ComfyUI Workflow Studio — Online Node Editor

Drag nodes · Click output, then input · Pinch or wheel to zoom · Drag background to pan

Choose a node or a starter workflow.

Inspector

Select a node to edit its values.

Prompt → workflow

Describe a checkpoint, Flux.1 Dev or SD3 text-to-image workflow. AI proposes settings; OnBeg builds a core-node graph for you to review. Three trial AI requests per signed-in account; Pro uses the monthly allowance.

Send only this description to Cloudflare AI. My graph and files stay local.

Installed checkpoint filenameComfyUI input image filename (image-to-image)

Flux / SD3 installed model files

Diffusion model filename (Flux / SD3)CLIP-L filenameCLIP-G filename (SD3)T5 encoder filenameVAE filename (Flux / SD3)

GGUF diffusion files require ComfyUI-GGUF. SD3 needs its matching VAE; Flux.1 uses ae.safetensors.

Reference image → visual prompt

A JPEG preview up to 768px is prepared locally without original metadata. Only the preview is sent after approval.

I approve sending this image preview to Cloudflare AI.

AI planning supports checkpoint text-to-image/image-to-image, Flux.1 Dev and SD3 text-to-image, optionally with interpolation resize. Image analysis suggests a visual prompt; it cannot recover the original workflow. Applying creates an undoable graph; set your real files and test in ComfyUI.

Edit raw node JSON

Export ComfyUI UI JSON