LoRA Stack Info

LoRA Stack Info (class AtomLoRAStackInspector) is the debugging node of the Atom_AI pack: it takes a LORA_STACK and prints what's in it as text. No model, no clip, no fusion - it just reads the list and tells you, entry by entry, which LoRA is there and at what model and clip strength.

It's the node you reach for when a stack applied but the result looks wrong, or when you inherited a workflow and want to know what someone actually wired up before you start touching it.

Why it exists

Here's the thing about the stack system: the list is invisible. A chain of LoRA Stack nodes passes a growing blob of data along, and by the time you're three entries deep you can't see at a glance whether entry two is set to strength 2.5 or whether it's even enabled. That's what this node is for - a mirror on a data type that has no visual of its own.

It's marked as an output node in the code, which in ComfyUI terms means its output text shows up directly in the UI when you run the graph. So you don't need to wire anything after it. You can also leave it dangling off the side of your stack (it doesn't consume or alter the stack) and it'll report on every run.

Inputs and outputs

One input, one output:

• lora_stack - the stack from your chain of LoRA Stack entries.
• stack_info (STRING) - a numbered list: position, LoRA filename, model strength, and clip strength for each entry. An empty stack returns "Stack is empty."

That's the entire node. There are no knobs, which is the point - you want the truth about what's in the pipe, not an opinion.

Installing it

Part of the Atom_AI pack, so install once: ComfyUI Manager, search Atom_AI, install, restart. Or manually:

cd ComfyUI/custom_nodes
git clone https://github.com/AtomAI-GR/Atom_AI

Restart ComfyUI after cloning. No extra dependencies, no model downloads. It lives under Atom_AI ▸ LoRA.

When you'd actually use it

• "The stack did nothing." Read the report: if the entry you thought was active shows a strength of 0 or the file's missing, there's your answer. A LORA_STACK is just data until LoRA Stack Apply fuses it, so confirming what survived the journey to that node is the first diagnostic step.
• "It applied but looks off." Check the strengths and order in the report against what you intended - inverted or doubled strengths are a classic slip that this node exposes instantly.
• Porting someone's workflow. A chain of stack entries tells you what's there, but the inspector turns it into something you can actually read and reason about.

Honest verdict: it's a tiny utility and it knows it. If you only ever run one LoRA, you'll never need it. But the moment you're chaining several LoRAs and chasing a mystery, a read-only view of the data costs nothing to add and saves the "which one of these four is the problem" guessing game. For that job it's genuinely the handiest node in the pack - the whole stack system would be harder to trust without it.