让我看看丨纯净

ComfyUI is great at hiding what's actually flowing through a wire. You see the cable, not the data. ShowMePure exists to fix exactly that: wire any output into its source input, run the graph, and the raw value appears right there in the node's widget - no formatting, no analysis, no ceremony.

The Chinese display name is "让我看看丨纯净" - "let me take a look, pure version." The "pure" is the point. Its sibling in the same pack, LetMeSee, gives you the forensic report: data type, run duration, system resources, the works. ShowMePure is the minimalist alternative: it just shows you the content and gets out of the way.

How it works

ShowMePure is a pass-through debug node. It takes one optional any-type input called source (declared as type *, the ComfyUI wildcard that accepts anything), and it returns that same source untouched on the output. In between, it formats the value for display - a string stays as-is, numbers and booleans get stringified, lists get printed, and anything else gets dumped as JSON. Then it writes that text into its own widget on the canvas via the workflow metadata, so the content you saw on the last run is sitting there staring at you.

Because the output is a literal passthrough of the input, you can splice the node into an existing connection without breaking the chain. That's the whole trick of the plumbing layer these utility nodes live in: a debug viewer that changes nothing downstream is one you can leave in place while you work (comfyui-node-plumbing.md is the full taxonomy of this layer).

The input and output that matter

• source (optional, any type) - plug in whatever you're curious about. It's the only input.
• source (output, any type) - the exact same value, passed through. Wire this onward if you spliced the node into a chain, or just leave it dangling.

It's marked as an output node (OUTPUT_NODE = True), which matters: ComfyUI treats output nodes as things that always run, so the widget reliably updates on each execution.

Installing it

ShowMePure ships in ComfyUI-QING, a 79-node all-in-one pack by a solo developer, GAO-SHIQING. Install the pack once and you get this node plus its siblings:

cd ComfyUI/custom_nodes
git clone https://github.com/GAO-SHIQING/ComfyUI-QING
cd ComfyUI-QING
python install_dependencies.py

Restart ComfyUI after installing. Or use ComfyUI Manager and search for "ComfyUI-QING" - it's a one-click install there. The dependencies (Pillow, opencv-python, scipy, scikit-image, cairosvg, openai) are heavier than a single debug node justifies, but you're paying that tax for the whole pack.

Common issues

• Nothing shows in the widget. The node only displays what ran. If it's not connected to anything that executes (or the graph was loaded fresh and never run), the widget is empty. Run the workflow once.
• The value is a blob you can't read. That's normal for this node - it's deliberately unformatted. If you need structure, that's what LetMeSee is for.
• Fresh pack caution. QING has essentially no English-speaking community footprint yet, and the debug-category nodes are exactly the shape of thing that gets installed without reading. The source is a thin, readable wrapper around print logic, but the general rule stands: with any brand-new pack, glance at what it does before trusting it with anything sensitive (the ecosystem's one big malware incident happened in this exact category, per comfyui-ecosystem.md).

When you just need to know what a value actually is, ShowMePure is the least-fuss answer in the pack. One wire in, one glance at the canvas, done.