Ricky's Prompt Mixer (5-Input)

Once you build prompts from parts - a base prompt loaded from one file, a character description from another, a stack of style tags from a third - you hit a classic annoyance: when one fragment is empty, a naive concatenator hands you , , or trailing commas that quietly pollute your output. SimpleTextMixer exists to fix exactly that. It takes up to five text inputs, joins them with a separator you choose, and silently drops anything that's empty. The name "Ricky's Prompt Mixer (5-Input)" is accurate about both the person and the cap.

It's the second node in the small Rathius_ComfyNodes pack, alongside LoadTextFile_Rathius. The two are designed to be used together: load prompt fragments from files, mix them here, feed the result into a CLIP Text Encode. Like its sibling, it's pure Python with zero external dependencies - no requirements.txt, no models, nothing to download. It's a hobby pack, so don't expect an active issue tracker or rapid updates; the code is short enough that it mostly just works.

How it works

The mechanism is deliberately simple. The node collects its five inputs into a list, filters out any string that's empty or only whitespace, then joins what's left with your separator and returns it as a single string. Nothing clever - which is the point. It's the kind of glue node you could write yourself in ten minutes and never want to again.

The inputs and outputs that matter

• separator (required) - the string inserted between fragments. Default is ", ", which is right for SD-style tag lists. Set it to "\n" for line breaks or " " for plain spaces.
• text_1 through text_5 (optional) - the fragments to combine. Here's the catch that trips people up: all five are marked forceInput, meaning they're wire-only ports. You can't type a fragment directly into this node; every input has to come from something upstream. If you want to hand-type a segment, feed it from a plain text node first.

The single output is mixed_text (STRING). Wire it into the positive or negative prompt of a CLIP Text Encode, or into anything else that wants a string.

Installing it

Same pack, same story: ComfyUI Manager → search Rathius_ComfyNodes → install → restart. Or manually:

cd ComfyUI/custom_nodes
git clone https://github.com/Rathius-Saranoth/Rathius_ComfyNodes
# restart ComfyUI

No dependencies beyond what ComfyUI already ships. You'll find it under Ricky Tools in the node menu - not the same RathiusNodes/Utils category as the pack's other node. The pack's naming is a little all over the place; that's the hobby-pack tax.

Where people get burned

• The inputs are all wire-only. The most common complaint pattern is expecting to type into text_1. You can't - drag in a text node (or one of the pack's file loaders) and wire it in.
• Five is a hard cap. Need six fragments? Chain two mixers: feed the first one's mixed_text into the second's text_1.
• All-empty input returns an empty string. The empty-filtering is convenient, but it means a mixer with nothing wired to it won't error - it just emits "". As a negative prompt that's harmless; as a positive prompt you'll get whatever the model does with no instruction, which is rarely what you wanted.

If your prompts come from files and fragments, pairing this with LoadTextFile_Rathius gives you a tidy prompt-building section with no copy-paste and no comma rot.