Any Type Switcher
A numbered dial for up to 10 inputs of any type
- any1
- any2
- any3
- any4
- any5
- any6
- any7
- any8
- any9
- any10
- which_any
Every ComfyUI workflow eventually hits the moment where you want to try two or three versions of something - two prompts, two personas, two model loaders - without rewiring the graph every time you switch. Any Type Switcher is comfyui_LLM_party's plumbing fix for that: plug in up to ten values of anything, and pick which one comes out with a single number.
How it works
It's deliberately dumb, and that's the point. Wire values into any1 through any10 - they can be strings, images, models, tool definitions, whatever your graph produces, since the node accepts a wildcard type. Set num to the slot you want, and that one value passes straight through as the output. Nothing gets transformed or inspected; it's a selector, not a processor. Change num from 1 to 2 and you've swapped which branch of your graph is "live" without touching a single wire.
The inputs and outputs that matter
num(1–10, default 1) - the only input you'll actually touch during normal use. This is your dial: whichever slot number you set here is what gets output.any1throughany10- all optional, all wildcard type. You don't need to fill all ten; just wire in as many alternatives as you're actually comparing.- One output:
which_any- the value from whichever slotnumselected, passed through untouched, ready to feed into whatever comes next in your graph.
This pairs naturally with the party's other agent-building nodes - flip between two persona strings from classify_persona, two different tool sets, or two model loaders - without maintaining duplicate downstream graphs for each option.
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 run 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 for portable Windows builds). This node has no dependencies of its own beyond the pack's baseline - it's pure Python logic, no models, no API calls.
Common issues & troubleshooting
Output is empty or errors downstream. If num points at a slot you never wired anything into, the output is empty - ComfyUI won't stop you from setting num to 7 when you only filled in any1 through any3. Double check the slot count matches what you actually connected.
Downstream node complains about type. Because the wildcard type accepts anything, it's easy to plug genuinely incompatible things into different slots - a string in any1, an image in any2 - and then be confused when switching num breaks whatever's downstream. The node won't stop you from mixing types across slots; keep your alternatives type-consistent if the same output wire needs to work for all of them.
The pack isn't loading at all. This is a known, current issue independent of this specific node - users installing comfyui_LLM_party via Manager and running requirements.txt manually have reported nodes failing to import, usually from a CUDA/PyTorch mismatch with no clear compatibility list published. If any_switcher isn't showing up in the node browser, check that first rather than assuming something's wrong with this node specifically - it's about as simple as nodes get.
Inputs (11)
| Name | Type | Default | Description |
|---|---|---|---|
| num | INT | 11–10 | — |
| any1opt | * | — | |
| any2opt | * | — | |
| any3opt | * | — | |
| any4opt | * | — | |
| any5opt | * | — | |
| any6opt | * | — | |
| any7opt | * | — | |
| any8opt | * | — | |
| any9opt | * | — | |
| any10opt | * | — |
Outputs (1)
| Name | Type | Description |
|---|---|---|
| which_any | * | — |