stop queue 停止绘图 (2lab)

ComfyUI gives you a red stop button in the corner, but it's manual - you have to be sitting there when things go sideways. StopQueue (2lab) is that same stop button, except it's a node you can wire into the graph and trigger from the workflow itself. Flip its stop input to true and the running queue gets aborted, no human required.

Why does that matter? Think about the batch pipelines and published-API workflows this pack is built for. If you're iterating over a list of inputs, some of them will be garbage - and you might know in the graph exactly which conditions mean "abort everything." A validation node upstream can set stop to true, and instead of burning ten minutes generating garbage, the queue dies on the spot. It's a guardrail you can express in the graph itself.

Inputs

• stop - a boolean. When it's true, the node triggers a queue stop. That's the entire input surface.

No outputs. It's a terminal action node - you set the flag, it does the thing, nothing comes out.

How it works

Mechanically it's a scripted version of the interrupt you'd trigger from the UI. When the node executes and stop is true, it asks ComfyUI's server to cancel the running job (and the rest of the queue). Wire it at the point in your graph where you want the check to happen - feed it from a comparison node, a list-length check, or anything else that can produce a boolean verdict about whether the run should continue.

Installing it

Standard for this pack. ComfyUI Manager → search "comfyUI-tool-2lab" → install, or:

cd ComfyUI/custom_nodes
git clone https://github.com/AI2lab/comfyUI-tool-2lab

Restart ComfyUI. No models, no extra dependencies.

Gotchas

• True means stop. The polarity trips people up: stop defaults to false, and the node only acts when the value is true. If you're thinking of it as a "continue" flag, you'll invert your logic and wonder why nothing ever stops.
• It stops the run, not the problem. The node cancels the queue, but the fix for the underlying condition still lives upstream in your graph. Use it to save compute, not to mask bugs.
• Repo visibility. The pack's GitHub page currently returns 404 - the author may have made it private or renamed it - so fresh clones can fail while existing installs keep working. It's a known state of the world, not a mistake in your setup steps.