CASCADE/docker-compose.yml at main · CederGroupHub/CASCADE · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
services:
  # MLflow Tracking Server for observability
  mlflow:
    image: ghcr.io/mlflow/mlflow:v3.3.2
    container_name: cascade_mlflow
    network_mode: host  # Use host network for localhost access
    user: "${UID:-1000}:${GID:-1000}"  # Run as current user to match permissions
    environment:
      - MLFLOW_HOST=0.0.0.0
    volumes:
      # Mount local mlruns directory for persistence across rebuilds
      - ./conversational_system/mlruns:/mlruns
    # Use file-based backend (same as non-Docker mlflow ui) for consistent trace display
    command: mlflow server --host 0.0.0.0 --port 5001 --backend-store-uri file:///mlruns --default-artifact-root file:///mlruns
    healthcheck:
      test: ["CMD", "python", "-c", "import urllib.request; urllib.request.urlopen('http://localhost:5001/')"]
      interval: 10s
      timeout: 10s
      retries: 5
      start_period: 10s
    restart: unless-stopped

  # CASCADE Conversational System
  # NOTE: Neo4j is NOT included - uses your local Neo4j instance for data sharing
  cascade:
    build:
      context: .
      dockerfile: docker/Dockerfile
    container_name: cascade_app
    network_mode: host  # Use host network for localhost access to Neo4j
    user: "${UID:-1000}:${GID:-1000}"  # Run as current user to match permissions
    environment:
      # Home directory for user-writable cache (crawl4ai, etc.)
      - HOME=/app/.home
      # Playwright browsers are installed in cascade_user's home during build
      - PLAYWRIGHT_BROWSERS_PATH=/home/cascade_user/.cache/ms-playwright
      # API Keys
      - OPENAI_API_KEY=${OPENAI_API_KEY}
      - TAVILY_API_KEY=${TAVILY_API_KEY}
      - MP_API_KEY=${MP_API_KEY}
      # Supabase
      - SUPABASE_URL=${SUPABASE_URL}
      - SUPABASE_SERVICE_KEY=${SUPABASE_SERVICE_KEY}
      - SUPABASE_DATABASE_URL=${SUPABASE_DATABASE_URL}
      # Neo4j
      - NEO4J_URI=${NEO4J_URI:-bolt://localhost:7687}
      - NEO4J_USER=${NEO4J_USER:-neo4j}
      - NEO4J_PASSWORD=${NEO4J_PASSWORD}
      # Memory configuration
      - MEM0_USE_LLM=${MEM0_USE_LLM:-true}
      - MEM0_LLM_MODEL=${MEM0_LLM_MODEL:-gpt-4o-mini}
      - ENABLE_GRAPH_MEMORY=${ENABLE_GRAPH_MEMORY:-true}
      # Knowledge graph for code parsing (requires Neo4j)
      - USE_KNOWLEDGE_GRAPH=${USE_KNOWLEDGE_GRAPH:-true}
      # Generate AI summaries for extracted code (increases API usage)
      - GENERATE_CODE_SUMMARY=${GENERATE_CODE_SUMMARY:-false}
      # Agent model configuration
      - AGENT_MODEL_NAME=${AGENT_MODEL_NAME:-o3}
      - OPENAI_BASE_URL=${OPENAI_BASE_URL:-https://api.openai.com/v1}
      - USE_ADAPTIVE_AGENTS=${USE_ADAPTIVE_AGENTS:-false}
      - REQUIRE_OPENAI_API_KEY=${REQUIRE_OPENAI_API_KEY:-true}
      # Application paths
      - PROJECT_ROOT=${PROJECT_ROOT:-/app}
      - CODE_STORAGE_DIR=${CODE_STORAGE_DIR:-/app/temp_code}
      - SAVED_FILES_DIR=${SAVED_FILES_DIR:-/app/saved_code}
      # MLflow tracking
      - MLFLOW_TRACKING_URI=${MLFLOW_TRACKING_URI:-http://localhost:5001}
      # Streamlit settings
      - STREAMLIT_BROWSER_GATHER_USAGE_STATS=false
      # Database path (mounted in /app/data for write access)
      - CONVERSATIONS_DB_PATH=/app/data/conversations.db
    volumes:
      # Mount local directories for persistence across rebuilds
      - ./conversational_system/conversations.db:/app/data/conversations.db
      - ./conversational_system/saved_code:/app/saved_code
      - ./conversational_system/mlruns:/mlruns  # Same path as MLflow server for trace data
    depends_on:
      mlflow:
        condition: service_healthy
    restart: unless-stopped