GR Prompt Selector Multi

Don't confuse this with its sibling GR Prompt Selector, which picks one of six prompts with a single number input. This node doesn't pick anything - it encodes all six, all the time, and hands you back six separate CONDITIONING outputs. Think of it less as a selector and more as six CLIP Text Encode nodes bundled behind one clip wire, saving you from placing and rewiring seven separate nodes (six positive plus one negative) every time you want this shape in a graph.

How it works

You write six prompts into positive_a1 through positive_a6. A separate field, alwayspositive_a1 - no underscore between "always" and "positive," that's the literal input name - holds text that gets folded into every one of the six encodes: your usual place for a style tag, a quality phrase, anything you want present no matter which of the six outputs you end up using. negative_a1 is encoded once and shared across the board. All of it runs through the same clip input.

Because nothing here decides a winner, the six outputs are genuinely independent and simultaneously live. That's the actual use case: six characters in a regional-prompting setup that each need their own conditioning, six variations you want queued as separate branches without disturbing the wiring on the ones you're not touching this run, or just a fast way to keep several prompt ideas encoded and ready so you can swap which output feeds your sampler by moving one wire instead of re-typing text. If what you actually want is "run whichever one number I set," reach for GR Prompt Selector instead - this node has no equivalent switch.

The inputs and outputs that matter

• clip - required, shared across all seven encodes this node performs internally.
• positive_a1 through positive_a6 - your six independent prompts.
• alwayspositive_a1 - text merged into every one of the six, regardless of which you use.
• negative_a1 - the single shared negative.

Seven outputs: positive1 through positive6 (CONDITIONING, one per slot) and negative (CONDITIONING). All are wired straight to a sampler's conditioning inputs - there's no combined "prompts" text log here the way there is on some of this pack's other prompt nodes, so if you need to remember which output came from which prompt text later, that's on you to track (label your prompt slots clearly, or note it outside the graph).

How to install it

ComfyUI Manager, search GraftingRayman. Manual:

cd ComfyUI/custom_nodes
git clone https://github.com/GraftingRayman/ComfyUI_GraftingRayman

Restart afterward. The pack won't import at all without OpenAI's CLIP installed separately - portable ComfyUI: .\python_embeded\python.exe -m pip install git+https://github.com/openai/CLIP.git; system Python: pip install git+https://github.com/openai/CLIP.git. Note this node's own clip input is your loaded CLIP checkpoint from a checkpoint loader - a completely different thing from the OpenAI CLIP Python package the pack needs at startup.

Common issues & troubleshooting

Pack won't load. Check your console at startup for the missing-CLIP-package import error and run the pip install above.

Only some outputs seem to matter. That's expected - you don't have to consume all six. Leave the ones you're not using this run unwired; nothing about the node requires you to route every output.

Wanted it to pick one automatically. This node has no selection logic - it always encodes and returns all six. For automatic picking or randomizing, use GR Prompty; for a manual single-number switch, use GR Prompt Selector.

Lost track of which output is which. There's no readback string here - write distinguishable text into each positive_aN slot (or keep notes elsewhere), since the node itself won't tell you afterward what you put where.