docker-compose.yml

# ══════════════════════════════════════════════════════════════════
#  TinAI v7 – docker-compose.yml (TOUT est optionnel)
# ══════════════════════════════════════════════════════════════════

services:

  # ── llama.cpp (optionnel)
  llama:
    profiles: ["llama"]
    image: ghcr.io/ggml-org/llama.cpp:server
    container_name: tinai-llama
    restart: unless-stopped
    ports:
      - "${PORT_LLAMA:-8081}:8080"
    volumes:
      - ${MODELS_DIR:-./models}:/models
      - ./scripts/llama-entrypoint.sh:/llama-entrypoint.sh
    entrypoint: ["sh", "/llama-entrypoint.sh"]
    environment:
      LLAMA_HF_REPO:          ${LLAMA_HF_REPO:-Qwen/Qwen3-1.7B-GGUF}
      LLAMA_HF_FILE:          ${LLAMA_HF_FILE:-Qwen3-1.7B-Q5_K_M.gguf}
      LLAMA_ARG_MODEL:        /models/${LLAMA_HF_FILE}
      LLAMA_ARG_HOST:         0.0.0.0
      LLAMA_ARG_PORT:         8080
      LLAMA_ARG_CTX_SIZE:     ${LLAMA_CTX_SIZE:-8192}
      LLAMA_ARG_N_PARALLEL:   ${LLAMA_N_PARALLEL:-1}
      LLAMA_ARG_THREADS:      ${LLAMA_THREADS:-4}
      LLAMA_ARG_N_GPU_LAYERS: ${LLAMA_GPU_LAYERS:-0}
      LLAMA_ARG_API_KEY:      ${TINAI_API_KEY:-sk-tinai}
      LLAMA_ARG_ALIAS:        ${LLAMA_MODEL_ALIAS:-qwen3-1.7b}
    mem_limit: ${MEM_LIMIT:-6g}
    healthcheck:
      test: ["CMD", "curl", "-sf", "http://localhost:8080/health"]
      interval: 30s
      timeout: 10s
      start_period: 120s
    networks:
      - tinai-net

  # ── OpenFang (optionnel)
  tinai:
    profiles: ["openfang"]
    build: .
    image: tinai:v7
    container_name: tinai-openfang
    restart: unless-stopped
    ports:
      - "${PORT_OPENFANG:-4200}:4200"
    volumes:
      - ${OPENFANG_DATA_DIR:-./data/openfang}:/root/.openfang
    environment:
      LLAMA_PORT:          8081
      LLAMA_HOST:          llama
      TINAI_API_KEY:       ${TINAI_API_KEY:-sk-tinai}
      ANTHROPIC_API_KEY:   ${ANTHROPIC_API_KEY:-}
      OPENAI_API_KEY:      ${OPENAI_API_KEY:-}
      GROQ_API_KEY:        ${GROQ_API_KEY:-}
    networks:
      - tinai-net

  # ── LiteLLM (optionnel)
  litellm:
    profiles: ["litellm"]
    image: ghcr.io/berriai/litellm:main-stable
    container_name: tinai-litellm
    restart: unless-stopped
    ports:
      - "${PORT_LITELLM:-4000}:4000"
    volumes:
      - ./litellm/config.yaml:/app/config.yaml:ro
    environment:
      LITELLM_MASTER_KEY: ${LITELLM_MASTER_KEY:-sk-tinai-litellm}
    networks:
      - tinai-net
    depends_on:
      llama:
        condition: service_healthy

  # ── Hermès Agent (optionnel)
  hermes:
    profiles: ["hermes"]
    image: nousresearch/hermes-agent:latest
    container_name: tinai-hermes
    restart: unless-stopped
    ports:
      - "${PORT_HERMES:-4201}:8080"
    volumes:
      - ${HERMES_DATA_DIR:-./data/hermes}:/opt/data
    environment:
      HERMES_LLM_BASE_URL: http://llama:8080/v1
      HERMES_LLM_API_KEY:  ${TINAI_API_KEY:-sk-tinai}
    command: gateway run
    networks:
      - tinai-net

  # ── Open WebUI (optionnel)
  webui:
    profiles: ["webui"]
    image: ghcr.io/open-webui/open-webui:main-slim
    container_name: tinai-webui
    restart: unless-stopped
    depends_on:
      llama:
        condition: service_healthy
    ports:
      - "${PORT_WEBUI:-3000}:8080"
    volumes:
      - ${WEBUI_DATA_DIR:-./data/webui}:/app/backend/data
    environment:
      OPENAI_API_BASE_URL: http://llama:8080/v1
      OPENAI_API_KEY:      ${TINAI_API_KEY:-sk-tinai}
      WEBUI_SECRET_KEY:    ${WEBUI_SECRET_KEY:-tinai-secret-change-me}
    networks:
      - tinai-net

  # ── Monitoring (optionnel)
  monitor:
    profiles: ["monitoring"]
    build: ./monitoring
    image: tinai-monitor:latest
    container_name: tinai-monitor
    restart: unless-stopped
    ports:
      - "9000:19999"
    networks:
      - tinai-net

networks:
  tinai-net:
    driver: bridge