RH Node Info List

This one confuses people because it looks like it should do something and it doesn't. RH_NodeInfoListNode doesn't generate an image, call an API, or need a key. It's a little clipboard. You use it to say "in the cloud workflow I'm about to run, set node 12's prompt to this and node 8's seed to that" - and then you hand that list to the node that actually does the work.

To make sense of it you need the pack it lives in. ComfyUI_RH_APICall is a client for RunningHub, a cloud ComfyUI service where you build and save a workflow on their servers. This plugin lets your local ComfyUI fire off one of those saved cloud workflows and pull the results back - images, video, latents, audio, text. RunningHub is a real and fast-growing platform (community mentions roughly 30x'd across 2025), so if you've landed here you're probably wiring your local graph up to run something on their GPUs instead of yours.

What it's actually for

Here's the problem it solves. Your cloud workflow is fixed - it lives on RunningHub with a specific set of nodes. But you don't want to run it with the same prompt and seed every time. You want to drive it from your local machine, swap the prompt, bump the seed, change the batch count. That's what this node does: it builds a list of overrides that get applied to the remote workflow at run time. The author's own words are that you use NodeInfoList to modify the value of each node in the RH workflow - "prompts, seeds, generation batches, etc."

The mechanism is a linked list, which is why the node feels weird in isolation. Each RH_NodeInfoListNode describes exactly one override. To override three fields you chain three of these nodes together, each feeding the next, and the final one hands a complete ARRAY to the main RunningHub execution node.

The inputs and outputs that matter

There are only four sockets and you'll use all of them:

• nodeId (INT) - the ID of the target node inside the RunningHub workflow, not your local graph. This is the part beginners miss. You get these IDs from your workflow on RunningHub's side.
• fieldName (STRING) - which field on that node to overwrite, e.g. text, seed, steps, batch_size. It has to match the field the cloud node actually exposes.
• fieldValue (STRING) - the new value. Note it's a string even when you're setting a number like a seed or step count; you type it as text and RunningHub coerces it on their end.
• previousNodeInfoList (ARRAY, optional) - plug the output of the previous NodeInfoList node in here to chain them. Leave it empty on the first one.

The single output is ARRAY - the accumulated list of overrides. Wire it into previousNodeInfoList of the next override in the chain, and the last node's ARRAY goes into the main RunningHub API-call node that runs the workflow.

So a typical setup is: NodeInfoList (set prompt) → NodeInfoList (set seed) → NodeInfoList (set batch) → the execute node. Order of the chain doesn't matter; the pairing of nodeId + fieldName is what targets each override.

Installing it

Two ways, both painless because there are no models to download and no heavy Python deps - the generation happens on RunningHub, not your box:

• ComfyUI Manager: open the Manager, search for ComfyUI_RH_APICall, install, restart.
• Manual: cd ComfyUI/custom_nodes && git clone https://github.com/HM-RunningHub/ComfyUI_RH_APICall, then restart ComfyUI.

Then register at runninghub.cn and grab an API key. Important: the key does not go on this node - it goes on the main execution node. RH_NodeInfoListNode stays inert and keyless. There are example workflows in the pack's examples/ folder; start from those rather than wiring it from scratch.

Where people get stuck

The usual failures are all targeting mistakes, not bugs. If an override silently doesn't apply, it's almost always the wrong nodeId (you used your local node's ID instead of the RunningHub one) or a fieldName that doesn't match the cloud node's actual field. When something's off, verify both against the workflow as it exists on RunningHub - that's the source of truth, not your local canvas.

One broader caution worth keeping in mind: this pack sends your prompts and inputs to a third-party server and runs them there, so it's a very different trust and privacy posture than staying fully local. The ComfyUI custom-node ecosystem has a real history of nodes doing more than they advertise, so if you're pasting an API key, know you're trusting RunningHub with whatever you send. Nothing shady here - just the honest cost of offloading to someone else's GPUs.