Int

WtlInt is the most honest node in the pack: it outputs the integer you type into it, and nothing else. No processing, no preview system, no apply_type dropdown. Just a value slider (default 0, full 32-bit range) on the front and an int output on the back.

So why does this exist when every ComfyUI workflow is already full of bare integer widgets? Because a bare widget lives on the node it configures. You can't run a wire from the steps slider of one KSampler into another node without the value being trapped there. WtlInt exists to turn a number into a first-class graph citizen - something you can feed into multiple nodes, wire through a switch or a scheduler, or drive from a shared control. It's the integer sibling of ComfyUI's Int primitive, and it does exactly what that does, with the same honest range of −2³¹ to 2³¹−1.

The practical use is consistency. When five nodes need the same steps value, one WtlInt feeding all five beats editing five widgets every time you tune. It also plays nice with the pack's math nodes - WtlAdd, WtlSubtract, WtlMultiply, WtlDivide all accept INT on their x socket, so you can build a tiny arithmetic chain starting from one of these constants without a type mismatch anywhere.

It ships in the WtlNodes pack (Scorpiosis0/ComfyUI-WtlNodes). Install via ComfyUI Manager by searching "WtlNodes", or:

cd ComfyUI/custom_nodes
git clone https://github.com/Scorpiosis0/ComfyUI-WtlNodes.git

then restart ComfyUI. The pack's dependencies are numpy, scipy, and pillow - no models, no downloads, nothing GPU-related.

Honestly, if you already use stock ComfyUI's INT primitive, you may never need this node. But the pack's logic nodes are built around the same dual-int/float socket convention, and WtlInt is the natural constant source for them. It's a placeholder node in the best sense: it occupies a socket so the graph stays honest about where a number comes from.