Add AI SDK compatibility via v0/ai-sdk and v0/react

[nandorojo]

Additive AI SDK v7 compatibility for the v0 SDK: two new entry points, v0/ai-sdk and v0/react, with optional peer deps on ai, @ai-sdk/react, and react. No changes to the existing API.

Server

// app/api/chat/route.ts
import { v0 } from 'v0'
import { toUIMessageStreamResponse } from 'v0/ai-sdk'

export async function POST(request: Request) {
  const { chatId, message, attachments } = await request.json()

  return toUIMessageStreamResponse(
    chatId
      ? await v0.messages.sendStream({ chatId, message, attachments })
      : await v0.chats.createStream({ message, attachments }),
  )
}

// stream resumption
export async function GET(request: Request) {
  const chatId = new URL(request.url).searchParams.get('chatId')
  if (!chatId) return new Response(null, { status: 204 })

  try {
    return toUIMessageStreamResponse(await v0.chats.resume({ chatId }))
  } catch {
    return new Response(null, { status: 204 })
  }
}

Client

'use client'

import { useV0Chat } from 'v0/react'

const { messages, sendMessage, status, chatId } = useV0Chat()

// or, with an existing chat + resumption of in-flight generations
useV0Chat({ chatId, messages, resume: true })

Prefer the plain hook? Same route handler:

import { useChat } from '@ai-sdk/react'
import type { V0UIMessage } from 'v0/ai-sdk'

useChat<V0UIMessage>() // just works

Fully typed messages, derived from the SDK

V0UIMessage derives entirely from the existing Message type. text/thinking map to the AI SDK's native text/reasoning parts; every other v0 part becomes a typed data-* part; the remaining Message fields are the metadata.

{message.parts.map((part, index) => {
  switch (part.type) {
    case 'text':
      return <p key={index}>{part.text}</p>
    case 'reasoning':
      return <em key={index}>{part.text}</em>
    case 'data-file-edit':
      // part.data.operation: 'create' | 'update' | 'delete' | 'rename' | 'patch'
      return <code key={index}>{`${part.data.operation} ${part.data.path}`}</code>
    case 'data-bash':
      return <pre key={index}>{`$ ${part.data.command}`}</pre>
    default:
      return null
  }
})}

message.metadata?.chatId
message.metadata?.usage?.creditsCost.total

Loading history

import { toUIMessages } from 'v0/ai-sdk'

const response = await v0.messages.list({ chatId, limit: 100 })
const messages = toUIMessages(response.data.messages) // pass to useV0Chat({ chatId, messages })

---

Also exports toUIMessageStream (raw chunk stream), toUIMessage, and fromUIMessage for custom setups. 55 tests cover the converters, the stream protocol adapter (driven through the SDK's real jsondiffpatch deltas), and useV0Chat rendered against a recorded fetch.

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated (UTC)
v0-sdk-clone-demo	Error	Open in v0	Jul 8, 2026 8:45pm
v0-sdk-playground	Error	Open in v0	Jul 8, 2026 8:45pm
v0-sdk-simple-demo	Error	Open in v0	Jul 8, 2026 8:45pm

[vercel]

Additional Suggestion:

Named function expression in useEffect shadows the memoized generateNextChatId, causing infinite recursion and a stack overflow on mount.

[socket-security]

Review the following changes in direct dependencies. Learn more about Socket for GitHub.

Diff	Package	Supply Chain Security	Vulnerability	Quality	Maintenance	License
	npm/​@​ai-sdk/​react@​4.0.16
	npm/​legid@​0.1.4
	npm/​ai@​7.0.15
	npm/​@​happy-dom/​global-registrator@​20.10.6

View full report