Simple LLM: Load Workflow
A workflow loader that (for now) just says 'loaded'
- STRING
Simple LLM: Load Workflow looks like it should load a saved LLM workflow into your graph. It's a one-option dropdown and a string output, sitting in the "Utils" corner of the pack's Simple LLM suite. And here's the thing you need to know up front: right now it returns the text "Loaded workflow: simpleLLM.json" - a confirmation string, not the workflow itself. If you were hoping for a node that materializes a graph, this isn't that. Not yet.
Let me be fair to it, though. What it is is a placeholder with a real purpose: the pack ships an example workflow (workflow/SimpleLLM_Basic_TextGeneration.json in the repo) showing how to wire the Simple LLM nodes together, and this loader reads from a nodes/LLM folder inside the pack that already contains a simpleLLM.json - the exact file named in its dropdown. It's scaffolding for an intended "load and run a saved LLM flow" feature, shipped early so the plumbing exists.
How it works
The node scans nodes/LLM/ inside the pack for *.json files and builds its dropdown from whatever it finds. In a stock install that's one entry: simpleLLM.json. When you run it, it checks the file exists and returns the confirmation string. That's the entire mechanism - and it's why the output is a STRING rather than a workflow graph or a config object.
The honest version: treat this node as a stub. The real value in this corner of the pack is the example workflow - if you open the pack's workflow/SimpleLLM_Basic_TextGeneration.json, you'll see the intended assembly (config → agent → run prompt → display) laid out as a working graph you can load into ComfyUI and adapt.
Inputs and output
- workflow - a dropdown listing JSON files found in
nodes/LLM/. In a fresh install, exactlysimpleLLM.json. - Output: one STRING, currently the "Loaded workflow: …" confirmation.
The caveat
Because it doesn't return the file's content, you can't wire this into anything expecting an actual workflow definition. If you add your own JSON to nodes/LLM/, it'll show up in the dropdown, but you'll still only get the confirmation string back. Expect this node to change meaning when the pack's LLM section matures - its README openly calls the LLM features "Implementing the tests now."
Installing it
It's part of ComfyUI-NS-Util:
cd ComfyUI/custom_nodes
git clone https://github.com/NakamuraShippo/ComfyUI-NS-Util
or ComfyUI Manager → Install via Git URL with the same URL, then restart. No model downloads, no API keys needed for this node itself - it's pure local file bookkeeping. Dependencies are the pack's standard light set.
When you'd actually use it
Honestly? For the confirmation it gives that the example workflow file is present, and as a starting point for understanding where the pack stores its LLM workflow JSONs. If you want to use the Simple LLM system, skip this and load the example workflow from the pack's workflow/ folder instead. One star for usefulness today, but it's a fair hint about where this suite is heading - and the "100% CTR" on its page suggests people keep clicking it hoping for more than it delivers.
Inputs (1)
| Name | Type | Default | Description |
|---|---|---|---|
| workflow | COMBO | simpleLLM.json | 1 options: simpleLLM.json |
Outputs (1)
| Name | Type | Description |
|---|---|---|
| STRING | STRING | — |