps()
• opts.frameSize: [width, height] e.g. ['700px', '500px']

createUI('dashboard', import.meta.resolve('./Dashboard.tsx'));  // ESM
createUI('dashboard', require.resolve('./Dashboard.tsx'));       // CommonJS

UI (mcp-ui-kit/ui)

• useProps(defaults) — Get props passed from the server
• sendPrompt(message) — Send a message to the AI chat
• callTool(name, params) — Invoke an MCP tool

Development

npm install
npm run dev:all  # Start demo server + inspector