EZ Comfy: ComfyUI for non-technical creators (auto model/workflow/VRAM, exports back to node graph)

[KarloffsGhost]

Hi ComfyUI team and community, I built EZ Comfy to make ComfyUI approachable for non-technical users while still
respecting power-user workflows.

What it does

• User types a prompt in a simple UI
• EZ Comfy auto-selects model, workflow, and settings based on installed checkpoints + available VRAM
• Handles prompt adaptation for model families
• Helps avoid VRAM conflicts (including local LLM/Ollama handoff scenarios)
• Exports workflow JSON back into ComfyUI so advanced users can keep editing nodes directly

Why I built it
I wanted a smoother on-ramp for new creators, without replacing ComfyUI’s flexibility.

Repo

• https://github.com/KarloffsGhost/EZ_Comfy

Would love feedback on

1. Where this aligns/conflicts with ComfyUI frontend direction
2. Onboarding UX for first-time users
3. Any workflow or model-selection edge cases I should test next

[aniruddhaadak80]

From my point of view, the most promising part is that EZ Comfy exports back into the node graph instead of trapping users in a simplified surface forever. That keeps the onboarding layer compatible with ComfyUI''s core strength, which is inspectable, editable workflows.

The key thing I would watch is provenance. Users should be able to tell exactly which model, workflow, and VRAM decisions were made automatically so the handoff back to power-user mode stays trustworthy.

[KarloffsGhost]

Thanks so much for taking the time to review it and share this — I really appreciate it. I’m glad the export-back-to-node-graph approach resonated, since preserving ComfyUI’s inspectable/editable workflow model is a big part of the goal. Your point about provenance and making automatic decisions more visible is a very good one too. My next iteration is going to focus on incorporating that feedback. Again much appreciated that you took the time to provide feedback!