🔶 Simple Prompt Text

This is the plainest node in the whole pack, and there isn't much mystery to it: two text boxes go in, one string comes out. What makes it worth having its own node instead of just typing everything into a single CLIPTextEncode is that the two boxes stay separate on the graph, so you can keep a "base" prompt and a "variable" add-on visually distinct - and, more usefully, wire something else into that second box later.

What it does

Prompt is a required multiline text field - your main prompt. add_prompt is an optional multiline field that gets appended onto it. The single output, PROMPT_TXT, is a plain STRING: the combined text, ready to feed into whatever actually turns it into conditioning (this node doesn't touch CLIP at all - it's pure text assembly, not encoding).

That split between the two fields is the actual point. Keep your subject and style description in Prompt, and use add_prompt for the part you expect to change - a scene detail, a mood word, a tag pulled from somewhere else in the graph. Because it's a separate input rather than one big text blob, you can swap out just that piece without touching the base prompt, or eventually replace the widget with a wired connection from another string-producing node if your workflow calls for it.

Why you'd reach for it

If you're just writing one prompt for one image, you don't need this - a stock CLIPTextEncode does the same job with fewer moving parts. Where it's genuinely handy is workflow organization: splitting "the prompt I don't want to keep retyping" from "the bit I'm actively iterating on" keeps your canvas readable, especially in the checkpoint- and prompt-changing animation setups this pack is built around, where you might have several of these stacked and only want to tweak the add-on text between runs.

Installing it

Through ComfyUI Manager: search "Chaosaiart-Nodes," or use "Install via Git URL" with https://github.com/chaosaiart/Chaosaiart-Nodes, then restart ComfyUI. Or manually: cd ComfyUI/custom_nodes && git clone https://github.com/chaosaiart/Chaosaiart-Nodes, restart. Linux users also need pip install opencv-python and pip install tqdm (with your venv active, if you use one) for the pack as a whole - this particular node doesn't need either, but skipping the step breaks the pack's import on load since it's all one Python module. Windows has an Install_windows script in the folder. Nothing to download model-wise.

Troubleshooting

There's not much that goes wrong with a text concatenation node. The one thing to remember: PROMPT_TXT is a plain STRING, not CONDITIONING - you still need a CLIPTextEncode (or Chaosaiart's own checkpoint-plus-prompt node) somewhere downstream to actually turn this text into something a KSampler can use. If your prompt isn't showing up in the generated image the way you expect, double-check you didn't leave add_prompt empty when you meant to fill it, or that you're not accidentally duplicating text between the two fields - since it's a straight append with no deduplication, repeating a phrase in both boxes just repeats it in the output.