UNPKG

ai

Version:

AI SDK by Vercel - build apps like ChatGPT, Claude, Gemini, and more with a single interface for any model using the Vercel AI Gateway or go direct to OpenAI, Anthropic, Google, or any other model provider.

302 lines (276 loc) • 8.31 kB
--- title: experimental_MCPAppRenderer description: API reference for rendering MCP Apps with @ai-sdk/react. --- # `experimental_MCPAppRenderer` <Note type="warning"> `experimental_MCPAppRenderer` is experimental and may change in a future release. </Note> `experimental_MCPAppRenderer` renders an MCP App for an AI SDK tool UI part. It detects MCP App metadata on the tool part, loads the app resource, renders the app in a sandbox proxy iframe, and bridges MCP Apps JSON-RPC messages between the iframe and your host application. For tool parts without MCP App metadata, the component renders the `fallback`. ## Import <Snippet text={`import { experimental_MCPAppRenderer as MCPAppRenderer } from "@ai-sdk/react"`} prompt={false} /> ## Example ```tsx 'use client'; import { experimental_MCPAppRenderer as MCPAppRenderer, type MCPAppBridgeHandlers, type MCPAppMetadata, type MCPAppResource, type MCPAppSandboxConfig, } from '@ai-sdk/react'; import { isToolUIPart } from 'ai'; const sandbox = { url: '/mcp-app-sandbox', className: 'h-80 w-full rounded-lg border', style: { border: 0 }, } satisfies MCPAppSandboxConfig; async function loadResource(app: MCPAppMetadata): Promise<MCPAppResource> { const response = await fetch('/api/mcp-app-host/read-resource', { method: 'POST', body: JSON.stringify({ uri: app.resourceUri }), }); if (!response.ok) { throw new Error('Failed to load MCP App resource'); } return response.json(); } const handlers: MCPAppBridgeHandlers = { callTool: params => fetch('/api/mcp-app-host/call-tool', { method: 'POST', body: JSON.stringify(params), }).then(response => response.json()), openLink: ({ url }) => { window.open(url, '_blank', 'noopener,noreferrer'); return {}; }, }; export function MessagePart({ part }: { part: unknown }) { if (!isToolUIPart(part)) { return null; } return ( <MCPAppRenderer part={part} loadResource={loadResource} handlers={handlers} sandbox={sandbox} fallback={null} /> ); } ``` ## Props <PropertiesTable content={[ { name: 'part', type: 'ToolUIPart<UITools> | DynamicToolUIPart', description: 'The AI SDK tool UI part. The renderer looks for MCP App metadata in `part.toolMetadata.mcp.app`.', }, { name: 'sandbox', type: 'MCPAppSandboxConfig', description: 'Configuration for the outer sandbox proxy iframe used to host the app.', }, { name: 'resource', type: 'MCPAppResource', isOptional: true, description: 'A preloaded MCP App resource. Provide either `resource` or `loadResource`.', }, { name: 'loadResource', type: '(app: MCPAppMetadata) => Promise<MCPAppResource>', isOptional: true, description: 'Loads the MCP App resource for the app metadata found on the tool part.', }, { name: 'handlers', type: 'MCPAppBridgeHandlers', isOptional: true, description: 'Callbacks used to handle iframe requests such as `tools/call`, `resources/read`, `ui/open-link`, and display mode changes.', }, { name: 'hostInfo', type: '{ name: string; version: string }', isOptional: true, description: 'Host identity returned to the app during the `ui/initialize` handshake.', }, { name: 'hostContext', type: 'MCPAppHostContext', isOptional: true, description: 'Host context sent to the app, such as theme, display mode, and available display modes.', }, { name: 'fallback', type: 'ReactNode', isOptional: true, description: 'Rendered while the resource is loading, when loading fails, or when the tool part is not an MCP App.', }, ]} /> ## Sandbox Config <PropertiesTable content={[ { name: 'url', type: 'string | URL', description: 'The URL of the sandbox proxy iframe. The proxy receives app HTML from the host and creates the inner app iframe.', }, { name: 'title', type: 'string', isOptional: true, description: 'Accessible iframe title.', }, { name: 'className', type: 'string', isOptional: true, description: 'Class name applied to the outer iframe.', }, { name: 'style', type: 'CSSProperties', isOptional: true, description: 'Inline styles applied to the outer iframe.', }, { name: 'targetOrigin', type: 'string', isOptional: true, description: 'Target origin used for `postMessage`. Defaults to `*`; set this to your sandbox origin in production.', }, { name: 'outerSandbox', type: 'string', isOptional: true, description: 'Sandbox attribute for the outer proxy iframe. Defaults to `allow-scripts allow-same-origin allow-forms`.', }, { name: 'innerSandbox', type: 'string', isOptional: true, description: 'Sandbox attribute sent to the proxy for the inner app iframe. Defaults to `allow-scripts allow-forms`.', }, ]} /> ## Bridge Handlers `experimental_MCPAppRenderer` uses these handlers to respond to iframe requests. In production, server-backed handlers should validate authorization and MCP Apps tool visibility before calling the MCP server. <PropertiesTable content={[ { name: 'allowedTools', type: 'string[]', isOptional: true, description: 'Optional client-side allowlist checked before forwarding `tools/call` requests.', }, { name: 'callTool', type: '(params: MCPAppToolCallParams) => Promise<unknown> | unknown', isOptional: true, description: 'Handles app-initiated `tools/call` requests.', }, { name: 'readResource', type: '(params: { uri: string }) => Promise<unknown> | unknown', isOptional: true, description: 'Handles app-initiated `resources/read` requests.', }, { name: 'listResources', type: '(params?: unknown) => Promise<unknown> | unknown', isOptional: true, description: 'Handles app-initiated `resources/list` requests.', }, { name: 'openLink', type: '(params: { url: string }) => Promise<unknown> | unknown', isOptional: true, description: 'Handles app-initiated `ui/open-link` requests.', }, { name: 'sendMessage', type: '(params: unknown) => Promise<unknown> | unknown', isOptional: true, description: 'Handles app-initiated `ui/message` requests.', }, { name: 'updateModelContext', type: '(params: unknown) => Promise<unknown> | unknown', isOptional: true, description: 'Handles app-initiated `ui/update-model-context` requests.', }, { name: 'requestDisplayMode', type: "(params: { mode: 'inline' | 'fullscreen' | 'pip' }) => Promise<{ mode: MCPAppDisplayMode }> | { mode: MCPAppDisplayMode }", isOptional: true, description: 'Handles app-initiated `ui/request-display-mode` requests.', }, { name: 'onSizeChange', type: '(params: { width?: number; height?: number }) => void', isOptional: true, description: 'Called when the app sends a size change notification.', }, { name: 'onInitialized', type: '() => void', isOptional: true, description: 'Called after the app sends `ui/notifications/initialized`.', }, { name: 'onRequestTeardown', type: '(params: unknown) => void', isOptional: true, description: 'Called when the app requests teardown.', }, { name: 'onLog', type: '(params: unknown) => void', isOptional: true, description: 'Called when the app sends a log notification.', }, { name: 'onError', type: '(error: Error) => void', isOptional: true, description: 'Called when a supported iframe request fails while being handled.', }, ]} /> ## See Also <ExampleLinks examples={[ { title: 'MCP Apps guide', link: '/docs/ai-sdk-core/mcp-apps', }, { title: 'MCP Apps helpers', link: '/docs/reference/ai-sdk-core/mcp-apps', }, ]} />