Docs

Streaming & Results

Every adapter exposes the AI SDK agent surface: stream() for live turns and generate() for one-shots. The stream is a typed async iterable, so your UI can render exactly what is happening — text, tool calls, results, approval requests — as it happens.

const result = await agent.stream({ messages })   // streaming
const result = await agent.generate({ messages }) // one-shot

The result object#

Property	What it is
result.fullStream	Async iterable of typed chunks: `text-delta`, `tool-call`, `tool-result`, `tool-approval-request`, `error`.
result.response	Resolves to the final messages — append them to your history.
result.text	The final assistant text.
result.totalUsage	Token counts for the turn.

Consuming the stream#

for await (const chunk of result.fullStream) {
  if (chunk.type === 'text-delta') process.stdout.write(chunk.text)
  else if (chunk.type === 'tool-call') console.log(`\n↳ ${chunk.toolName}`, chunk.input)
  else if (chunk.type === 'tool-result') console.log('  done')
  else if (chunk.type === 'tool-approval-request') {
    // Pause point: respond and continue — see Tool Approval.
  } else if (chunk.type === 'error') {
    console.error(chunk.error)
  }
}

Keeping history#

The agent is stateless between turns by design — you own the conversation. Keep a messages array and push result.response.messages after each turn:

const messages = []

async function turn(text) {
  messages.push({ role: 'user', content: text })
  const result = await agent.stream({ messages })
  for await (const chunk of result.fullStream) { /* render */ }
  const response = await result.response
  messages.push(...response.messages)
}

Building a chat UI? The stream shape is AI SDK-native, so it plugs into useChat-style frontends with a thin transport — Sharables wires it exactly that way.