🎨 快速保存Prompt
Keep the prompts that actually worked, with one click
This is the "don't lose the good one" node. You generate something you like, you've fiddled with the prompt for an hour, and the temptation is to copy it into a text file you'll never open again. Instead you wire the winning prompt text into this node, give it a name, pick a category, and run once - it's now a saved prompt in your gallery, searchable, cover-bearing, ready to be selected by PromptsSelector in any future workflow. It's the inbound half of the loop that SaveToGallery closes on the image side.
It's a thin utility and it knows it: no outputs, no processing, just "put this string in the library under that name."
How it works
Three inputs, all straightforward:
prompt_name- the display name the prompt gets in the gallery. The node refuses to save with an empty name.category- the target category from a dropdown of your existing categories. The dropdown is snapshotted at load, so a brand-new category may need a page refresh to appear.prompt_value- the actual prompt content. This one isforceInput, so it's meant to be wired from a string output (your CLIP Text Encode's text, or a note node) rather than typed - that's what makes it "quick."
The save runs in a background thread so the workflow isn't blocked. The logic is simple upsert: if a prompt with the same name already exists in that category, its value is updated to the new text; otherwise a new prompt is created. Same name, same category → overwrite. That's the whole contract, and it's exactly right for a node whose job is "keep this current version."
Errors - like an empty name or empty value - print to the ComfyUI console and don't crash the graph. Check the console if the save silently didn't happen.
Where you'd use it
Two patterns worth stealing. One: keep a "current best" prompt per subject - wire your sampler's text into QuickSavePrompt with a fixed name, and every good run quietly refreshes your library's canonical version. Two: use it mid-iteration as a save point - liked a variation? Name it portrait v3, rerun, move on. The gallery stays the source of truth without you ever hand-copying text.
Install
Same pack-wide steps as everything in Prompt Gallery. ComfyUI Manager → search "Prompt Gallery", or:
cd ComfyUI/custom_nodes
git clone https://github.com/Lotus0614/ComfyUI-Prompt-Gallery
cd ComfyUI-Prompt-Gallery
pip install -r requirements.txt
Windows portable: ..\..\python_embeded\python.exe -m pip install -r requirements.txt. Light deps (Pillow, optional pyahocorasick), no model downloads. Restart ComfyUI and hard-refresh the browser afterwards.
Common issues
- Saved prompt not in the gallery - it's asynchronous, so give it a beat and refresh. If it never appears, check the ComfyUI console for a "[QuickSavePrompt]" error; empty name/value are the usual culprits.
- Name collision surprise - same name in the same category updates the value. Use distinct names if you want history preserved.
- Category dropdown missing a category you just made - reload the page; the list is built when ComfyUI loads its node definitions.
Keep the prompt, throw away the guilt about not filing it. That's the whole point of this node.
Inputs (3)
| Name | Type | Default | Description |
|---|---|---|---|
| prompt_name | STRING | — | |
| category | COMBO | 1 options: 全部 | |
| prompt_value | STRING | — |
Outputs (0)
No outputs