CWK Infos Extractor

CWK Infos Extractor is the companion to CWK Model Preset Manager, and it does exactly one thing: it takes the infos JSON string your main node spits out and fans it into fifteen individual STRING outputs. No model loading, no settings, no UI - just a tiny text-splitting utility. It exists because the parent node already bundles all its resolved settings (sampler, scheduler, CFG, steps, clip skip, resolution, even the RNG mode) into a single machine-readable string, and if you want to use any of those values downstream, you'd otherwise have to parse JSON yourself or wire up a generic JSON-to-text node.

Here's the flow: your CWK Model Preset Manager emits an infos output containing a JSON dict of everything it resolved - model name, sampler, scheduler, CFG, steps, clip skip, RNG source, width, height, batch size, and more. Connect that pin to this node's single infos input and you get clean, typed STRING pins for each value. It's the difference between one opaque blob and fifteen things you can actually wire into display nodes, logging nodes, or any custom node that accepts a string input.

The one thing worth knowing

The node isn't just copying strings around - it cleans them. Model, VAE, and CLIP names get their subfolder prefix and file extension stripped, so Pony\autismmixSDXL_autismmixPony.safetensors arrives at the other end as autismmixSDXL_autismmixPony. That's the field you want for a display label or a filename-based workflow. The remaining outputs are self-explanatory once you see them: sampler_name, scheduler, cfg, steps, clip_skip, rng, model_sampling, width, height, resolution (formatted as 1024x1024), batch_size, plus clip_type.

Two practical notes. First, everything comes out as a STRING, even numeric-looking fields like cfg and steps - so if you're feeding them into a node that wants an INT or FLOAT, you'll need a conversion node in between. Second, if the input isn't valid JSON (or is empty), the node gracefully returns empty strings for every output rather than crashing the graph. It's defensive in exactly the way you want a utility node to be.

Installing it

There's no separate install - CWK Infos Extractor ships in the same pack as the model manager. Search CWK Checkpoints Preset Manager in ComfyUI Manager, or:

cd ComfyUI/custom_nodes
git clone https://github.com/cowneko/CWK_Checkpoints_Preset_Manager.git

Then restart ComfyUI. It has zero dependencies of its own; if the main pack runs, this runs.

Honest take

This is a convenience node, not a requirement. If you're just building a standard image workflow, you'll never touch it - the parent node already gives you typed steps, cfg, sampler_name, scheduler, width, and height outputs directly. Reach for CWK Infos Extractor when you want those values as text: a workflow that logs its own settings, a label that shows the sampler used, or a custom node downstream that only accepts strings. That's a narrow job, and it does it cleanly. For everyone else, it's the node you'll add once, wonder about, and quietly leave in the back of your graph.