Random Number

This one's actually well documented in WAS's own README, more so than most of its neighbors, and it's worth quoting directly because the node does more than its name suggests. Set a min and max, and Random Number gives you, per the pack's own description: a random integer between min and max (inclusive, uniformly distributed), a random float between min and max (also inclusive and uniform), a random 0-or-1 value that reads as a boolean if you take it off the int output, and a random shuffled list of the integers in that range, returned as a string - the README's own example: with min=0 and max=3, a possible result is the string '3,1,2,0'.

Why it's not "just" the sampler's seed

It's tempting to think ComfyUI's noise seed already covers randomness and stop there, but the seed only randomizes generation - the actual diffusion noise. Random Number is a general-purpose random value you can wire anywhere a NUMBER or a shuffled sequence is useful: randomizing a crop offset between runs, picking among several LoRA strengths without hardcoding one, driving a random choice of prompt fragment, or generating a shuffled batch order via the list output. It's decoupled from sampling entirely - you can randomize structure and layout decisions in your graph independent of whatever the KSampler's own seed is doing.

The inputs and outputs that matter

min and max define the range everything else draws from. From there the node hands back several flavors at once, per the README: an integer roll, a float roll, a 0/1 value usable as a boolean, and - distinct from the other three - a shuffled string list of every integer in the range, not just a single draw.

Worth knowing about too: WAS also ships a separate True Random.org Number Generator node that pulls genuinely non-deterministic entropy from atmospheric noise via random.org's API (it needs an API key from your random.org account). Random Number, by contrast, is the fast, local, no-network option - reach for random.org's node specifically if you need cryptographic-grade randomness rather than a standard pseudo-random draw, which is what the vast majority of graphs actually want.

Installing it

ComfyUI Manager: search "WAS Node Suite," install, restart. Manually:

cd ComfyUI/custom_nodes
git clone https://github.com/WASasquatch/was-node-suite-comfyui

install requirements.txt against your ComfyUI Python, restart. No model download, no heavy dependency.

Common issues & troubleshooting

Getting a different value every queue when I wanted it fixed. That's expected - it's a random draw, re-rolled on execution. If you need a value to stay put while you iterate on the rest of the graph, generate it once, note the result, and swap in a Constant Number with that fixed value instead of leaving the random node wired in.

Wired the wrong output into the wrong slot. The node hands back several types at once (int, float, boolean-ish, string list) - plugging the shuffled-list string into something expecting a plain integer, or vice versa, is a type mismatch, not a bug. Double-check which specific output you dragged the wire from.

Whole pack fails to import. WAS Node Suite has been unmaintained since the author marked it retired in December 2023; the recurring failure is the entire suite throwing "Import Failed" after a ComfyUI update from a pinned dependency clash. Reinstall requirements.txt against the correct, embedded Python interpreter and restart.