FL System Check

Drop this node in, run it, and it tells you what you're actually running on: Python version, OS, CPU, RAM, GPU, and the versions of the libraries that matter (PyTorch, transformers, and friends). No inputs, no outputs - it's a diagnostics readout. When something's broken and you need to know your environment before you can even start guessing why, this is the node that gives you the numbers.

That sounds mundane until you remember what daily life in ComfyUI is actually like. The single most common category of pain in this ecosystem is dependency and version hell - custom nodes each drag in their own Python requirements, and one pack wanting transformers==4.30 while another wants >=4.40 is how a working setup quietly breaks. When you're debugging one of those, or filing a bug report, or following a workflow that "only works on torch 2.x," the first question anyone asks is what versions do you have? FL_SystemCheck answers it without you hunting through pip and nvidia-smi.

How it works

It gathers system and environment information - interpreter version, platform details, hardware, installed library versions, relevant environment variables - and exposes it for display (it also serves the info via an HTTP endpoint). It's read-only: it inspects, it reports, it changes nothing.

The inputs and outputs

There are no inputs and no outputs - it's an output/display node. You add it to the graph, run, and read what it shows. That's the entire interface, and that's fine: its whole job is to surface facts, not to feed data anywhere.

How to install it

Part of filliptm's Fill-Nodes pack. ComfyUI Manager → search ComfyUI_Fill-Nodes → Install → restart, or:

cd ComfyUI/custom_nodes
git clone https://github.com/filliptm/ComfyUI_Fill-Nodes
pip install -r ComfyUI_Fill-Nodes/requirements.txt

No model, nothing extra.

Common issues & troubleshooting

It reports the server's environment, not your laptop's. If you're using a hosted or remote ComfyUI, the numbers describe the machine ComfyUI is actually running on - the server - which is usually exactly what you want when debugging that instance, but it's not your local box. Keep that straight when you're comparing setups.

It tells you what's wrong, it doesn't fix it. This is a thermometer, not medicine. If it shows a torch/CUDA mismatch or an unexpectedly old library, the node's job is done - the fix is on you (reinstall the right versions into ComfyUI's Python environment, resolve the conflicting requirement). Its value is turning "something's off" into a specific, paste-able list of versions.

Use it before you ask for help. When a workflow won't run and you're about to post in a Discord or open an issue, run this first and include the output. Half of dependency debugging is establishing the baseline, and this collapses that step into one node.