Prompt text input

Don't overthink this one. SeargePromptText is a multiline text field that outputs whatever you typed as a plain string. That's the entire node. You write a prompt, it hands the text downstream. No encoding, no CLIP, no tokens - just text in, text out.

Why does a node this simple exist? Because in the Searge workflow, prompts get routed and combined before they're encoded. Keeping the raw text in its own little node means you can wire the same prompt into several places, join it with other snippets, or reuse it across a graph without retyping. It's a plumbing convenience. Its category is Searge/_deprecated_/Prompting, so it belongs to the older building-block set; the current v4.x workflow generally handles prompt text inside its own UI nodes, but this still works.

How it works

There's one widget: a multiline string box. Whatever you type comes out the other side as a STRING. Anything that accepts a string input - Searge's prompt combiner, its prompt encoders, or standard ComfyUI text nodes - can take that output. Think of it as a labelled sticky note you can plug into the workflow.

The inputs and outputs that matter

• prompt - the multiline text box. Type your prompt here.
• prompt (output) - the same text as a STRING.

That's genuinely all there is. There are no hidden parameters to discover.

A quick word on SDXL prompting

Since you're here to write a prompt, one thing worth knowing: SDXL doesn't reward the wall-of-negatives habit people carried over from SD 1.5. Shorter negatives tend to work better. And the "right" prompting style is really set by your checkpoint, not the base model - Danbooru-style tags on Illustrious, score tags on Pony, short prose on realism finetunes. This node doesn't care what you type; it's just the box. But what you type is where the actual results come from.

How to install it

ComfyUI Manager: search SeargeSDXL, install, restart. Manual: python -m pip install opencv-python in ComfyUI's Python env first (a required dependency), then cd ComfyUI/custom_nodes && git clone https://github.com/SeargeDP/SeargeSDXL.git and restart ComfyUI. Windows portable users have an installer script that also fetches the workflow's models.

No model download for this node - it's pure text.

Where people get burned

• Expecting it to encode. This node does not turn text into conditioning. It outputs a string; a prompt encoder (SDXL Prompt Encoder, or ComfyUI's CLIP Text Encode) is what converts that into something the sampler can use. Wiring this straight into a sampler won't work.
• Reaching for it in a fresh graph. It's deprecated. If you're building new, just use the text fields in the current workflow's prompt nodes rather than a separate text box.
• Type mismatches. Its output is a plain STRING, not CONDITIONING. If a socket refuses the connection, that's the type system telling you an encoder needs to go in between.