Next and Vercel Runtime

Use the serverless runtime when work must continue after a request ends and the deployment may be cold between wakes.

Install

pnpm add @use-crux/core @use-crux/postgres @use-crux/upstash pg @upstash/qstash

Configure runtime

import { config } from '@use-crux/core'
import { serverless } from '@use-crux/core/runtime'
import { postgres } from '@use-crux/postgres/runtime'
import { qstash } from '@use-crux/upstash/runtime'

export default config({
  runtime: serverless({
    store: postgres({ url: process.env.DATABASE_URL }),
    wake: qstash(),
    publicUrl: process.env.CRUX_PUBLIC_URL,
    namespace: process.env.CRUX_RUNTIME_NAMESPACE,
    retention: {
      events: '24h',
      terminalWork: '7d',
      terminalSnapshots: '30d',
    },
  }),
})

publicUrl can be omitted when Crux can infer a stable URL from Vercel environment variables. Production setup fails with PUBLIC_URL_UNRESOLVED when no stable URL exists.

Generate the route

Run generation before framework build:

crux runtime generate

Generated Next files target the public handler API:

import { createRuntimeHandler } from '@use-crux/core/runtime'
import { reviewFlow } from '@/flows/review'
import { embedDocument } from '@/tasks/embed-document'

export const { GET, POST } = createRuntimeHandler({
  targets: [reviewFlow, embedDocument],
})

POST verifies the wake request before durable writes. GET returns health metadata for setup and preflight without exposing secrets.

Verify setup

crux runtime setup --check
crux runtime status

Use crux runtime setup --apply only when you want Crux to create missing Crux-owned Postgres resources. It is additive only.

Production notes

• Keep DATABASE_URL, QStash credentials, and CRUX_PUBLIC_URL in deployment secrets.
• Use an explicit CRUX_RUNTIME_NAMESPACE when preview and production share the same substrate.
• Keep retention.idempotencyKeys at least as long as QStash's max wake delay, or set it to false.
• Regenerate artifacts in CI and fail builds on ARTIFACTS_STALE.
• Inspect stuck work with crux runtime inspect <workId>; retry blocked/dead-letter work with crux runtime retry <workId> after fixing the cause.

Related

• Reference: Runtime Engine
• Reference: @use-crux/postgres
• Reference: @use-crux/upstash