Large File Combine

Exactly what it sounds like: File Combine's three input slots, expanded to ten. If you're building a knowledge base out of a genuine document set - a handful of reference files feeding a RAG pipeline, several attachments bundled for one downstream node - three slots runs out fast, and this is the node for that case instead of chaining multiple File Combine nodes together.

What it does

Ten optional string inputs, file1 through file10, plus is_enable for the usual bypass toggle. Wire in as many as you actually have - none are required - and the single output, files, is all of them merged into one value. Same underlying behavior as File Combine, just scaled up: no validation of what the strings actually are, no filtering, just a combine step for whatever content you connect.

Where it fits

This is the node to reach for once a workflow genuinely needs to merge more than three sources - a folder's worth of document excerpts headed into this pack's Word Vector Search node, or a set of files being bundled ahead of a node (like one of the Gitee file-bed or Feishu send nodes) that only exposes a single file-path input. If you consistently need fewer than four inputs, plain File Combine is the lighter option and functionally identical below its three-slot ceiling.

Installing it

Ships with the pack, no separate step:

• ComfyUI Manager: search "comfyui_LLM_party", install, restart.
• Manual: cd ComfyUI/custom_nodes && git clone https://github.com/heshengtao/comfyui_LLM_party, then pip install -r requirements.txt from inside the pack folder using ComfyUI's Python, then restart.

No external dependencies beyond the pack itself - this is a pure combine utility.

Common issues & troubleshooting

Combined output is hard to parse back apart. Same caveat as File Combine: this node doesn't insert any particular delimiter logic you can rely on being a specific format - check what your downstream node actually expects and confirm this node's output matches it, rather than assuming a separator.

Only using two or three of the ten slots. Perfectly fine - every input is optional, and there's no requirement to fill all ten before the node produces useful output. If you consistently only need a few, plain File Combine does the identical job with a smaller, tidier node on the canvas.

Genuinely need more than ten sources. At that point this pair of nodes has hit its ceiling - you'd be looking at chaining two Large File Combine nodes together (feeding one's files output into another one's input slots), or restructuring the workflow to loop over a folder instead of wiring each file in individually.