LZ Pipe Info

Pipes are great until they aren't. An LZ pipe bundles your model, CLIP, VAE, conditionings, latent, prompt texts, seed, and hashes into a single connection - and the moment an image comes out wrong, you're staring at a bundle wondering what's actually in it. LZ Pipe Info is the answer: it dumps the pipe's contents to the UI as a readable report, and passes the pipe through untouched so you can park it in the middle of a chain as a debug probe.

How it works

Plug an lz_pipe in and it builds a text report of every field inside: tensors render as <tensor (shape)>, nested dicts as summaries, short strings in full, long ones truncated to 200 chars, and internal keys (those starting with _) skipped so the noise stays low. The report comes out on the info_text STRING output, which you can also route anywhere you'd send text.

Two nice touches. There's an optional text input that prints as a section header (=== your note ===) at the top of the report - so when you have three debug probes in a graph you can label which is which instead of guessing. And because info_text is a real output, you can save it, log it, or pipe it into a Text Preview node.

The outputs are lz_pipe (identical passthrough - nothing is mutated) and info_text. The node is marked as an output node, so its report shows up in the UI even mid-graph.

When you'd use it

Mostly debugging. The classic case: you wired EZCheckpointLoader → LZKSamplerDecode, the image is wrong, and you suspect the pipe is carrying stale data from an earlier branch. Park an LZPipeInfo between them and read off exactly which ckpt, seed, and prompt text are travelling. It's also the friendliest way to learn the pipe format - run one after every pack node you meet and you'll quickly internalize what each one stuffs into the bundle.

The one genuine alternative use: since it's an output node that shows text, some people use it as a pipe-aware note/comment node to document what a chain is supposed to carry. Workable, though a plain Note node is lighter.

Installing it

cd ComfyUI/custom_nodes
git clone https://github.com/liz-ils/ComfyUI-LZNodes

restart → MyCustomNodes/Pipe. ComfyUI Manager: search ComfyUI-LZNodes. No extra dependencies.

The honest take

It's a debug utility, not a workflow staple - you won't leave it in every graph. But when a pipe chain misbehaves, it's the difference between staring at an opaque bundle and reading a one-screen summary of what's wrong. In a pack whose whole bet is "pipes simplify your graph," a tool that makes pipes inspectable is the missing half of that bet. Keep one handy.