Recipe Manager

Prompt Manager is for saving prompts. Recipe Manager (class RecipeManager) is for saving recipes - the whole generation configuration: prompts, LoRAs, sampler settings, resolution, seed, the works. It's the Prompt Manager sibling stripped down to the one thing the recipe pipeline needs: a save/load library for RECIPE_DATA payloads. No use-switches, no trigger words, no LoRA stack sockets - the author calls it "workflow-data focused," and it shows.

How it works

Like the Prompt Manager node, it's organized around category and name dropdowns. Pick an entry and the saved recipe data flows out the single recipe_data output; pick a blank name, and the current state flows in to be saved. The difference is in what gets saved: instead of a prompt string plus LoRA stacks, it saves a full recipe payload - the structured bundle that includes positive/negative prompts, model slots, sampling parameters, resolution, and seed.

The optional recipe_data input is what makes it useful as an inline node: connect a recipe in progress (from a Recipe Builder, Extractor, or Renderer), and you can tweak the text field or just forward the payload through, then save the result as a named entry. The text input is a direct editor for the prompt portion - handy for a quick tweak before saving. There's also thumbnail support built in, so saved entries show you a preview instead of a filename.

It's an output node, which means it's happy as the tail of a recipe workflow: the recipe runs, the renderer produces its recipe_data output, and this node archives the successful configuration for next time.

Where it fits

The recipe-first flow looks like this: extract or build a recipe, render it, and when you like the result, run it through Recipe Manager to save it. Later, reload that entry and feed it straight into a Recipe Renderer or Recipe Relay to reproduce the run. That's the whole pitch of the recipe toolset - repeatable pipelines instead of "I saved a prompt and now I have to redo the sampler settings by hand." Use Prompt Manager when you're organizing prompts for a hand-built graph; use Recipe Manager when you want the graph itself to be reproducible.

Installing

It ships with the pack:

cd ComfyUI/custom_nodes
git clone https://github.com/FranckyB/ComfyUI-Prompt-Manager.git
cd ComfyUI-Prompt-Manager
pip install -r requirements.txt

Or install via ComfyUI Manager (search "Prompt Manager") and restart. No models or extra dependencies.

Gotchas

Because it saves full configurations, entries are coupled to the models and LoRAs you had installed at save time. Reload a recipe on a machine missing one of its LoRAs and you'll get a missing-asset warning rather than a silent failure - which is the right behavior, but be ready to re-check your installs after moving setups. And if you're comparing it to Prompt Manager: if you never plan to touch the recipe toolset, this node has no real job - its only output is recipe_data, which standard ComfyUI nodes can't consume. It's only useful as part of the recipe workflow.