Express (Node.js)

@react-ai-stream/express is a one-line Express middleware that handles RAIS-compliant SSE streaming. It reads req.body.messages, calls your chosen provider, and pipes the stream back to the client with proper headers.

Install

npm install @react-ai-stream/express

Express itself is a peer dependency — install it separately if you haven't already.

Usage

import express from 'express'
import { raisMiddleware } from '@react-ai-stream/express'
 
const app = express()
app.use(express.json())
 
app.post('/api/chat', raisMiddleware({
  provider: 'openai',
  apiKey: process.env.OPENAI_API_KEY,
  model: 'gpt-4o-mini',
}))
 
// Anthropic:
// app.post('/api/chat', raisMiddleware({
//   provider: 'anthropic',
//   apiKey: process.env.ANTHROPIC_API_KEY,
//   model: 'claude-sonnet-4-6',
// }))
 
app.listen(3001)

What it does

Reads req.body.messages (array of {role, content} objects)
Sets Content-Type: text/event-stream, Cache-Control: no-cache, X-Accel-Buffering: no
Pipes AsyncIterable<StreamChunk> from provider → SSE response
Aborts the LLM stream when the client disconnects

API

`raisMiddleware(options)`

Option	Type	Description
`provider`	`'openai' \| 'anthropic'`	Use a built-in provider
`apiKey`	`string`	Provider API key
`model`	`string`	Model name (optional, has sensible default)
`handler`	`(messages, signal) => AsyncIterable<StreamChunk>`	Custom stream handler

Pass either provider + apiKey or a handler — not both.

Client

Point useAIChat at your Express server:

const chat = useAIChat({ endpoint: 'http://localhost:3001/api/chat' })

For same-origin Next.js apps, use a relative URL:

const chat = useAIChat({ endpoint: '/api/chat' })

Add cors middleware if your React app runs on a different port than Express.

Abort handling

The middleware listens for req.on('close') and aborts the provider stream when the client disconnects or clicks Stop. This cancels the upstream LLM call so you don't pay for tokens that nobody reads.

Overview Vue 3