This was trained pretty poorly due to captioning. The clothing was captioned out, which is good when listed for you, which it was not, but the character's traits/look was as well sadly thus you must add it properly. I listed the prompts below that most of you will want to save instead of just "maggie".

maggie, black hair, 1girl, long hair, solo, breasts, blunt bangs, jewelry, earrings, freckles, black eyes, large breasts, shirt, skirt, open mouth, clothes around waist,

I've not tested on comfyui, as it has pissed me off recently not saving hours of making a workflow due to a bug, but webui will need to double or triple load the lora like so in prompts.

<lora:Maggie_IL:1> <lora:Maggie_IL:1> <lora:Maggie_IL:1>

Also, it MAY or may not effect your output adding additional prompts like "theloudhouse, the loud house," in case your checkpoint actually knows what it is, which some do and they don't seem to be too aggressive and just improve the quality slightly.

So please use this knowledge to make this work correctly for you. I don't suggest increasing the weight over 1 as <lora:Maggie_IL:1.5> or higher will work by itself but it also causes more issues than just loading it multiple times which causes none. And yes, you can load the lora multiple times at 0.75 instead of 1 and get good results that act like a normal lora being loaded that way once.