Flake Data Split (All)

FlakeDataSplitAll is the pack's escape hatch. The whole Flakes ecosystem runs on the flake_data bundle - a single wire that secretly carries your model, CLIP, VAE, both conditionings, the latent, the resolution, and the sampler settings. That's great while you stay inside the pack, but the moment you want to touch any of those values yourself, the bundle is a black box. This node opens it: feed it one flake_data pin and get all thirteen pieces back as ordinary, individually wireable outputs.

The full list: model, clip, vae, positive, negative, latent, filename_prefix, width, height, steps, cfg, sampler_name, scheduler. That's every pin you'd need to drive your own KSampler, your own VAE Decode, your own Save Image - which is the primary reason you'd reach for this node. You get a bundle from a Flake Model Preset or Flake Stack, but you want to run a custom sampler, inspect the latent, or feed one specific value into a node that expects a plain number. Split it, grab what you need, wire the rest into the void.

A few useful details from the source. filename_prefix isn't stored as a plain string in the bundle - it's built from the preset name plus the stack's output_stem entries, with the date folder and time stamp, so what comes out of this pin is the actual prefix Save Image will use. And sampler_name and scheduler come out as plain strings, not dropdowns - the node uses string outputs deliberately because ComfyUI doesn't allow combo lists as output types. In practice that's harmless: they plug straight into a KSampler's corresponding inputs.

It's also just a great debug tool. If a bundle isn't behaving, split it and preview the values, or drop the filename_prefix output somewhere visible to see what the output path will be before you commit to a long queue.

Inputs and outputs. One required input: flake_data. Thirteen outputs, listed above, always all present. If the all-or-nothing clutter bothers you, FlakeDataSplitSelect is the same node with a dropdown and +/- buttons so you only expose the pins you actually need.

Install. ComfyUI Manager → search ComfyUI Flakes → Install, or:

cd ComfyUI/custom_nodes
git clone https://github.com/JeyzerMC/comfyui-flakes

then restart. PyYAML is the only dependency and it already ships with ComfyUI core.