TanStack
AI
UI

React Chat UI

createChatHook from @tanstack/ai-react/ui binds your chat options and components once. Tool calls and approvals stay typed.

Each screen then renders the chat with one hook call. You supply every visible component. There is no default markup or style.

Note

Upgrading from @tanstack/ai-react-ui? Read Upgrade to 1.0.

Quick setup

  1. Install the packages:
shell
npm i @tanstack/ai @tanstack/ai-react @tanstack/ai-openai
  1. Put OPENAI_API_KEY in your server environment. Then add a POST route at /api/chat:
ts
import {
  chat,
  chatParamsFromRequest,
  toServerSentEventsResponse,
} from '@tanstack/ai'
import { openaiText } from '@tanstack/ai-openai'

export async function POST(request: Request) {
  const { messages, threadId, runId } = await chatParamsFromRequest(request)

  const stream = chat({
    adapter: openaiText('gpt-6.1-sol'),
    messages,
    threadId,
    runId,
  })

  return toServerSentEventsResponse(stream)
}
  1. Create src/chat.tsx:
tsx
import { fetchServerSentEvents } from '@tanstack/ai-react'
import { createChatHook } from '@tanstack/ai-react/ui'

const chatOptions = {
  connection: fetchServerSentEvents('/api/chat'),
}

const { useAppChat } = createChatHook({
  options: chatOptions,
  components: {
    layout: ({ Messages }) => (
      <section>
        <Messages />
      </section>
    ),
    message: ({ message, Parts }) => (
      <article data-role={message.role}>
        <Parts />
      </article>
    ),
  },
  partsComponents: {
    text: ({ part }) => <p>{part.content}</p>,
    fallback: () => null,
  },
})

export function ChatScreen() {
  const chat = useAppChat({ threadId: 'support-chat' })

  return (
    <main>
      <chat.AppChat />
      <form
        onSubmit={(event) => {
          event.preventDefault()
          const field = event.currentTarget.elements.namedItem('message')
          if (!(field instanceof HTMLInputElement)) return
          const text = field.value.trim()
          if (!text) return
          field.value = ''
          void chat.sendMessage(text)
        }}
      >
        <input name="message" />
        <button type="submit">Send</button>
      </form>
    </main>
  )
}
  1. Render <ChatScreen /> and send a message.

The reply streams into the page as the model writes it.

How it works

  • createChatHook runs once, at module scope. It binds chatOptions to your components.
  • useAppChat({ threadId }) creates one chat instance. chat.AppChat renders that instance with your components.
  • layout places the message list. message wraps one message. Parts picks a component for each part of that message.
  • partsComponents.text renders text. fallback catches every part type that you did not register.

Next steps

Each step adds one thing to the quick setup. The chat UI recipes have the full file for each one.

Render a tool

Add the tool to tools in chatOptions. Then register a component under the tool name in toolsComponents. part.input and part.output are typed from the tool schemas.

tsx
toolsComponents: {
  getWeather: ({ part }) => {
    if (part.state !== 'complete') return <p>Looking up weather</p>
    return (
      <p>
        {part.input?.city}: {part.output?.temperature}
      </p>
    )
  },
},

The factory needs one entry for every tool in chatOptions. Full example: Make a tool look right.

Ask before a tool runs

Set needsApproval: true on the tool. The tool component then gets an interrupt prop:

tsx
toolsComponents: {
  deleteFile: ({ part, interrupt }) =>
    interrupt?.status === 'pending' ? (
      <button onClick={() => interrupt.resolveInterrupt(true)}>
        Delete {part.input?.path}
      </button>
    ) : (
      <p>{part.input?.path}</p>
    ),
},

resolveInterrupt(true) approves. resolveInterrupt(false) denies. Full example: Ask before a tool runs.

Ask the user your own question

Define the question with defineInterrupt. Add it to interrupts in chatOptions. Register a component under its id in interruptsComponents.generic, and render Interrupts in the layout:

tsx
const choosePlan = defineInterrupt({
  id: 'choosePlan',
  payloadSchema: z.object({ title: z.string() }),
  responseSchema: z.string(),
})

// In createChatHook:
interruptsComponents: {
  generic: {
    choosePlan: ({ interrupt }) => (
      <button onClick={() => interrupt.resolveInterrupt('pro')}>
        {interrupt.payload?.title}
      </button>
    ),
    fallback: ({ interrupt }) => <p>{interrupt.reason}</p>,
  },
},

The factory needs one entry for every interrupt id in chatOptions. Full example: Ask the user your own question.

Read chat state inside a widget

Mapped components do not get chat as a prop. Call useChatContext() inside the widget. This also moves the composer into the layout: register it as input, then render Input.

tsx
import { fetchServerSentEvents } from '@tanstack/ai-react'
import { createChatHook } from '@tanstack/ai-react/ui'

const chatOptions = {
  connection: fetchServerSentEvents('/api/chat'),
}

const { useAppChat, useChatContext } = createChatHook({
  options: chatOptions,
  components: {
    input: () => {
      const chat = useChatContext()
      return (
        <form
          onSubmit={(event) => {
            event.preventDefault()
            const field = event.currentTarget.elements.namedItem('message')
            if (!(field instanceof HTMLInputElement)) return
            const text = field.value.trim()
            if (!text) return
            field.value = ''
            void chat.sendMessage(text)
          }}
        >
          <input name="message" disabled={chat.isLoading} />
          <button type="submit">Send</button>
        </form>
      )
    },
    layout: ({ Messages, Input }) => (
      <main>
        <Messages />
        <Input />
      </main>
    ),
    message: ({ Parts }) => (
      <article>
        <Parts />
      </article>
    ),
  },
  partsComponents: {
    text: ({ part }) => <p>{part.content}</p>,
    fallback: () => null,
  },
})

export function ChatScreen() {
  const chat = useAppChat({ threadId: 'support-chat' })
  return <chat.AppChat />
}

Call useChatContext() only inside chat.AppChat. A call outside that tree throws.

Render markdown

Model replies arrive as markdown. Register TextPart from @tanstack/ai-react/ui as the text part:

tsx
import { TextPart } from '@tanstack/ai-react/ui'

partsComponents: {
  text: ({ part }) => <TextPart content={part.content} />,
  fallback: () => null,
},

TextPart uses TanStack Markdown. It escapes raw HTML and removes executable URLs. To color fenced code, read Highlight markdown code.

Type a widget in its own file

Move a widget to its own file and type its props with ToolProps, PartProps, or InterruptProps. Pass typeof chatOptions and a name:

tsx
import { fetchServerSentEvents } from '@tanstack/ai-react'
import { createChatHook, type ToolProps } from '@tanstack/ai-react/ui'
import { toolDefinition } from '@tanstack/ai'
import { z } from 'zod'

const getWeather = toolDefinition({
  name: 'getWeather',
  description: 'Look up weather',
  inputSchema: z.object({ city: z.string() }),
  outputSchema: z.object({ temperature: z.number() }),
}).client()

const chatOptions = {
  connection: fetchServerSentEvents('/api/chat'),
  tools: [getWeather],
}

export function WeatherTool({
  part,
}: ToolProps<typeof chatOptions, 'getWeather'>) {
  if (part.state !== 'complete') return <p>Looking up weather</p>
  return (
    <p>
      {part.input?.city}: {part.output?.temperature}
    </p>
  )
}

export const { useAppChat } = createChatHook({
  options: chatOptions,
  components: {
    layout: ({ Messages }) => <Messages />,
    message: ({ Parts }) => (
      <article>
        <Parts />
      </article>
    ),
  },
  partsComponents: { fallback: () => null },
  toolsComponents: { getWeather: WeatherTool },
})

Put chatOptions in a shared module so every widget file can import its type.

You now have a typed chat UI in React. Each widget is your own component.

Advanced

Every widget slot

Config keyKeysProps
componentslayout, message, input, queueLayoutProps, MessageProps, InputProps, QueueProps
partsComponentsa part key, or fallbackPartProps<typeof chatOptions, 'text'>
toolsComponentseach tool name in chatOptions.toolsToolProps<typeof chatOptions, 'getWeather'>
interruptsComponents.toolsan approval tool nameInterruptProps<typeof chatOptions, 'deleteFile'>
interruptsComponents.genericeach interrupt id, plus an optional fallbackInterruptProps<typeof chatOptions, 'choosePlan'>
subagentsComponentseach subagent nameSubagentProps. See Subagents

Part keys:

  • text, thinking, activity
  • image, audio, video, document
  • toolCall, toolResult
  • structuredOutput, uiResource, subagent

layout gets these components. Render them as tags, such as <Messages />:

  • Messages, Interrupts, Queue, Subagents
  • Input, only when you register input. Without it, <Input /> is a compile error.

useAppChat() takes per-instance options: threadId (required), initialMessages, live, and forwardedProps. Keep tools, interrupts, and outputSchema on chatOptions. Two instances with different threadId values keep separate histories:

tsx
const chat = useAppChat({ threadId: 'support-1' })

Tool part states

part.stateMeaning
awaiting-inputThe call started. No arguments yet.
input-streamingThe model is still writing the arguments.
input-completeThe arguments are final.
approval-requestedThe tool waits for approval.
approval-respondedThe user approved or denied.
completeThe result arrived.
errorThe call failed.

A tool component also gets result, the matching tool-result part, when one exists.

Approval next to the tool, or in the list

A tool with needsApproval: true renders its approval in one of two places:

  • Next to the tool. Read interrupt in the tool component. This is the default.
  • In the list. Register a component under interruptsComponents.tools with the tool name. Render Interrupts in the layout. The approval then leaves the message.

Do not do both for one tool. A list approval component reads interrupt.originalArgs, typed from the tool inputSchema.

Generic interrupts and fallback

Generic interrupts always render in Interrupts. They never render inside a tool. interruptsComponents.generic.fallback renders:

  • an interrupt id that you did not register.
  • an unbound interrupt. Something outside this chat owns it, so it cannot be resolved here.

If the copy for an unbound interrupt must be different, branch on interrupt.kind === 'unbound'.

Show queued messages

By default, a message sent while a run streams waits in a queue. Register queue on components and render Queue in the layout. Each item has cancelQueued() bound to it:

tsx
components: {
  layout: ({ Messages, Queue }) => (
    <main>
      <Messages />
      <Queue />
    </main>
  ),
  message: ({ Parts }) => <Parts />,
  queue: ({ item }) => (
    <div>
      {typeof item.content === 'string' ? item.content : 'Queued'}
      <button type="button" onClick={() => item.cancelQueued()}>
        Cancel
      </button>
    </div>
  ),
},

Type a queue item with QueueProps<typeof chatOptions>.

List input before layout

If input is an inline function expression written after layout, TypeScript cannot see that an input is registered. Then Input is missing from the layout props.

To prevent this, do one of these:

  • Put input before layout.
  • Use an arrow function or a named component for input.

If it still slips through, <Input /> without a registered input warns once in development.

Widgets in other files

A widget file that imports useChatContext from the factory module makes a circular import. To prevent this, call createChatHookContexts() in its own module first:

tsx
// chat-contexts.ts
import { createChatHookContexts } from '@tanstack/ai-react/ui'

export const { chatContext, partContext, interruptContext, useChatContext } =
  createChatHookContexts()

Pass chatContext, partContext, and interruptContext to createChatHook under context. Widget files then import useChatContext from the contexts module.

Part and interrupt widgets get part and interrupt as props. Do not read those from context.

Walk messages yourself

createChatUI is the lower-level kit. If you map messages yourself, use it. It takes the same component maps:

tsx
import { fetchServerSentEvents, useChat } from '@tanstack/ai-react'
import { createChatUI } from '@tanstack/ai-react/ui'
import { toolDefinition } from '@tanstack/ai'
import { z } from 'zod'

const getWeather = toolDefinition({
  name: 'getWeather',
  description: 'Look up weather',
  inputSchema: z.object({ city: z.string() }),
  outputSchema: z.object({ temperature: z.number() }),
}).client()

const chatOptions = {
  connection: fetchServerSentEvents('/api/chat'),
  tools: [getWeather],
}

const UI = createChatUI(chatOptions, {
  components: {
    layout: ({ Messages }) => <Messages />,
    message: ({ Parts }) => <article><Parts /></article>,
  },
  partsComponents: {
    text: ({ part }) => <p>{part.content}</p>,
    fallback: () => null,
  },
  toolsComponents: {
    getWeather: ({ part }) => <p>{part.input?.city}</p>,
  },
})

export function ManualChat() {
  const chat = useChat({ ...chatOptions, threadId: 'support-chat' })
  return (
    <UI.Provider chat={chat}>
      <UI.Messages>
        {(messages) =>
          messages.map((message) => (
            <UI.Message key={message.id} message={message}>
              {(parts) =>
                parts.map((part, index) => (
                  <span key={index}>{part.key}</span>
                ))
              }
            </UI.Message>
          ))
        }
      </UI.Messages>
    </UI.Provider>
  )
}

You can also pick a registered widget at the call site:

tsx
import type { MessageProps } from '@tanstack/ai-react/ui'

function PickedWidgets({
  message,
}: {
  message: MessageProps<typeof chatOptions>['message']
}) {
  return (
    <UI.Message message={message}>
      {(parts) =>
        parts.map((part, index) => (
          <UI.Part key={index} part={part}>
            {(p) => (part.key === 'toolCall' ? <p.getWeather /> : <p.Render />)}
          </UI.Part>
        ))
      }
    </UI.Message>
  )
}
  • p.getWeather and p.text are the widgets that you passed to the factory.
  • p.Render renders this one part the automatic way.
  • UI.Input exists when you register input.

An unknown tool name at runtime warns once in development and renders nothing.

TextPart options

  • highlighter: a synchronous CodeHighlighter. Its output is trusted HTML, so use createTanStackMarkdownHighlighter or another highlighter that escapes source text.
  • extensions: more TanStack Markdown extensions. The streaming extension is always on.
  • components: a tag-name map with normal HTML props (href, className). inline and node are not passed.
  • className, userClassName, assistantClassName: classes for the wrapper. Pass role to apply the role class.
tsx
import { TextPart } from '@tanstack/ai-react/ui'

export function Reply({ content }: { content: string }) {
  return (
    <TextPart
      content={content}
      components={{
        a: (props) => <a {...props} target="_blank" rel="noreferrer" />,
      }}
    />
  )
}

If you still pass remarkPlugins, rehypePlugins, or disableDefaultPlugins, read TextPart markdown.

Edit on GitHub
Compaction: Advanced
Solid