Add ComfyUI as an optional service in the installer

- Updated .env.example to include COMFYUI_HOSTNAME for configuration.
- Modified Caddyfile to add a reverse proxy for ComfyUI at the specified hostname.
- Enhanced docker-compose.yml to include a new ComfyUI service with persistent storage and health checks.
- Updated scripts to integrate ComfyUI into the setup wizard and final report.
- Documented the integration process and reflections in the memory bank for future reference.

.env.example

@@ -138,6 +138,7 @@ PROMETHEUS_HOSTNAME=prometheus.yourdomain.com
 PORTAINER_HOSTNAME=portainer.yourdomain.com
 LETTA_HOSTNAME=letta.yourdomain.com
 QDRANT_HOSTNAME=qdrant.yourdomain.com
+COMFYUI_HOSTNAME=comfyui.yourdomain.com
 LETSENCRYPT_EMAIL=
 
 # Everything below this point is optional.

Caddyfile

@@ -73,6 +73,11 @@
     reverse_proxy qdrant:6333
 }
 
+# ComfyUI
+{$COMFYUI_HOSTNAME} {
+    reverse_proxy comfyui:8188
+}
+
 # Neo4j
 {$NEO4J_HOSTNAME} {
     reverse_proxy neo4j:7474

README.md

@@ -30,6 +30,8 @@ The installer also makes the following powerful open-source tools **available fo
 
 ✅ [**Flowise**](https://flowiseai.com/) - A no-code/low-code AI agent builder that complements n8n perfectly, allowing you to create sophisticated AI applications with ease.
 
+✅ [**ComfyUI**](https://github.com/comfyanonymous/ComfyUI) - A powerful, node-based UI for Stable Diffusion workflows. Build and run image-generation pipelines visually, with support for custom nodes and extensions.
+
 ✅ [**Dify**](https://dify.ai/) - An open-source AI application development platform that provides comprehensive LLMOps capabilities, including workflow management, prompt engineering, RAG pipelines, and AI agent orchestration. Perfect for building production-ready AI applications.
 
 ✅ [**Qdrant**](https://qdrant.tech/) - A high-performance open-source vector store, specialized for AI. While Supabase also offers vector capabilities, Qdrant is included for its speed, making it ideal for demanding AI tasks.
@@ -134,6 +136,7 @@ After successful installation, your services are up and running! Here's how to g
     - **SearXNG:** `searxng.yourdomain.com`
     - **Prometheus:** `prometheus.yourdomain.com` (Typically used as a data source for Grafana)
     - **Portainer:** `portainer.yourdomain.com` (Protected by Caddy basic auth; on first login, complete Portainer admin setup)
+    - **ComfyUI:** `comfyui.yourdomain.com` (Node-based Stable Diffusion UI)
 
 2.  **Explore n8n:**
 

docker-compose.yml

@@ -16,6 +16,7 @@ volumes:
   letta_data:
   weaviate_data:
   portainer_data:
+  comfyui_data:
 
 x-n8n: &service-n8n
   build:
@@ -214,6 +215,7 @@ services:
       - SEARXNG_USERNAME=${SEARXNG_USERNAME}
       - SEARXNG_PASSWORD_HASH=${SEARXNG_PASSWORD_HASH}
       - PORTAINER_HOSTNAME=${PORTAINER_HOSTNAME}
+      - COMFYUI_HOSTNAME=${COMFYUI_HOSTNAME}
     cap_drop:
       - ALL
     cap_add:
@@ -581,3 +583,18 @@ services:
     volumes:
       - portainer_data:/data
       - ${DOCKER_SOCKET_LOCATION:-/var/run/docker.sock}:/var/run/docker.sock
+
+  comfyui:
+    image: yanwk/comfyui-boot:cu124-slim
+    container_name: comfyui
+    profiles: ["comfyui"]
+    restart: unless-stopped
+    environment:
+      - CLI_ARGS=--listen 0.0.0.0
+    volumes:
+      - comfyui_data:/home/runner
+    healthcheck:
+      test: ["CMD", "wget", "-qO", "/dev/null", "http://localhost:8188"]
+      interval: 10s
+      timeout: 5s
+      retries: 5

memory-bank/reflection/reflection-comfyui-integration.md

@@ -0,0 +1,34 @@
+# Reflection: ComfyUI Integration (Level 2)
+
+## Summary
+Added ComfyUI as an optional service using Docker Compose profile `comfyui`, proxied via Caddy at `COMFYUI_HOSTNAME`. Default CPU support, optional GPU planned. Updated `.env.example`, `docker-compose.yml`, `Caddyfile`, `scripts/04_wizard.sh`, `scripts/06_final_report.sh`, and `README.md`.
+
+## What Went Well
+- Followed existing installer patterns (profiles, Caddy host blocks, env generation/wizard/reporting) with minimal, clear edits
+- Simple reverse proxy through Caddy; WebSocket support expected to work without extra config
+- Compose config validated successfully
+
+## Challenges
+- No single “official” Docker image; community images differ in volume layout and flags
+- Volume paths for models/output/custom_nodes vary by image; chose a conservative mount point for persistence
+- GPU enablement requires NVIDIA toolkit and compose device reservations (not universally available)
+
+## Lessons Learned
+- Keep defaults CPU-first to minimize friction; add GPU as an opt-in
+- Abstract image details behind a validation checklist (port 8188, volume paths, CLI flags)
+- Document model storage and persistence expectations explicitly
+
+## Improvements / Next Steps
+- Consider adding a GPU-specific profile variant (e.g., `comfyui-gpu-nvidia`) when environment supports it
+- Evaluate switching to a more widely adopted/maintained image and standardize volume mappings
+- Extend final report with quick pointers to model directories and basic usage tips
+
+## Verification Checklist
+- Implementation reviewed end-to-end: YES
+- Successes documented: YES
+- Challenges documented: YES
+- Lessons learned documented: YES
+- Process/Technical improvements identified: YES
+
+## Impact
+- New optional service enabling visual Stable Diffusion workflows in the installer with minimal complexity

memory-bank/tasks.md

@@ -466,3 +466,87 @@ Integrate Portainer Community Edition as an optional service to manage the local
 ### Archiving Status (Portainer)
 - [x] Archive document created: `memory-bank/archive/feature-portainer-integration_20250808.md`
 - [x] tasks.md marked COMPLETE for Portainer
+
+---
+
+## New Task: Add ComfyUI Service (Node-based UI for SD Workflows)
+
+### Description
+Integrate ComfyUI as an optional service in the installer, proxied by Caddy at a configurable hostname. Default to CPU-only for simplicity; allow optional GPU via NVIDIA Container Toolkit if present.
+
+### Complexity
+- Level: 2 (Simple Enhancement)
+- Type: Add-on service via Docker Compose profile + Caddy
+
+### Options Considered (Research)
+- Dockerized ComfyUI service using a maintained community image (e.g., ghcr.io/ai-dock/comfyui or equivalent)
+- Bare-metal Python install managed by scripts (higher maintenance, not aligned with project patterns)
+- Integrate as an extension of AUTOMATIC1111 (not applicable to this project’s stack)
+- Build our own image from source (heavier maintenance)
+
+→ Recommended: Use a maintained ComfyUI Docker image exposing port 8188, mount persistent volumes for models and custom nodes, reverse-proxy with Caddy. GPU support remains optional via compose flags when host supports NVIDIA.
+
+### Technology Stack
+- Image: Maintained ComfyUI Docker image (to be validated during tech gate)
+- Port: 8188 (internal)
+- Reverse proxy: Caddy with HTTPS at `COMFYUI_HOSTNAME`
+- Storage: Named volume `comfyui_data` (models, input, output, custom_nodes)
+- GPU: Optional via NVIDIA toolkit (compose device reservations)
+
+### Files to Modify
+- `.env.example`: add `COMFYUI_HOSTNAME`
+- `docker-compose.yml`: add `comfyui` service with `profiles: ["comfyui"]`, volumes, healthcheck, optional GPU stanza
+- `Caddyfile`: add host block for `{$COMFYUI_HOSTNAME}` → `reverse_proxy comfyui:8188`
+- `scripts/04_wizard.sh`: add `comfyui` option with description
+- `scripts/06_final_report.sh`: add ComfyUI section with URL
+- `scripts/03_generate_secrets.sh`: generate default hostname `COMFYUI_HOSTNAME`
+
+### Implementation Steps
+1) `.env.example`
+   - Add `COMFYUI_HOSTNAME=comfyui.yourdomain.com`
+2) Wizard
+   - Add `comfyui` to selectable services list
+3) docker-compose
+   - Add `comfyui` service (image, port 8188, volumes: `comfyui_data:/data` or image-appropriate paths)
+   - Add `comfyui_data:` to top-level volumes
+   - Optional GPU: add NVIDIA device reservations when available
+4) Caddy
+   - Add site for `{$COMFYUI_HOSTNAME}` with `reverse_proxy comfyui:8188`
+5) Final report
+   - Print ComfyUI URL when profile active
+6) Secrets script
+   - Generate/populate `COMFYUI_HOSTNAME` similar to other hostnames
+
+### Potential Challenges
+- Large model storage footprint; ensure persistent volume and document where to place models
+- GPU optionality: only enable when NVIDIA toolkit exists; keep CPU default to avoid install friction
+- WebSockets: Caddy generally handles WS automatically; verify UI works via proxy
+
+### Technology Validation Checkpoints
+- [ ] Confirm maintained image name and tag
+- [ ] Verify port 8188 and container paths for volumes (models/custom_nodes)
+- [ ] Validate Caddy reverse proxy works (incl. websockets)
+- [ ] Optional: Validate GPU flags on a host with NVIDIA toolkit
+
+### Testing Strategy
+- Start only Caddy + ComfyUI with profile enabled
+- Access `https://COMFYUI_HOSTNAME` and verify UI loads and basic workflow runs
+- Confirm persistence of uploads/outputs in `comfyui_data`
+
+### Next Mode Recommendation
+- Implement Mode (no creative phase required)
+
+### Reflection Status (ComfyUI)
+- [x] Implementation thoroughly reviewed
+- [x] Successes documented
+- [x] Challenges and solutions analyzed
+- [x] Lessons Learned documented
+- [x] Process/Technical improvements identified
+- [x] reflection-comfyui-integration.md created
+- [x] tasks.md updated with reflection status
+
+### Reflection Highlights (ComfyUI)
+- **What Went Well**: Minimal changes following established patterns (profiles, Caddy, wizard, README, final report); compose validated successfully.
+- **Challenges**: Lack of a clearly “official” image; differing volume paths across images; optional GPU support trade-offs.
+- **Lessons Learned**: Default to CPU for broad compatibility; keep image choice abstract to allow swapping; add validation checklist for volumes and websockets.
+- **Improvements**: Consider adding a GPU sub-profile and documenting model storage locations; later evaluate swapping to a more authoritative image with stable volume conventions.

scripts/04_wizard.sh

@@ -66,6 +66,7 @@ base_services_data=(
     "open-webui" "Open WebUI (ChatGPT-like Interface)"
     "searxng" "SearXNG (Private Metasearch Engine)"
     "ollama" "Ollama (Local LLM Runner - select hardware in next step)"
+    "comfyui" "ComfyUI (Node-based Stable Diffusion UI)"
 )
 
 services=() # This will be the final array for whiptail

scripts/06_final_report.sh

@@ -135,6 +135,13 @@ if is_profile_active "portainer"; then
   echo "(Note: On first login, Portainer will prompt to set up an admin user.)"
 fi
 
+if is_profile_active "comfyui"; then
+  echo
+  echo "================================= ComfyUI ============================="
+  echo
+  echo "Host: ${COMFYUI_HOSTNAME:-<hostname_not_set>}"
+fi
+
 if is_profile_active "qdrant"; then
   echo
   echo "================================= Qdrant =============================="