Deploying to Production

Using the CLI

# Generate Docker deployment files
Kernova deploy docker

# Or generate a Node.js production entry
Kernova deploy node

Docker Deployment

After running Kernova deploy docker:

npm run build
docker build -t my-app .
docker run -p 4200:4200 \
  -e OPENAI_API_KEY=sk-... \
  -e Kernova_AUTH=true \
  -v ./data:/app/Kernova-data \
  my-app

The generated Dockerfile uses Node.js 24 Alpine, installs only production dependencies, and exposes port 4200.

Environment Variables

Variable	Description	Default
`PORT`	Studio/API port	4200
`Kernova_DATA_DIR`	Persistence directory	./Kernova-data
`Kernova_AUTH`	Require auth for API	true
`NODE_ENV`	Environment	production
Provider keys	See Providers	—

Production Checklist

Set NODE_ENV=production
Set Kernova_AUTH=true (enforces API key auth)
Use a persistent volume for Kernova-data/
Set model provider API keys
Configure budget limits
Set up TLS (reverse proxy or tlsCert/tlsKey)
Configure event retention (applyRetention(30))

Health Check

The /health endpoint returns 200 OK without authentication — use it for load balancer probes:

GET /health → { "status": "healthy", "timestamp": "..." }

Scaling

v0.1 is single-process. For scaling:

Use a reverse proxy (nginx) for TLS and load balancing
Run multiple instances with shared PostgreSQL (planned for v0.2)
Use Redis for distributed event bus (planned for v0.3)