Large Classify Function

Multi-agent workflows have a wiring problem: if you want the same piece of context - a shared instruction, a retrieved knowledge chunk, a user's original question - feeding into ten different downstream branches, you end up dragging the same wire across your graph ten times, or littering it with Reroute nodes. Large Classify Function solves exactly that plumbing headache: one text input, up to ten identical outputs.

How it works

You give it a single text value, and it comes back out through as many of the ten output ports as you've wired up. That's the whole mechanism, based on what the node actually exposes - there's no separate parameter for splitting logic, delimiters, or category labels, so don't expect it to route content differently to different ports based on its meaning. Think of it as a broadcast node rather than an actual classifier: the "classify" in the name is really about pairing it with the party's persona-building nodes (classify_persona and classify_persona_plus), which take multiple text inputs and merge them the opposite direction - this one splits one thing into many, they combine many things into one.

The inputs and outputs that matter

• text - the single string you want available on multiple output ports.
• is_enable - turn the node off without removing it from the graph.
• Ten outputs, response1 through response10, each carrying the same text value. You only need to wire up as many of these as you actually use - leaving the rest disconnected is fine.

How to install it

• ComfyUI Manager - search "comfyui_LLM_party", install, restart.
• Manual - cd ComfyUI/custom_nodes && git clone https://github.com/heshengtao/comfyui_LLM_party, then restart.

Then pip install -r requirements.txt from inside the comfyui_LLM_party folder, in your ComfyUI Python environment (python_embeded\python.exe -m pip install -r requirements.txt on portable Windows). Nothing special beyond the pack's baseline requirements - this node is pure Python plumbing with no API calls or models involved.

Common issues & troubleshooting

Expecting real classification behavior and not getting it. If you came to this node hoping it would route text to different outputs based on content or category, that logic isn't in the node's exposed inputs - it broadcasts the same value to every connected port rather than sorting it. If you need actual conditional routing, pair an LLM classification step with any_switcher or bool_logic instead, and use this node purely for the fan-out plumbing.

Only some outputs seem to update. All ten carry the same value on every run - if one downstream branch looks stale, check that branch's own wiring rather than this node, since there's no per-output logic here to go wrong.

The pack isn't loading. Separate from this node, comfyui_LLM_party has a real, current install issue where nodes fail to import after installing through Manager - usually a CUDA/PyTorch mismatch with no clear compatibility list published. Check that first if classify_function_plus isn't showing up in your node browser.