Environment_INFO(mki-环境信息显示)

Ever tried to help someone debug a ComfyUI problem and gotten "it just doesn't work"? The first thing you actually need is their environment: OS, Python version, GPU, driver, ComfyUI version, which packages are installed. Environment_INFO_makki (display name "Environment_INFO(mki-环境信息显示)") dumps exactly that report straight into a node in your graph.

It's a diagnostics node. You run it once, copy the output, paste it into a GitHub issue or a Discord help thread, and the person on the other end can actually see what they're dealing with. That alone makes it worth having around when you're on both ends of support requests.

How it works

The node gathers the report using a mix of platform, psutil, nvidia-smi (if present), and ComfyUI's own version module, then renders it as a text block you can read on the node itself - the pack's JS puts it in a scrollable text widget, so you can grab it with Ctrl+C instead of chasing console output.

Six boolean toggles control what's included, all on by default:

• SYSTEM_INFO - OS, platform, processor, Python version, ComfyUI version, date.
• HARDWARE_INFO - CPU cores, CPU usage, RAM total/available/used.
• GPU_INFO - from nvidia-smi: GPU name, driver, VRAM total/used, utilization.
• DEEP_LEARNING_FRAMEWORKS_INFO - the DL stack (torch, CUDA, etc.).
• ALL_INSTALLED_PACKAGES_INFO - the full pip package list, which is what makes "conflicting dependency" bugs visible.
• CUSTOM_NODES_FOLDERS_INFO - your installed custom node folders.

Outputs are INFO (the report as a STRING) and anything - an optional any-type passthrough so you can chain this node into an existing wire without breaking the graph. The report also lands in the UI widget, so you don't even need to wire the STRING anywhere to read it.

The bits that matter

• GPU_INFO is the one people actually need. Driver version + VRAM + utilization explains most "out of memory" or "torch not using my GPU" reports in one line.
• ALL_INSTALLED_PACKAGES_INFO is the other half of the debugging story - the ecosystem essay in the KB makes the point that most custom-node pain is conflicting dependencies, and this list is how you prove it.
• psutil and comfyui_version are assumed present. ComfyUI ships both, so on a normal install it just works. If you're on a stripped-down environment, pip install psutil fixes the rare failure.

Installing

It's part of ComfyUI-MakkiTools:

cd ComfyUI/custom_nodes
git clone https://github.com/MakkiShizu/ComfyUI-MakkiTools
cd ComfyUI-MakkiTools
pip install -r requirements.txt

Restart ComfyUI. One caution: the full package list can be long, and with nvidia-smi absent (AMD-only or headless setups) the GPU section quietly drops rather than erroring. Don't read a missing GPU line as a problem - read it as "this machine has no NVIDIA tooling visible."

It won't fix anything by itself, and you won't use it daily. But the next time you're copy-pasting "what's your environment" into a support thread, this node turns a two-hour back-and-forth into one paste.