runpod-ai-orchestrator/arty.yml

name: "RunPod AI Service Orchestrator"
version: "0.1.0"
description: "Process-based AI service orchestrator for RunPod GPU instances with ComfyUI integration"
author: "valknar@pivoine.art"
license: "MIT"

# Environment profiles for selective repository management
envs:
  # RunPod environment variables
  default:
    COMFYUI_ROOT: services/comfyui
    AUDIOCRAFT_ROOT: services/audiocraft
    UPSCALE_ROOT: services/upscale

# Git repositories to clone for a fresh RunPod deployment
references:
  # AudioCraft Studio
  - url: https://dev.pivoine.art/valknar/audiocraft-ui.git
    into: $AUDIOCRAFT_ROOT
  # Upscale UI
  - url: https://dev.pivoine.art/valknar/upscale-ui.git
    into: $UPSCALE_ROOT
  # ComfyUI base installation
  - url: https://github.com/comfyanonymous/ComfyUI.git
    into: $COMFYUI_ROOT
    description: "ComfyUI - Node-based interface for image/video/audio generation"

  # ComfyUI Essential Custom Nodes
  - url: https://github.com/ltdrdata/ComfyUI-Manager.git
    into: $COMFYUI_ROOT/custom_nodes/ComfyUI-Manager
    description: "ComfyUI Manager - Install/manage custom nodes and models"

  - url: https://github.com/Kosinkadink/ComfyUI-VideoHelperSuite.git
    into: $COMFYUI_ROOT/custom_nodes/ComfyUI-VideoHelperSuite
    description: "Video operations and processing"

  - url: https://github.com/Kosinkadink/ComfyUI-AnimateDiff-Evolved.git
    into: $COMFYUI_ROOT/custom_nodes/ComfyUI-AnimateDiff-Evolved
    description: "AnimateDiff for video generation"

  - url: https://github.com/cubiq/ComfyUI_IPAdapter_plus.git
    into: $COMFYUI_ROOT/custom_nodes/ComfyUI_IPAdapter_plus
    description: "IP-Adapter for style transfer"

  - url: https://github.com/ltdrdata/ComfyUI-Impact-Pack.git
    into: $COMFYUI_ROOT/custom_nodes/ComfyUI-Impact-Pack
    description: "Auto face enhancement and detailer"

  # ComfyUI Optional Custom Nodes
  - url: https://github.com/kijai/ComfyUI-CogVideoXWrapper.git
    into: $COMFYUI_ROOT/custom_nodes/ComfyUI-CogVideoXWrapper
    description: "CogVideoX integration for text-to-video"

  - url: https://github.com/ltdrdata/ComfyUI-Inspire-Pack.git
    into: $COMFYUI_ROOT/custom_nodes/ComfyUI-Inspire-Pack
    description: "Additional inspiration tools"

  - url: https://github.com/Kosinkadink/ComfyUI-Advanced-ControlNet.git
    into: $COMFYUI_ROOT/custom_nodes/ComfyUI-Advanced-ControlNet
    description: "Advanced ControlNet features"

  - url: https://github.com/MrForExample/ComfyUI-3D-Pack.git
    into: $COMFYUI_ROOT/custom_nodes/ComfyUI-3D-Pack
    description: "3D asset generation"

  - url: https://github.com/ssitu/ComfyUI_UltimateSDUpscale.git
    into: $COMFYUI_ROOT/custom_nodes/ComfyUI_UltimateSDUpscale
    description: "Ultimate SD Upscale for high-quality image upscaling"

  - url: https://github.com/kijai/ComfyUI-KJNodes.git
    into: $COMFYUI_ROOT/custom_nodes/ComfyUI-KJNodes
    description: "Kijai optimizations for HunyuanVideo and Wan2.2 (FP8 scaling, video helpers, model loading)"

  - url: https://github.com/Fannovel16/comfyui_controlnet_aux.git
    into: $COMFYUI_ROOT/custom_nodes/comfyui_controlnet_aux
    description: "ControlNet preprocessors (Canny, Depth, OpenPose, MLSD) for Wan2.2 Fun Control"

  - url: https://github.com/city96/ComfyUI-GGUF.git
    into: $COMFYUI_ROOT/custom_nodes/ComfyUI-GGUF
    description: "GGUF quantization support for memory-efficient model loading"

  - url: https://github.com/11cafe/comfyui-workspace-manager.git
    into: $COMFYUI_ROOT/custom_nodes/comfyui-workspace-manager
    description: "Workspace manager for ComfyUI - workflow/model organization"

  - url: https://dev.pivoine.art/valknar/comfyui-extras.git
    into: $COMFYUI_ROOT/custom_nodes/comfyui-extras

# Deployment scripts for RunPod instances
scripts:
  setup: |
    arty deps
    arty setup/services
    arty setup/comfyui
    arty models/download
    arty models/link

  setup/services: |
    echo "========================================="
    echo "  Setting up services python venvs"
    echo "========================================="
    echo ""

    # Install dependencies for each custom node
    echo "Setting up venvs for all services..."
    cd services

    installed_count=0
    skipped_count=0

    for dir in */; do
      cd $dir
      python3 -m venv venv && source venv/bin/activate
      pip install --upgrade pip
      if [ -f "requirements.txt" ]; then
        echo ""
        echo "📦 Installing ${dir%/} dependencies..."
        if pip install -r "requirements.txt"; then
          installed_count=$((installed_count + 1))
          echo "  ✓ ${dir%/} dependencies installed"
        else
          echo "  ⚠ Warning: Some dependencies for ${dir%/} may have failed"
          installed_count=$((installed_count + 1))
        fi
      else
        skipped_count=$((skipped_count + 1))
      fi
      deactivate
      cd -
    done

    echo ""
    echo "✓ Service environments configured successfully"
    echo "  Services with dependencies: $installed_count"
    echo "  Services without requirements.txt: $skipped_count"

  setup/comfyui: |
    echo "========================================="
    echo "  Setting up ComfyUI"
    echo "========================================="
    echo ""

    cd $COMFYUI_ROOT

    # Activate ComfyUI venv
    if [ ! -d "venv" ]; then
      echo "❌ ERROR: ComfyUI venv not found. Run setup/venvs first!"
      exit 1
    fi

    source venv/bin/activate

    # Install common extension dependencies
    echo "Installing common extension dependencies..."
    pip install GitPython opencv-python-headless diffusers insightface onnxruntime

    # Create model directory structure
    echo "Creating ComfyUI model directories..."
    mkdir -p $COMFYUI_ROOT/models/{checkpoints,unet,vae,loras,clip,clip_vision,controlnet,ipadapter,embeddings,upscale_models,video_models,animatediff_models,animatediff_motion_lora,audio_models,configs,diffusers,diffusion_models,musicgen,latent_upscale_models}

    # Install dependencies for each custom node
    echo "Installing dependencies for all custom nodes..."
    cd custom_nodes

    installed_count=0
    skipped_count=0

    for dir in */; do
      if [ -f "${dir}requirements.txt" ]; then
        echo ""
        echo "📦 Installing ${dir%/} dependencies..."
        if pip install -r "${dir}requirements.txt"; then
          installed_count=$((installed_count + 1))
          echo "  ✓ ${dir%/} dependencies installed"
        else
          echo "  ⚠ Warning: Some dependencies for ${dir%/} may have failed"
          installed_count=$((installed_count + 1))
        fi
      else
        skipped_count=$((skipped_count + 1))
      fi
    done

    deactivate

    echo ""
    echo "✓ Extension dependencies installation complete"
    echo "  Extensions with dependencies: $installed_count"
    echo "  Extensions without requirements.txt: $skipped_count"

  models/download: |
    artifact_civitai_download.sh download -c ./models/models_civitai.yaml --cache-dir ./.cache/civitai
    artifact_huggingface_download.sh download -c ./models/models_huggingface.yaml --cache-dir ./.cache/huggingface

  models/link: |
    artifact_civitai_download.sh link -c ./models/models_civitai.yaml --cache-dir $PWD/.cache/civitai --output-dir ./services/comfyui/models
    artifact_huggingface_download.sh link -c ./models/models_huggingface.yaml --cache-dir $PWD/.cache/huggingface --output-dir ./services/comfyui/models

  tailscale/start: |
    echo "========================================="
    echo "  Starting tailscale"
    echo "========================================="
    echo ""

    if [ ! "$TAILSCALE_AUTHKEY" ]; then
      echo "  ⚠ Tailscale disabled (no TAILSCALE_AUTHKEY in env)"
      echo "  Services requiring VPN connectivity will not work"
      exit 1
    fi

    # Persist Tailscale state to network volume for stable IP across restarts
    TAILSCALE_STATE="/workspace/.tailscale"
    mkdir -p "$TAILSCALE_STATE"

    echo "  Starting Tailscale daemon..."
    tailscaled --tun=userspace-networking --socks5-server=localhost:1055 --statedir="$TAILSCALE_STATE" &
    sleep 3

    echo "  Connecting to Tailscale network..."
    HOSTNAME="runpod-ai-orchestrator"
    tailscale up --authkey="$TAILSCALE_AUTHKEY" --advertise-tags=tag:gpu --hostname="$HOSTNAME" || {
        echo "  ⚠ Tailscale connection failed, continuing without VPN"
    }

    # Get Tailscale IP if connected
    TAILSCALE_IP=$(tailscale ip -4 2>/dev/null || echo "not connected")
    if [ "$TAILSCALE_IP" == "not connected" ]; then
        echo "  ⚠ Tailscale failed to obtain IP"
        exit 1
    fi

    echo "  ✓ Tailscale connected"
    echo "    Hostname: $HOSTNAME"
    echo "    IP: $TAILSCALE_IP"

  #
  # Supervisor Control Scripts
  #
  supervisor/start: |
    mkdir -p .logs/
    supervisord -c supervisord.conf

  supervisor/stop: |
    supervisorctl -c supervisord.conf shutdown

  supervisor/status: |
    supervisorctl -c supervisord.conf status

  supervisor/restart: |
    supervisorctl -c supervisord.conf restart all

  supervisor/reload: |
    supervisorctl -c supervisord.conf reload