GALIAIS-Nodes Prompt Combine

Every workflow eventually needs to stitch prompt pieces together, and while the GALIAIS-Nodes pack has composer nodes for structured character work, sometimes you just have ten arbitrary strings that need to become one prompt. That's Prompt Combine: up to ten string inputs (段1 through 段10), joined into one output, with dedupe and a custom separator.

It's the pack's generic plumbing node - the one you reach for when the fancy structured nodes are overkill and you just need section_a + section_b in a defined order.

How it works

The two required inputs are the join behavior: 去重 (dedupe, on by default) removes tags that appear in more than one section, and 分隔符 (separator) is the character used between pieces - leave it empty for the pack's normal comma-plus-space joining, or set your own. The ten sections are all optional; empty ones are skipped entirely, so you don't need to populate all ten to get a clean output. Sections can be typed in directly or wired from other nodes (they're marked as force-input in the source, so ComfyUI lets you connect them).

Outputs: 提示词 (the joined string) and 元信息JSON (which parts were non-empty and whether dedupe ran). For a beginner, 提示词 is the one that matters - it feeds straight into your CLIP text encode or into a node that expects a single positive prompt, like the Prompt Orchestrator or the enricher.

Installing it

It's part of the GALIAIS-Nodes pack:

cd ComfyUI/custom_nodes
git clone https://github.com/GALIAIS/GALIAIS-Nodes.git

Restart ComfyUI, or install via ComfyUI Manager ("GALIAIS-Nodes"). No Python dependencies, no database, no AI setup - this is one of the few nodes in the pack that works fully standalone.

Notes

Two small things worth knowing. First, dedupe is token-level, and it's why you should leave it on: when you're composing a prompt from sections that each contain 1girl (character identity, appearance, pose - it happens constantly), dedupe is the difference between a clean prompt and a redundant one. Second, order is preserved as you fill the slots - 段1 always lands before 段2. If your workflow needs pieces reordered dynamically, this isn't the node; you'd gate the sections upstream. And one honest caveat: because everything is joined at the same weight, Prompt Combine is a joiner, not a formatter - if you need per-section weights, the pack's Prompt Section or Prompt Weight nodes are the structured alternative. For everything else, it's the right tool: boring, reliable, and always in the drawer.