AI Core/chat Experience

Chat Experience

This skill builds on ai-core. Read it first for critical rules.

Setup — Minimal Chat App

Server: API Route (TanStack Start)

// src/routes/api.chat.ts
import { createFileRoute } from '@tanstack/react-router'
import { chat, toServerSentEventsResponse } from '@tanstack/ai'
import { openaiText } from '@tanstack/ai-openai'

export const Route = createFileRoute('/api/chat')({
  server: {
    handlers: {
      POST: async ({ request }) => {
        const abortController = new AbortController()
        const body = await request.json()
        const { messages } = body

        const stream = chat({
          adapter: openaiText('gpt-5.5'),
          messages,
          systemPrompts: ['You are a helpful assistant.'],
          abortController,
        })

        return toServerSentEventsResponse(stream, { abortController })
      },
    },
  },
})

Client: React Component

// src/routes/index.tsx
import { useState } from 'react'
import { useChat, fetchServerSentEvents } from '@tanstack/ai-react'
import type { UIMessage } from '@tanstack/ai-react'

function ChatPage() {
  const [input, setInput] = useState('')

  const { messages, sendMessage, isLoading, error, stop } = useChat({
    connection: fetchServerSentEvents('/api/chat'),
  })

  const handleSubmit = () => {
    if (!input.trim()) return
    sendMessage(input.trim())
    setInput('')
  }

  return (
    <div>
      <div>
        {messages.map((message: UIMessage) => (
          <div key={message.id}>
            <strong>{message.role}:</strong>
            {message.parts.map((part, i) => {
              if (part.type === 'text') {
                return <p key={i}>{part.content}</p>
              }
              return null
            })}
          </div>
        ))}
      </div>

      {error && <div>Error: {error.message}</div>}

      <div>
        <input
          value={input}
          => setInput(e.target.value)}
          => {
            if (e.key === 'Enter' && !e.shiftKey) {
              e.preventDefault()
              handleSubmit()
            }
          }}
          disabled={isLoading}
          placeholder="Type a message..."
        />
        {isLoading ? (
          <button
        ) : (
          <button disabled={!input.trim()}>
            Send
          </button>
        )}
      </div>
    </div>
  )
}

Vue/Solid/Svelte/Preact have identical patterns with different hook imports (e.g., import { useChat } from '@tanstack/ai-solid').

Core Patterns

1. Streaming Chat with SSE

Server returns a streaming SSE Response; client parses it automatically.

Server:

import { chat, toServerSentEventsResponse } from '@tanstack/ai'
import { anthropicText } from '@tanstack/ai-anthropic'

export async function POST(request: Request) {
  const { messages } = await request.json()
  const abortController = new AbortController()

  const stream = chat({
    adapter: anthropicText('claude-opus-5'),
    messages,
    modelOptions: {
      temperature: 0.7,
      max_tokens: 2000, // Anthropic-native key
    },
    systemPrompts: ['You are a helpful assistant.'],
    abortController,
  })

  return toServerSentEventsResponse(stream, { abortController })
}

To make the SSE response resumable (reconnect after a drop/refresh without re-running the provider), pass a delivery-durability adapter: toServerSentEventsResponse(stream, { durability: { adapter: memoryStream(request) } }) (memoryStream from @tanstack/ai is process-local, for dev/tests) or durableStream(request, { server }) from @tanstack/ai-durable-stream (Durable Streams protocol, production). Each SSE event gets an opaque adapter-owned id:; fetchServerSentEvents auto-reconnects with Last-Event-ID and exposes joinRun(runId) to replay a run from the start. See docs/resumable-streams/overview.md.

Client:

import { useChat, fetchServerSentEvents } from '@tanstack/ai-react'

const { messages, sendMessage, isLoading, error, stop, status } = useChat({
  connection: fetchServerSentEvents('/api/chat'),
  body: { provider: 'anthropic', model: 'claude-opus-5' },
  onFinish: (message) => {
    console.log('Response complete:', message.id)
  },
  onError: (err) => {
    console.error('Stream error:', err)
  },
})

The body field is merged into the POST request body alongside messages, letting the server read data.provider, data.model, etc.

The status field tracks the chat lifecycle: 'ready' | 'submitted' | 'streaming' | 'error'.

2. Rendering Thinking/Reasoning Content

Models with extended thinking (Claude, Gemini) emit ThinkingPart in the message parts array.

import type { UIMessage } from '@tanstack/ai-react'

function MessageRenderer({ message }: { message: UIMessage }) {
  return (
    <div>
      {message.parts.map((part, i) => {
        if (part.type === 'thinking') {
          const isComplete = message.parts
            .slice(i + 1)
            .some((p) => p.type === 'text')
          return (
            <details key={i} open={!isComplete}>
              <summary>
                {isComplete ? 'Thought process' : 'Thinking...'}
              </summary>
              <pre>{part.content}</pre>
            </details>
          )
        }

        if (part.type === 'text' && part.content) {
          return <p key={i}>{part.content}</p>
        }

        if (part.type === 'tool-call') {
          return (
            <div key={part.id}>
              Tool call: {part.name} ({part.state})
            </div>
          )
        }

        return null
      })}
    </div>
  )
}

Server-side, enable thinking via modelOptions on the adapter:

import { chat, toServerSentEventsResponse } from '@tanstack/ai'
import { geminiText } from '@tanstack/ai-gemini'

export async function POST(request: Request) {
  const { messages } = await request.json()

  const stream = chat({
    adapter: geminiText('gemini-3.8-flash'),
    messages,
    modelOptions: {
      thinkingConfig: {
        includeThoughts: true,
        thinkingLevel: 'HIGH', // Gemini 3.x; Gemini 2.x uses thinkingBudget
      },
    },
  })

  return toServerSentEventsResponse(stream)
}

3. Sending Multimodal Content (Images)

Use sendMessage with a MultimodalContent object instead of a plain string.

import { useChat, fetchServerSentEvents } from '@tanstack/ai-react'
import type { ContentPart } from '@tanstack/ai'

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

function sendImageMessage(text: string, imageBase64: string, mimeType: string) {
  const contentParts: Array<ContentPart> = [
    { type: 'text', content: text },
    {
      type: 'image',
      source: { type: 'data', value: imageBase64, mimeType },
    },
  ]

  sendMessage({ content: contentParts })
}

function sendImageUrl(text: string, imageUrl: string) {
  const contentParts: Array<ContentPart> = [
    { type: 'text', content: text },
    {
      type: 'image',
      source: { type: 'url', value: imageUrl },
    },
  ]

  sendMessage({ content: contentParts })
}

Render image parts in received messages:

import type { UIMessage } from '@tanstack/ai-react'

function ImagePart({ part }: { part: UIMessage['parts'][number] }) {
  if (part.type !== 'image') return null
  const src =
    part.source.type === 'url'
      ? part.source.value
      : `data:${part.source.mimeType};base64,${part.source.value}`
  return <img src={src} alt="Attached image" />
}

4. Sending Audio Messages (Browser Recording)

Use useAudioRecorder from @tanstack/ai-react (or createAudioRecorder in Svelte) to capture audio in the browser. The resolved AudioRecording includes a ready-to-use part that slots directly into sendMessage.

import {
  useAudioRecorder,
  useChat,
  fetchServerSentEvents,
} from '@tanstack/ai-react'

const { isRecording, isSupported, start, stop } = useAudioRecorder()
const { sendMessage } = useChat({
  connection: fetchServerSentEvents('/api/chat'),
})

async function toggle() {
  if (!isRecording) {
    await start()
    return
  }
  const recording = await stop()
  await sendMessage({ content: [recording.part] })
}

recording.part is { type: 'audio', source: { type: 'data', value: base64, mimeType } }. Returns the recorder's native format (audio/webm or audio/mp4) with no transcoding.

5. HTTP Stream Format (Alternative to SSE)

Use toHttpResponse + fetchHttpStream for newline-delimited JSON instead of SSE.

Server:

import { chat, toHttpResponse } from '@tanstack/ai'
import { openaiText } from '@tanstack/ai-openai'

export async function POST(request: Request) {
  const { messages } = await request.json()
  const abortController = new AbortController()

  const stream = chat({
    adapter: openaiText('gpt-5.6'),
    messages,
    abortController,
  })

  return toHttpResponse(stream, { abortController })
}

Client:

import { useChat, fetchHttpStream } from '@tanstack/ai-react'

const { messages, sendMessage } = useChat({
  connection: fetchHttpStream('/api/chat'),
})

The only difference is swapping toServerSentEventsResponse / fetchServerSentEvents for toHttpResponse / fetchHttpStream. Everything else stays identical.

This includes resumability: pass the same durability adapter to toHttpResponse(stream, { durability: { adapter: memoryStream(request) } }) and each NDJSON line becomes an { id, chunk } envelope. fetchHttpStream auto-reconnects with Last-Event-ID, de-dupes the replayed prefix, and exposes joinRun(runId) — the same guarantees as resumable SSE. The XHR adapters (xhrServerSentEvents / xhrHttpStream) are resumable too.

6. MCP Tool Discovery via chat({ mcp })

Pass mcp to let chat() own discovery and lifecycle for one or more MCP clients. Useful when you want minimal boilerplate and don't need to reuse the clients across calls.

// Prop shape:
// chat({
//   ...,
//   mcp: {
//     clients: Array<MCPClient | MCPClients>,
//     connection?: 'close' | 'keep-alive',  // default: 'close'
//     lazyTools?: boolean,
//     onDiscoveryError?: (error: unknown, source) => void,
//   }
// })

• clients — one or more MCPClient / MCPClients instances.
• connection — 'close' (default) closes each client when the run ends (after the agent loop completes and the stream is drained); with 'keep-alive', chat() never closes the clients — the caller owns their lifecycle (keep connections warm across requests).
• lazyTools — forwarded to tools({ lazy: true }) so tool schemas are sent to the LLM on demand.
• onDiscoveryError — throw (or re-throw) to fail the entire call fast; return normally to skip that source and continue. Omit to rethrow (fail-fast).

When to use mcp vs. the tools spread:

Server-side example:

import { chat, toServerSentEventsResponse } from '@tanstack/ai'
import { openaiText } from '@tanstack/ai-openai'
import { createMCPClient } from '@tanstack/ai-mcp'

export async function POST(request: Request) {
  const { messages } = await request.json()

  const mcpClient = await createMCPClient({
    transport: { type: 'http', url: 'https://mcp.example.com/mcp' },
  })

  const stream = chat({
    adapter: openaiText('gpt-5.6'),
    messages,
    mcp: {
      clients: [mcpClient],
      connection: 'keep-alive', // chat() won't close it — reuse across requests
    },
  })

  return toServerSentEventsResponse(stream)
  // connection: 'keep-alive' — chat() never closes mcpClient; it stays open for reuse across runs.
}

7. Queueing Messages Sent While Streaming

By default, a sendMessage call that arrives while a stream is in flight is queued and sent automatically once the run settles successfully — this is a behavior change: such sends used to be silently dropped. Configure it with the queue option on useChat:

import { useChat, fetchServerSentEvents } from '@tanstack/ai-react'

const { messages, queue, sendMessage, cancelQueued, isLoading } = useChat({
  connection: fetchServerSentEvents('/api/chat'),
  queue: { whenBusy: 'queue', drain: 'fifo', maxSize: 5, onOverflow: 'reject' },
})

• whenBusy — 'queue' (default) holds the message until a successful settle; 'drop' ignores the send (never appears in queue/messages); 'interrupt' aborts the current stream and sends immediately (unlike stop(), does not flush already-queued items — they drain after the interrupting send succeeds).
• drain — 'fifo' (default) sends queued items one at a time in order; 'batch' merges everything queued into a single send once the run settles successfully.
• maxSize / onOverflow — cap the queue length; 'reject' (default) silently ignores overflow sends (does not throw), 'drop-oldest' evicts the oldest queued item to make room.

The top-level queue option also accepts a plain WhenBusy string shorthand (e.g. queue: 'interrupt') or a QueueStrategy function for per-send action control. Strategy form always drains FIFO. Actions use the WhenBusy type.

Drain vs flush: queued messages auto-send only after a successful settle. They are discarded on stream error/abort of the active generation, stop(), clear(), unsubscribe(), and reload(). interrupt does not flush.

queue: Array<QueuedMessage> ({ id, content, createdAt }) is separate from messages — render pending sends distinctly and cancel with cancelQueued(id):

import { useChat, fetchServerSentEvents } from '@tanstack/ai-react'

function QueuedMessages() {
  const { queue, cancelQueued } = useChat({
    connection: fetchServerSentEvents('/api/chat'),
  })

  return (
    <div>
      {queue.map((q) => (
        <div key={q.id}>
          {typeof q.content === 'string' ? q.content : '[attachment]'}
          <button => cancelQueued(q.id)}>Cancel</button>
        </div>
      ))}
    </div>
  )
}

Override the configured policy for a single send with the second argument to sendMessage:

import { useChat, fetchServerSentEvents } from '@tanstack/ai-react'

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

sendMessage('Never mind, do this instead', { whenBusy: 'interrupt' })

8. Browser-Refresh Durability (client persistence)

By default a ChatClient / useChat keeps messages in memory only, so a full page reload loses the conversation. The optional persistence option (a ChatClientPersistence adapter) fixes this from the client side: it stores one combined record — { messages, resume? } (ChatPersistedState) — per chat id, so a reload restores the transcript and rehydrates any pending interrupt / rejoins a run that was still streaming. No manual initialMessages + onFinish boilerplate.

Three storage adapters ship from @tanstack/ai-client: localStoragePersistence (survives reloads and browser restarts), sessionStoragePersistence (scoped to the tab), and indexedDBPersistence (async, structured-clone storage — no codec needed for Date/Map/etc.). Give the chat a stable threadId so the reload finds the same record. Persistence keys on threadId; the storage adapters are re-exported from each framework package, so a single import works:

import {
  useChat,
  fetchServerSentEvents,
  localStoragePersistence,
} from '@tanstack/ai-react'

// Defaults to the ChatPersistedState shape and a JSON codec, so no type
// argument or serialize/deserialize is needed. indexedDBPersistence stores via
// structured clone (a Date round-trips exactly).
const persistence = localStoragePersistence()

function Chat() {
  const { messages, sendMessage } = useChat({
    threadId: 'support-chat',
    connection: fetchServerSentEvents('/api/chat'),
    persistence,
  })
  // ...render messages, call sendMessage(text)
}

Keep large transcripts off the client. persistence also accepts true (server-authoritative): the client caches nothing, and on mount it hydrates the thread from the server by threadId (transcript plus a cursor to any run still generating). An adapter is client-authoritative; true leaves history on the server and needs a connection with a hydrate handler plus a server GET endpoint (reconstructChat), since the delivery log only holds one run.

Mid-stream reload rejoin. If the run was still streaming when the page reloaded, the client re-attaches instead of showing a frozen half-reply — but only when the connection is resumable: a delivery-durability-backed route that records the stream and exposes a GET replay handler (see docs/resumable-streams/overview.md and Pattern 1's durability adapter). Given that, useChat finds the persisted in-flight run on load and auto-rejoins it via joinRun, replaying from the server's log so the reply finishes where it left off. No extra client code beyond the resumable connection.

Every framework, no extra code. Durability rides the existing persistence option, so it works identically in @tanstack/ai-react, -solid, -vue, -svelte, -angular, and -preact — pass persistence (and a stable threadId, which is the chat's identity) to the framework's useChat / createChat / injectChat; nothing is framework-specific.

Client vs. server durability. This is the client (per-browser) half. The authoritative, multi-user, server-side copy is the withPersistence middleware — see ai-core/middleware/SKILL.md. The two are independent; use both for instant reload restore plus a durable record of record.

Common Mistakes

a. CRITICAL: Using Vercel AI SDK patterns (streamText, generateText)

// WRONG
import { streamText } from 'ai'
import { openai } from '@ai-sdk/openai'

const messages = [{ role: 'user' as const, content: 'Hello' }]
const result = streamText({ model: openai('gpt-5.6'), messages })

// CORRECT
import { chat } from '@tanstack/ai'
import { openaiText } from '@tanstack/ai-openai'

const messages = [{ role: 'user' as const, content: 'Hello' }]
const stream = chat({ adapter: openaiText('gpt-5.6'), messages })

b. CRITICAL: Using Vercel createOpenAI() provider pattern

// WRONG
import { streamText } from 'ai'
import { createOpenAI } from '@ai-sdk/openai'

const messages = [{ role: 'user' as const, content: 'Hello' }]
const openai = createOpenAI({ apiKey: process.env.OPENAI_API_KEY })
streamText({ model: openai('gpt-5.6'), messages })

// CORRECT
import { openaiText } from '@tanstack/ai-openai'
import { chat } from '@tanstack/ai'

const messages = [{ role: 'user' as const, content: 'Hello' }]
chat({ adapter: openaiText('gpt-5.6'), messages })

c. CRITICAL: Using monolithic openai() instead of openaiText()

// WRONG — `openai()` is no longer exported from @tanstack/ai-openai
import { openai } from '@tanstack/ai-openai'
chat({ adapter: openai(), model: 'gpt-5.6', messages })

// CORRECT
import { chat } from '@tanstack/ai'
import { openaiText } from '@tanstack/ai-openai'

const messages = [{ role: 'user' as const, content: 'Hello' }]
chat({ adapter: openaiText('gpt-5.6'), messages })

The monolithic openai() adapter no longer exists. Use tree-shakeable adapters: openaiText(), openaiImage(), openaiSpeech(), etc.

d. HIGH: Using toResponseStream instead of toServerSentEventsResponse

// WRONG — toResponseStream does not exist
import { toResponseStream } from '@tanstack/ai'
return toResponseStream(stream, { abortController })

// CORRECT
import { chat, toServerSentEventsResponse } from '@tanstack/ai'
import { openaiText } from '@tanstack/ai-openai'

export async function POST(request: Request) {
  const { messages } = await request.json()
  const abortController = new AbortController()
  const stream = chat({
    adapter: openaiText('gpt-5.6'),
    messages,
    abortController,
  })
  return toServerSentEventsResponse(stream, { abortController })
}

e. HIGH: Passing model as separate parameter to chat()

// WRONG
chat({ adapter: openaiText(), model: 'gpt-5.6', messages })

// CORRECT
import { chat } from '@tanstack/ai'
import { openaiText } from '@tanstack/ai-openai'

const messages = [{ role: 'user' as const, content: 'Hello' }]
chat({ adapter: openaiText('gpt-5.6'), messages })

The model is passed to the adapter factory, not to chat().

f. HIGH: Passing sampling options at the root of chat()

Sampling options (temperature, token limits, top_p/topP) are not top-level fields on chat(). They live inside modelOptions using the provider's native key.

// WRONG — temperature/maxTokens are not root options
chat({ adapter, messages, temperature: 0.7, maxTokens: 1000 })

// WRONG — there is no `options` field either
chat({ adapter, messages, options: { temperature: 0.7, maxTokens: 1000 } })

// CORRECT — inside modelOptions, provider-native keys (OpenAI shown)
import { chat } from '@tanstack/ai'
import { openaiText } from '@tanstack/ai-openai'

const messages = [{ role: 'user' as const, content: 'Hello' }]

chat({
  adapter: openaiText('gpt-5.6'),
  messages,
  modelOptions: { temperature: 0.7, max_output_tokens: 1000 },
})

temperature works on most models (Claude 5 models reject sampling parameters; see ai-core/adapter-configuration/SKILL.md). Token limits use provider-native keys (max_output_tokens for OpenAI, max_tokens for Anthropic/Grok, maxOutputTokens for Gemini, max_completion_tokens for Groq, maxCompletionTokens for OpenRouter, and num_predict nested under modelOptions.options for Ollama). See ai-core/adapter-configuration/SKILL.md.

g. HIGH: Using providerOptions instead of modelOptions

// WRONG
chat({
  adapter,
  messages,
  providerOptions: { text: { format: { type: 'json_object' } } },
})

// CORRECT — provider-native option under modelOptions (OpenAI Responses shown)
import { chat } from '@tanstack/ai'
import { openaiText } from '@tanstack/ai-openai'

const messages = [{ role: 'user' as const, content: 'Hello' }]

chat({
  adapter: openaiText('gpt-5.6'),
  messages,
  modelOptions: { text: { format: { type: 'json_object' } } },
})

h. HIGH: Implementing custom SSE stream instead of using toServerSentEventsResponse

// WRONG
import { chat } from '@tanstack/ai'
import { openaiText } from '@tanstack/ai-openai'

export async function POST(request: Request) {
  const { messages } = await request.json()
  const stream = chat({ adapter: openaiText('gpt-5.6'), messages })

  const readable = new ReadableStream({
    async start(controller) {
      const encoder = new TextEncoder()
      for await (const chunk of stream) {
        controller.enqueue(encoder.encode(`data: ${JSON.stringify(chunk)}\n\n`))
      }
      controller.enqueue(encoder.encode('data: [DONE]\n\n'))
      controller.close()
    },
  })
  return new Response(readable, {
    headers: { 'Content-Type': 'text/event-stream' },
  })
}

// CORRECT
import { chat, toServerSentEventsResponse } from '@tanstack/ai'
import { openaiText } from '@tanstack/ai-openai'

export async function POST(request: Request) {
  const { messages } = await request.json()
  const abortController = new AbortController()
  const stream = chat({
    adapter: openaiText('gpt-5.6'),
    messages,
    abortController,
  })
  return toServerSentEventsResponse(stream, { abortController })
}

toServerSentEventsResponse handles SSE formatting, abort signals, error events (RUN_ERROR), and correct headers automatically.

i. HIGH: Implementing custom onEnd/onFinish callbacks instead of middleware

// WRONG — chat() has no onEnd/onFinish option
chat({
  adapter,
  messages,
  onEnd: (result) => {
    trackAnalytics(result)
  },
})

// CORRECT
import { chat } from '@tanstack/ai'
import type { ChatMiddleware } from '@tanstack/ai'
import { openaiText } from '@tanstack/ai-openai'
import { trackAnalytics, trackTokens } from './analytics'

const analytics: ChatMiddleware = {
  name: 'analytics',
  onFinish(ctx, info) {
    trackAnalytics({ reason: info.finishReason, iterations: ctx.iteration })
  },
  onUsage(ctx, usage) {
    trackTokens(usage.totalTokens)
  },
}

const messages = [{ role: 'user' as const, content: 'Hello' }]
chat({ adapter: openaiText('gpt-5.6'), messages, middleware: [analytics] })

chat() has no onEnd/onFinish option. Use middleware for lifecycle events. See also: ai-core/middleware/SKILL.md.

j. HIGH: Importing from @tanstack/ai-client instead of framework package

// WRONG
import { fetchServerSentEvents } from '@tanstack/ai-client'
import { useChat } from '@tanstack/ai-react'

// CORRECT
import { useChat, fetchServerSentEvents } from '@tanstack/ai-react'

Framework packages re-export everything needed from @tanstack/ai-client. Import from @tanstack/ai-client only in vanilla JS (no framework).

k. MEDIUM: Not handling RUN_ERROR events in streaming context

Streaming errors arrive as RUN_ERROR events in the stream, not as thrown exceptions. The useChat hook surfaces these via the error state and onError callback. If you consume the stream manually (without useChat), check for RUN_ERROR chunks:

import { chat } from '@tanstack/ai'
import { openaiText } from '@tanstack/ai-openai'

const messages = [{ role: 'user' as const, content: 'Hello' }]
const stream = chat({ adapter: openaiText('gpt-5.6'), messages })

for await (const chunk of stream) {
  if (chunk.type === 'RUN_ERROR') {
    console.error('Stream error:', chunk.message)
    break
  }
  if (chunk.type === 'TEXT_MESSAGE_CONTENT') {
    process.stdout.write(chunk.delta)
  }
}

If not handled, the UI appears to hang with no feedback.

Cross-References

• See also: ai-core/tool-calling/SKILL.md -- Most chats include tools
• See also: ai-core/adapter-configuration/SKILL.md -- Adapter choice affects available features
• See also: ai-core/middleware/SKILL.md -- Use middleware for analytics and lifecycle events
• See also: @tanstack/ai-persistence skills (skills/ai-persistence/SKILL.md in that package) -- Server + client state persistence, store contracts, adapter recipes (deeper than Pattern 8)