🔶 Lora +add_lora

A LoRA loader with two things going for it beyond the stock ComfyUI one: it chains with other Chaosaiart LoRA nodes into a stack, and it surfaces the LoRA's own trigger-word metadata so you don't have to go hunting on CivitAI to remember what to type. The README credits idrirap's Lora-Auto-Trigger-Words node for the idea - "all tags are included in the metadata," and this node reads that same metadata off your LoRA file and reports it back to you.

lora_name is a dropdown built from whatever's in your LoRA folder - populated at ComfyUI startup, so a file you just added won't show up until you restart. strength_model and strength_clip (both default 1, range -20 to 20) control how strongly the LoRA affects the diffusion model and the CLIP text encoder respectively. Default 1.0 is often stronger than you want - 0.5 to 0.8 is the common range for most LoRAs before they start overpowering your base checkpoint or introducing artifacts, and it's worth backing off from 1.0 as your first troubleshooting step if a LoRA's results look off rather than assuming the file itself is bad.

The add_lora input (type LORA) is where the chaining happens: plug another chaosaiart_lora or chaosaiart_lora_advanced node in here and you're stacking LoRAs - style plus character plus concept, for instance - the same way you'd chain multiple LoraLoader nodes in a standard workflow, just built as a daisy chain instead of separate parallel nodes. Worth knowing before you go wild stacking: interactions between multiple LoRAs are genuinely unpredictable, and two LoRAs that look fine individually can fight each other once combined, so it's worth testing additions one at a time rather than stacking five and troubleshooting blind.

Outputs: LORA (the chain - plug this into another Lora node's add_lora, or into chaosaiart_Prompt's add_lora input to get it applied downstream by chaosaiart_MainPromptCLIPEncode), and INFO (a text string with the trigger words/tags pulled from the LoRA's metadata, care of the idrirap-inspired feature the README mentions - wire this to a Show Text node if you want it visible while you work).

Install: search "Chaosaiart-Nodes" in ComfyUI Manager, or use "Install via Git URL" with https://github.com/chaosaiart/Chaosaiart-Nodes. Manually: cd ComfyUI/custom_nodes && git clone https://github.com/chaosaiart/Chaosaiart-Nodes, then restart ComfyUI. On Linux, also run pip install opencv-python and pip install tqdm inside your venv - the Windows install script handles this automatically.

One real limitation: LORA is a type this pack defines for itself, so this node only chains with other 🔶Chaosaiart LoRA/Prompt nodes downstream - you can't wire its output into a standard LoraLoader or another pack's stacker node. If you need to mix Chaosaiart's LoRA chain into a non-Chaosaiart pipeline, this isn't the node for that.