Large Classify Persona

For a simple assistant, three text slots (comfyui_LLM_party's plain classify_persona) is plenty. For something more elaborate - a roleplay character with backstory, speech patterns, relationship notes, and a set of hard rules all kept separate so you can tune each independently - three isn't enough. Large Classify Persona is the same idea scaled up: up to ten text pieces plus a file, merged into one system prompt.

How it works

Exactly like its smaller sibling: it concatenates whatever text inputs you've connected - plus an optional block of file content - into a single string shaped for a system prompt. The value of keeping ten separate slots rather than one big text box is editing discipline: you can update "the character's speech quirks" (say, text4) without scrolling through and risking a typo in "the character's backstory" (text1). For anything with real depth - a persistent agent persona you're going to iterate on over weeks, not a one-off - this structure pays for itself quickly.

The inputs and outputs that matter

• file_content (optional) - pull persona text from a file rather than typing it directly, handy if a chunk of your character sheet already lives as a document.
• text1 through text10 (all optional) - your persona pieces. Use as few or as many as the character actually needs; nothing forces you to fill all ten.
• is_enable - toggle the whole node off without deleting it.
• One output: system_prompt_input - feed this into the system prompt slot on your LLM/agent node.

If ten slots is overkill for what you're building, classify_persona (three slots) is the simpler version of this same node.

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). No extra dependency for this node beyond the pack's baseline requirements.

Common issues & troubleshooting

Persona feels bloated or the LLM starts ignoring parts of it. Ten text slots make it easy to accumulate more instruction than the model can actually hold onto - long, sprawling system prompts are a known way to get an LLM to quietly drop or blend instructions rather than follow all of them. If your agent's behavior gets inconsistent, trim rather than add another slot.

Wrong text slot updated, persona doesn't seem to change. With ten inputs, it's easy to edit text3 when the trait you meant to change actually lives in text7. Keep a note of what each slot holds as your persona grows past a couple of pieces - the node itself doesn't label them for you beyond the numbers.

Output is empty. Every field here is optional, so it's possible to run this node with nothing actually wired in and get an empty system_prompt_input. Check that at least one text slot or file_content is connected if your agent isn't showing any persona-driven behavior.

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