rais-server

Zero-config reference server for the RAIS Protocol v1.

Stream AI responses from OpenAI, Anthropic, or Groq with a single command — no code required.

Usage

# OpenAI
OPENAI_API_KEY=sk-... npx rais-server

# Anthropic
ANTHROPIC_API_KEY=sk-ant-... npx rais-server

# Groq
GROQ_API_KEY=gsk_... npx rais-server

# Server starts on http://localhost:3001/api/chat

Options

npx rais-server \
  --provider anthropic \
  --model claude-sonnet-4-6 \
  --port 4000 \
  --system "You are concise" \
  --max-tokens 256 \
  --no-cors

| Flag | Default | Description | |---|---|---| | --provider | auto-detected | openai, anthropic, or groq | | --model | provider default | Model name | | --port | 3001 | Port to listen on | | --system | — | System prompt | | --max-tokens | 1024 | Max tokens (Anthropic / Groq) | | --no-cors | CORS enabled | Disable CORS headers |

Provider is auto-detected from environment variables: ANTHROPIC_API_KEY takes priority, then OPENAI_API_KEY, then GROQ_API_KEY.

Test your endpoint

# Send a message
curl -N -X POST http://localhost:3001/api/chat \
  -H 'Content-Type: application/json' \
  -d '{"messages":[{"role":"user","content":"say hello in one word"}]}'

# Run full compliance check
npx rais-compliance http://localhost:3001/api/chat

Point useAIChat at it

import { useAIChat } from '@react-ai-stream/react'

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

Wire format

Every response is a RAIS-compliant SSE stream:

data: {"type":"text","text":"Hello"}

data: {"type":"text","text":"!"}

data: {"type":"done"}

License

MIT — part of react-ai-stream