Show Any

ComfyUI is a graph you mostly trust until something goes wrong, and then the whole debugging experience is staring at wires. Show Any exists for those moments: plug anything into it - a string, an int, a latent tensor, a conditioning tuple, a dictionary - and it prints what that value actually is as readable text in the node's UI, while passing the original value through untouched on its output. It's a logging node that doesn't get in the way.

Why you'd reach for it: a downstream node is failing with a type error and you're not sure whether you're feeding it a list, a tuple, a tensor, or an empty something. Wire Show Any in front of the failing node, queue once, and the answer is right there in plain text. It's the ComfyUI equivalent of a print() you never have to remove because it doesn't alter the data.

How it works

The trick is in the type system: the anything input is a match-type wildcard, so ComfyUI lets you connect literally anything to it. Inside, the node walks the value and builds a display string by type - strings print as themselves, numbers as numbers, lists and dicts as their Python repr, and torch tensors print with their values truncated (edge items shown) so you can see the shape and a hint of the data without drowning in a 4D wall of numbers. Conditioning, which is famously a tuple of tensor-plus-dict, gets special handling: it prints the tuple length and the shape of each tensor, which is precisely the information you need when a sampler is rejecting your conditioning. If serialization fails entirely, you get the type name as a fallback, which still tells you something.

The output is a list with is_list semantics, and it carries the original values through unmodified - so the node is safe to leave in place as a permanent tap on a wire without changing behavior.

The one honest limitation: this is a "what is this" tool, not a "look at this image" tool. For a tensor's visual content you want a preview node; Show Any is for the structure and values.

Where it fits

ComfyUI is ~70% plumbing, and debugging plumbing is mostly answering "what came out of that node." Show Any is the answer. It's the plainest member of a family in this pack - Show Text is the same idea with a typed STRING output, and Show Any Stop / Show Text Stop add a review-break. For most "what is on this wire" questions, Show Any is the one.

Installing it

Ships with ComfyUI_Eclipse. ComfyUI Manager → search "ComfyUI_Eclipse" → Install, restart. Or:

git clone https://github.com/r-vage/ComfyUI_Eclipse custom_nodes/ComfyUI_Eclipse

No models, no extra Python deps beyond what ComfyUI already has (torch/numpy, which it checks for gracefully).

Gotchas

Nothing exotic. The display text is just UI feedback - it's not written anywhere persistent, so don't expect it to survive a workflow reload (use Show Text if you want the value embedded in the saved graph). And since it accepts anything, a truly enormous list or tensor can make the node's text area unwieldy; the edge-items truncation on tensors keeps that mostly in check. If you came to Eclipse from the older RvTools pack, note that old node IDs were removed in v4.0.0 - fresh workflows are unaffected.