I Made The Most Advanced Auto-Prompt Node In ComfyUI (Patreon)

I stopped typing prompts by hand months ago and I'm not going back. This node pack is why. Two nodes, AI Image Prompt Builder and AI Prompt Library, and together they cover everything from character setup to final saved prompt library. This is the full breakdown of what's in the pack and how to actually use it. Check out the Youtube video breakdown as well for a live demo:

Watch Now

Character & LoRA Panel

Sits at the top of the Builder node no matter which mode you're in. Trigger word, LoRA picker, model and CLIP strength, all in one place. The LoRA picker is a searchable dropdown, type to filter, and it scrolls clean no matter how many LoRAs you've got installed.

The trigger word field auto-prepends to your prompt the moment a LoRA is selected. That's the actual invocation of the trained concept, separate from the strength sliders entirely.

Saving Characters

Once you've got a trigger word, LoRA, and strengths dialed in, save the whole combo as a named character preset with its own reference thumbnail. Load it from the Saved dropdown any time and every prompt from that point carries the setup automatically. No re-entering the same values every session.

Every saved character lives in its own Characters tab inside the main category bar, right alongside the rest of the categories. Each one shows up as a card with its reference thumbnail, the character name, and the LoRA it's tied to right underneath. Click the card and everything, trigger word, LoRA, strengths, loads in at once. Building out a new one is just as fast, hit the New Character card, attach a LoRA, and set a thumbnail so you can identify it at a glance instead of hunting through names.

Two Modes, Built for Different Work

One toggle, two completely different category sets. Professional/Business mode gives you 9 broad categories: Medium, Subject, Clothing, Pose, Background, Camera, Lighting, Processing, Mood. Good for commercial, editorial, straightforward portrait work.

Influencer mode restructures around physical presentation. Clothing splits into 7 sub-categories instead of one flat list. Backgrounds cover 75+ real locations across 8 sub-categories. Body, Face, Hair, and Makeup each get their own category, 43 separate physical trait fields total.

They're split instead of combined because commercial work doesn't need 43 physical trait sliders, and influencer content needs way more granularity on appearance and setting than a single Subject category could ever hold. Every category, built in or custom, works the same way: thumbnail grid, click to select, right-click to customize.

Full Customization

Right-click anything, thumbnail card, text-only option, physical trait field, and you get Add/Change Thumbnail, Edit label, Delete, and Drag-to-reorder. Delete is fully reversible even on built-in defaults. Add entirely new categories and subcategories from the + tab, and they slot straight into the Randomizer's Custom Preset checklist automatically. The Manage Custom Content modal is where you see and prune everything you've added so a big personal library doesn't turn into tab chaos.

AI Prompt Library

The companion node. Saves, organizes, and reuses prompts whether they came from the Builder or you typed them yourself. Connect the Builder's output into the Library, queue once, and the resolved prompt lands in an editable text box, so you can tweak it before saving or before it feeds onward without going back through the Builder.

Save with an optional name, drop it into a folder if you want, hit Save to Library. Folders can be organized by character, project, client, whatever fits your workflow. Loading pulls it back into the preview box for review or another tweak. Editing a saved prompt is just opening it, editing the text, and saving over the original.

Why This Matters

Character and LoRA setup, two purpose built prompt modes, full customization, and a library to save your best work. All in two nodes. You will save yourself hours and hours of writing your own prompts, or save yourself MILLIONS of precious tokens to your favorite LLM by having this node handle it.

Bug Reports/Fixes

If you run into anything weird, a node throwing an error, a category not loading right, thumbnails not saving, whatever it is, report it directly to me here on the Patreon post or in the Discord. I'm actively fixing things as they come up, so the faster I know about it the faster it gets patched.

2026-07-25 — Bug Fix: Popups for saving/deleting characters, thumbnails, folders, and prompts have been rebuilt as native in-node popups instead of the browser's own, fixing cases where typed names or confirmations wouldn't register. Re install to fix.

Install

Download LoRAtech-advanced-prompt-builder.zip from this post.

Extract it into your ComfyUI/custom_nodes/ folder so the extracted folder sits directly inside custom_nodes, not nested another level deep.

Restart ComfyUI (or use Manager's restart if you're running it that way).

Both nodes will show up under the LoRAtech category when you right-click to add a node, or search "AI Image Prompt Builder" and "AI Prompt Library" directly.

If either node fails to load, check your console for a missing dependency error and drop it in the Discord, I'll help you sort it.

This one's for Patreon members first. Full workflow build using it is coming to YouTube soon.