ComfyUI Extension: ComfyUI-LoraSlider-LM
Run ComfyUI workflows without the setup
No installs, no CUDA version roulette, no GPU sitting idle on your bill. Bring a workflow and run it in the browser.
A self-contained LoRA Slider node for ComfyUI designed to work alongside ComfyUI-Lora-Manager with bipolar strength sliders and batch LoRA loading.
README
ComfyUI-LoraSlider-LM
A self-contained LoRA Slider node for ComfyUI that is designed to sit alongside ComfyUI-Lora-Manager (LM) and reuse its rich metadata — without modifying LM in any way.
It brings the slider UX of ComfyUI-Flex-Pack into the LM ecosystem, while fixing Flex's biggest pain point: adding and configuring sliders one at a time.
What it does
- One bipolar strength slider per LoRA, with editable min/max bounds and semantic end-labels — e.g. label the low end "skinny" and the high end "obese". The slider shows the actual strength value (not a 0–100 %).
- Batch-add many LoRAs at once, three ways:
- + Add LoRAs — a searchable multi-select picker backed by LM's library. Search
by name or path: include a
/(e.g.turbo/slider) to match folders. - + Text — type names or paste
<lora:folder/name:0.8>codes (e.g. copied from the LoRA Manager UI); each becomes a slider. - Import from Pool — connect LM's Lora Pool node and pull in every LoRA matched by its folder/tag filters.
- + Add LoRAs — a searchable multi-select picker backed by LM's library. Search
by name or path: include a
- Set values manually (drag the slider or type a number), or randomize:
- per-slider 🎲, or 🎲 Strength (all unlocked) — re-roll strengths in range;
- 🎲 Enabled — randomly enable/disable each unlocked slider;
- randomize each run toggles — re-roll strength and/or enabled state on every queued run (great for batch variation);
- 🔒 lock any slider and it is never touched by any randomize (manual or each-run).
- Best-effort auto-fill of min/max (and labels) from metadata. Slider LoRAs rarely expose their usable range as structured data — it's usually buried in the CivitAI description, e.g. "Strength: Safe zone (-3 - +2.5) … Higher is thinner." The node reads LM's stored description and parses the range + direction. Always editable via the per-row ⚙ config dialog; silently falls back to defaults when nothing is found.
- Outputs a standard
LORA_STACK(+ aggregatedtrigger_words) that plugs straight into LM's Lora Loader (LoraManager).
Install
Clone into your ComfyUI/custom_nodes/ directory (next to ComfyUI-Lora-Manager):
cd ComfyUI/custom_nodes
git clone https://github.com/EnragedAntelope/ComfyUI-LoraSlider-LM
Restart ComfyUI. The node appears under the Lora Slider category as LoRA Slider (LoraManager). ComfyUI-Lora-Manager must be installed and running for the picker / pool import / metadata auto-fill features (they call LM's HTTP routes); the node still works for manual sliders without them.
Typical wiring
Lora Pool (LoraManager) ──POOL_CONFIG──▶ LoRA Slider ──LORA_STACK──▶ Lora Loader (LoraManager) ──▶ KSampler
└────────▶ trigger_words (STRING)
How it stays decoupled from LoRA Manager
This package imports no LM Python modules and patches nothing. All integration is through LM's public HTTP API and shared ComfyUI types:
| Need | LM route / mechanism |
| --- | --- |
| Picker search (by name or folder path) | GET /api/lm/loras/relative-paths |
| Expand a Lora Pool folder/tag selection | POST /api/lm/loras/cycler-list |
| Resolve a LoRA's file path | GET /api/lm/loras/list |
| Description for min/max parsing | GET /api/lm/loras/model-description |
| Apply the result to MODEL/CLIP | output LORA_STACK → LM's Lora Loader |
The browser widget fetches metadata and serializes everything it needs into the node's saved value, so execution requires no network and no LM imports.
Development
# Python (node + metadata parser)
python -m pytest
# JS (strength-parser mirror)
node --test tests/strength_parser.test.mjs
nodes/metadata_parser.py and web/strength_parser.js implement the same heuristics —
keep them in sync.
License
MIT
Run ComfyUI workflows without the setup
No installs, no CUDA version roulette, no GPU sitting idle on your bill. Bring a workflow and run it in the browser.