Noodle Soup Term Edit

Noodle Soup Prompts is WAS Node Suite's wildcard system - a shared pantry of roughly 17,500 words across 84 terminologies. Type __animals__ in a prompt and the parse node swaps in a random entry from that list, which is how you get a "one prompt, endless variations" workflow without retyping anything. Noodle Soup Term Edit is the node that lets you take that pantry apart and put your own words in.

It does four things: adds words to an existing terminology, removes words from one, replaces a terminology's contents outright, or creates a brand-new one that only you have. Name a term the pantry doesn't know, hit create, and __your-term__ becomes a real draw from your own list everywhere the parser runs.

The inputs that matter

Only three, which makes this one of the friendlier nodes in the pack:

• term - the terminology name without the __ markers. my-animals is what a prompt draws as __my-animals__. A name the pantry doesn't have yet gets created.
• action - add entries, remove entries, replace entries, create the term, or delete the term. Add stores words the term doesn't already hold; remove takes them out and no refresh puts them back; replace leaves the term holding exactly what you typed.
• entries - the words, one per line. Blank lines are skipped. Note that a line is one entry, so red fox asleep in long grass is a single multi-word phrase - exactly what you want a wildcard to be able to draw.

What you get back

noodle is the term wrapped in markers - __my-animals__ - ready to paste into a prompt for Text Parse Noodle Soup Prompts or Prompt Parse. term is the stored name, useful for chaining into another edit or a pantry export. entries is every entry the term holds after the edit, in draw order (Text List to Text turns it into one line per entry), and entry_count is the running total.

The part people worry about, answered

The tooltip's reassurance is worth quoting: entries you add are kept apart from the ones the published pantry supplied, and Noodle Soup Pantry Refresh never removes them. The refresh node only fetches and merges the published pantry; your additions survive it, your removals don't come back, and terminologies you created are untouched. So editing the shared pantry is safe - the node keeps your changes in a separate layer (stored in the pack's was_state.db) rather than mutating the published data.

One delete is permanent within that stored state: delete the term removes it and everything in it, and its outputs come back empty.

Where the pantry lives and how to install

The whole terminology system is stored in <ComfyUI user dir>/was-node-suite/was_state.db alongside the style library - a SQLite file that two ComfyUI instances pointed at the same config directory can share. You can move a pantry between machines by exporting it as JSON with Noodle Soup Pantry Export/Import.

Noodle Soup Term Edit ships in WAS Node Suite v3:

cd ComfyUI/custom_nodes
git clone https://github.com/WASasquatch/was-node-suite-comfyui

or ComfyUI Manager → search WAS Node Suite v3, then restart. Needs ComfyUI 0.14.0+, Python 3.10+, and nothing else installs - v3's whole point was shedding the old pack's twenty-package dependency load. If you want your custom terms to persist a clean reinstall, back up was_state.db (or export the pantry) before wiping the pack's config directory.

The workflow to reach for: a seed list of subjects you actually generate - characters, moods, settings - saved as a terminology, then __subjects__, __moods__ in every prompt. That's prompt variation with a vocabulary you control, which is the entire reason the wildcard pattern is worth adopting.