Jyotisha/deploy

Railway deployment

Create services named web and api from this repository. Leave both Root Directory and Start Command empty; the Dockerfiles bind 0.0.0.0 and read $PORT.

Service	Required variable	Healthcheck Path
web	RAILWAY_DOCKERFILE_PATH=/deploy/railway-web.Dockerfile	/login
api	RAILWAY_DOCKERFILE_PATH=/deploy/railway-api.Dockerfile	/api/health

api variables

PORT=5200

PORT is set explicitly so the web service can reference the private API port. Railway healthchecks and the container command use the same value. No public API domain or legacy jyotish-app/ service is needed.

web variables

Required:

JYOTISH_API_BASE=http://${{api.RAILWAY_PRIVATE_DOMAIN}}:${{api.PORT}}
NEXT_PUBLIC_SUPABASE_URL=...
NEXT_PUBLIC_SUPABASE_ANON_KEY=...
SUPABASE_SERVICE_ROLE_KEY=...

For AI responses, configure either:

OPENAI_API_KEY=...
MASTRA_MODEL=openai/gpt-5-mini

or an OpenAI-compatible provider:

LLM_BASE_URL=...
LLM_API_KEY=...
LLM_MODEL=...
# LLM_PROVIDER_ID=third-party

ADMIN_EMAILS=admin@example.com,ops@example.com is required to use /admin/codes. Generate a public domain only for web.

After the first deploy

1. Add the generated web domain to Supabase Auth Site URL and Redirect URLs.
2. Open /login, sign in with an address listed in ADMIN_EMAILS, then verify /admin/codes.
3. Keep api private; verify its /api/health from Railway logs or the web service.