services:
  neuroforge:
    build:
      context: .
      target: server
    ports:
      - "127.0.0.1:8080:8080"
    environment:
      NEUROFORGE_ADMIN_TOKEN: ${NEUROFORGE_ADMIN_TOKEN}
      NEUROFORGE_APP_API_KEY: ${NEUROFORGE_APP_API_KEY}
      NEUROFORGE_WORKER_TOKEN: ${NEUROFORGE_WORKER_TOKEN}
      NEUROFORGE_METRICS_TOKEN: ${NEUROFORGE_METRICS_TOKEN}
      NEUROFORGE_CLUSTER_TOKEN: ${NEUROFORGE_CLUSTER_TOKEN:-}
      OPENAI_API_KEY: ${OPENAI_API_KEY:-}
    volumes:
      - neuroforge-data:/app/data
    restart: unless-stopped
    read_only: true
    tmpfs:
      - /tmp:size=64m,mode=1777
    security_opt:
      - no-new-privileges:true
    cap_drop:
      - ALL
    healthcheck:
      test: ["CMD", "wget", "-q", "-O", "-", "http://127.0.0.1:8080/readyz"]
      interval: 15s
      timeout: 3s
      retries: 4
      start_period: 10s

  worker-cpu:
    build:
      context: .
      target: worker
    command: ["-server", "http://neuroforge:8080", "-id", "cpu-compose-1"]
    environment:
      NEUROFORGE_WORKER_TOKEN: ${NEUROFORGE_WORKER_TOKEN:?Set a unique NeuroForge worker token}
      NEUROFORGE_WORKER_RESOURCE_CLASS: cpu
      NEUROFORGE_WORKER_CAPABILITIES: cpu,vector.relink
      NEUROFORGE_WORKER_MAX_CONCURRENCY: ${NEUROFORGE_CPU_WORKER_CONCURRENCY:-2}
    depends_on:
      neuroforge:
        condition: service_healthy
    restart: unless-stopped
    read_only: true
    tmpfs: ["/tmp:size=32m,mode=1777"]
    security_opt: ["no-new-privileges:true"]
    cap_drop: ["ALL"]

  worker-gpu:
    build:
      context: .
      target: worker
    command: ["-server", "http://neuroforge:8080", "-id", "gpu-compose-1"]
    environment:
      NEUROFORGE_WORKER_TOKEN: ${NEUROFORGE_WORKER_TOKEN:?Set a unique NeuroForge worker token}
      NEUROFORGE_WORKER_RESOURCE_CLASS: gpu
      NEUROFORGE_WORKER_CAPABILITIES: gpu,model.chat,model.embed
      NEUROFORGE_WORKER_MAX_CONCURRENCY: ${NEUROFORGE_GPU_WORKER_CONCURRENCY:-1}
      NEUROFORGE_WORKER_OLLAMA_URL: ${OLLAMA_BASE_URL:-http://ollama:11434}
      NEUROFORGE_WORKER_OLLAMA_CHAT_MODEL: ${OLLAMA_MODEL:-gemma3}
      NEUROFORGE_WORKER_OLLAMA_EMBEDDING_MODEL: ${OLLAMA_EMBEDDING_MODEL:-embeddinggemma}
    depends_on:
      neuroforge:
        condition: service_healthy
    restart: unless-stopped
    read_only: true
    tmpfs: ["/tmp:size=32m,mode=1777"]
    security_opt: ["no-new-privileges:true"]
    cap_drop: ["ALL"]

volumes:
  neuroforge-data: