BKWILDCARDS

A ComfyUI custom node that turns a bundled wildcard library into scoped dropdowns and emits a finished prompt string.

No wildcard syntax to learn. No __token__ to type. Pick a theme, choose the categories you want, wire one output into your prompt, and generate.

Ships with 5,277 hand-written entries across 54 categories and 8 themes — art styles, cybernetics, outfits, Halloween costumes, environments, poses, ancestry, species, hair, eyes, physical features, camera framing and more, all written for natural-language prompting.

Includes all current and updated version of my wilcard sets that are part of the BKWILDCARDS collection:

• All the Dresses

• All the Dresses, Eastern Edition

• _ghost.runner (cyberpunk)

• Whimsical Woods (dark fantasy)

• COZY SEXY LACY RACY Sleepwear (lingerie)

• Autumnal Oxidation (goth)

• Cassette Futurism (retro sci-fi)

• Nettie Necket (steam punk)

• Haunted Hallows (Halloween) 🆕

Install

Install via ComfyUI-Manager (recommended)

The easiest way. BKWILDCARDS is published on the Comfy Registry, so ComfyUI-Manager can install it for you; no git, no manual file copying.

1. Open Manager → Custom Nodes Manager.

2. Search for BKWILDCARDS.

3. Click Install.

4. Restart ComfyUI when prompted.

There are no dependencies to install as the node uses only the Python standard library.

Updating

• Via ComfyUI-Manager: open Manager → Custom Nodes Manager, find BKWILDCARDS, and click Update (or use Update All). Restart ComfyUI to load the new version. Manager pulls the latest release from the Comfy Registry.

• Via git (if you installed with git clone): run git pull inside the BKWILDCARDS folder, then restart ComfyUI.

The node title shows the build number (e.g. BKWILDCARDS Selector 0.9.16), so you can confirm the update took.

Manual install (git)

Git Repository: https://github.com/bkidderz/BKWILDCARDS

cd ComfyUI/custom_nodes
git clone https://github.com/bkidderz/BKWILDCARDS.git

Restart ComfyUI. There are no dependencies to install, the node uses only the Python standard library.

Manual install (ZIP)

1. Download this repository as a ZIP (Code → Download ZIP).

2. Extract it into ComfyUI/custom_nodes/.

3. Make sure the folder is named BKWILDCARDS and contains __init__.py directly inside it.

4. Restart ComfyUI.

Where is custom_nodes?

• ComfyUI Desktop (Windows)

  • %LOCALAPPDATA%\Comfy-Desktop\ComfyUI-Installs\ComfyUI\ComfyUI\custom_nodes

• Portable / manual

  • <your ComfyUI folder>\ComfyUI\custom_nodes

How to use

Add BKWILDCARDS Selector to your workflow (right-click → Add Node → BKWILDCARDS).

Important

You must wire the node's prompt output into a prompt block.

The node does not talk to your model on its own — it only produces text. Connect its prompt output to whatever feeds your positive prompt: a CLIPTextEncode node's text input, or a string node (such as StringFunction) that is already feeding your prompt.

The node tells you if you forgot: it turns red and displays NOT ready, please wire the node to your prompt block. Once wired, it turns green and reads ready — press Run.

Basic workflow

1. Pick a theme (top of the node) — e.g. _ghost.runner, Whimsical Woods, All the Dresses. Only that theme's categories are shown and can contribute.

2. Pick a gender — Female, Male, — random — (rolls per image), Fluid (both sets available), or — off —.

3. Choose your categories. Most are dropdowns with three kinds of choice:

   • — off — — this category contributes nothing

   • — random — — draw from the whole file

   • a named section (e.g. Netrunner / Decker, Interiors, Cybergoth) — draw only from that section

4. Wire prompt into your prompt block (see the callout above).

5. Press Run. The node draws one line from each active category and joins them into a single prompt.

Getting a new result each run

Set the seed widget's control to randomize for a fresh combination every queue, or fixed to lock a result you like. The seed fully determines the output — the same seed always reproduces the same prompt.

Reading what it produced

The output box at the bottom of the node fills in with the exact prompt, updating the moment you press Run — before the image renders. That same text is embedded in generated PNGs, so dragging a finished image back into ComfyUI restores the prompt that made it.

Features

Labeled output

By default the node tags each selection so the renderer reads structured attributes instead of one run-on sentence:

gender: a woman,
ancestry: korean, warm-toned East Asian features, monolid eyes, …,
hair: soft black hair, with loose open S-shaped waves, half-up …,
outfit: a sleeveless heavy canvas coverall unzipped low over …,
color palette: muted olive-drab and warm-tan accents …,
scene/background: an exterior landing pad, a boxy retrofuturist craft …,
pose: the subject holds one wrist up close to check a strapped readout …

Flip label_output to plain for a single comma-joined string instead.

🤪 Mayhem mode

Flip mayhem on in Settings for one-click chaos: the node ignores every selection, the theme and the gender, and composes a fully random cross-theme image, a cyberpunk outfit in a fantasy shrine with a goth pose, and so on. Queue again for a new one. It stays seed-deterministic, so any result you like can be reproduced or recovered from its PNG.

Live output preview

The output box updates at queue time, before generation starts, so you can see exactly what the run will use. The previewed text is guaranteed identical to what the image is generated from.