💤 Isulion Shutdown

The 💤 IsulionShutdown node does exactly what it says: when the workflow executes it, it shuts down your computer. There's no confirmation dialog, no "are you sure," no undo. The pack author knows how this sounds - the moment the node loads, ComfyUI's console prints a big yellow SECURITY WARNING telling you to only use it in trusted workflows and that you can delete shutdown_node.py to disable it.

So why would anyone want this? Because long render jobs happen at night. Queue up 300 images before bed, and instead of leaving a roaring GPU rig running until you wake up, you run one workflow that finishes, hits this node, and powers down. That's the entire use case, and for that one job it's perfect.

How it works

It's a couple of lines of os.system under the hood. On Windows it runs shutdown /s /f /t <delay>. On Linux/macOS it runs shutdown -h +<minutes>. That's it - no API, no package, nothing to configure beyond the delay.

The inputs that matter

• any - a wildcard * input. This is the trigger. You wire the tail of your workflow into it (a final image, a string, whatever) so the node only fires after everything upstream has finished. Because ComfyUI runs nodes when their inputs are ready, this is your "run me last" guarantee.
• delay - seconds before shutdown, default 30, minimum 0. On Windows it's exact. On Linux the value is converted to minutes (delay / 60, rounded, minimum 1) - so a delay of 30 becomes one minute, and a delay of 0 still becomes one minute. That's a genuine cross-platform gotcha worth knowing before you set 30 and expect 30 seconds.

Output

None. It's an output/terminal node; you don't wire anything out of it.

Installing it

One pack, all Isulion nodes:

1. ComfyUI Manager → Custom Nodes Manager → search Isulion → Install → Restart.
2. Or clone it manually:

cd "your_ComfyUI_install_dir/custom_nodes"
git clone https://github.com/Isulion/ComfyUI_Isulion

No requirements.txt, no models. And if the idea of a shutdown node in your ComfyUI makes you uneasy, you're not wrong to be cautious - the source itself suggests removing Core_Nodes/shutdown_node.py if you don't want it loaded.

Common issues

The scary one first: this fires the moment its inputs are ready, and it fires on every queue of the workflow. Forget to wire the any input and it can shut down mid-batch, because a node with a ready input runs as soon as the graph reaches it - actually no, it waits for its input, but if you leave any unconnected and queue, it will fire immediately. Wire it strictly last. Also: no confirmation, so test with a delay you can live with, or test with the workflow's sampler muted. On Linux, remember the seconds-to-minutes rounding. And if you're on a machine with other people's jobs running - a shared box, a server - this node will happily shut it down without asking. It's a niche tool for a solo render-rig use case, and it's excellent at exactly that.