# Minimal deployment: velox-ui alone, SQLite in a volume, no other services.
#
#   docker compose up -d
#
# Then open http://localhost:8080 and create the first account, which becomes the
# administrator. If Ollama is already running on the host, velox-ui finds it by itself.

services:
  velox-ui:
    image: lordraw/velox-ui:latest
    # Comment `image` and uncomment `build` to run from a source checkout.
    build: .
    container_name: velox-ui
    restart: unless-stopped
    ports:
      - "9080:8080"
    volumes:
      - velox-data:/data
    environment:
      # Generated into /data on first run when unset. Set it explicitly to keep
      # stored provider credentials readable across a rebuilt volume.
      # VELOX_SECRET_KEY: "change-me"
      VELOX_LOG_FORMAT: json
      VELOX_PROVIDER_OLLAMA_HOSTS: ${OLLAMA_HOSTS}
      # Read from a .env file next to this one (docker compose loads it
      # automatically), so the key itself is never written here or committed.
      VELOX_PROVIDER_NVIDIA_API_KEY: ${NVIDIA_API_KEY}
    # Lets the container reach an Ollama or llama.cpp server running on the host.
    # extra_hosts:
    #   - "host.docker.internal:host-gateway"
    env_file:
      - .env
volumes:
  velox-data: