String Text

String_Text is the pack's bare-bones text literal node: one multiline box, one STRING output, nothing else. It exists because ComfyUI is stubbornly bad at handing you a plain string. You can convert a node to primitive input, sure, but when you want an explicit, typed text source that clearly feeds an LLM's prompt input, a dedicated node like this is clearer and keeps the graph readable.

It's the dullest node in the pilcothink pack and that's fine - every workflow needs a place to type the question you're asking the model. In the pack's own architecture it's the natural companion to the generators: type a prompt here, wire it into the SLM generator's user_prompt or the Ollama generator's prompt, and you have a self-documenting text source instead of a hidden primitive.

How it works

There's no processing to speak of. The text input is a multiline STRING field; the node returns exactly what you typed, unmodified, as its STRING output. One input, one output, zero surprises.

The input that matters

• text - the multiline field with your string. Default empty.

That's the whole schema. The single STRING output plugs into any text input in the graph.

Install

This is part of the comfyui_pilcothink_VisionSLM pack, so you get it alongside the vision and Ollama nodes:

cd ComfyUI/custom_nodes
git clone https://github.com/gpdev-Pilcothink/comfyui_pilcothink_VisionSLM

Then restart ComfyUI. Or use ComfyUI Manager and search for comfyui_pilcothink_VisionSLM. Fair warning: the pack's requirements.txt is heavy - transformers, accelerate, faiss-cpu, sentence-transformers, opencv, and friends. If you only wanted a string node, this is a lot of machinery to drag in for it; you'd be better served by a dedicated text-utility pack. If you're here for the LLM nodes anyway, you won't notice.

Gotchas

Keep expectations calibrated: this is a passthrough with a multiline box. ComfyUI already has primitive nodes for string input, and several text-utility packs give you the same thing with extras (case conversion, variable substitution). Reach for String_Text because it's in this pack and keeps your LLM workflow self-contained - not because it does anything fancy. And as with any fresh custom node, install from the official repo. The pack is new, it has no community review behind it yet, and given the ecosystem's history with LLM-vision nodes specifically, that's exactly the category worth being careful about.