There are some few things to achieve consistency. IDK what AI you use, i will explain how i do.
I use Krita with the plugin 'Krita AI Diffusion'. If you don't know them. Krita is a free 'photoshop clone' and the plugin i mentioned is a unofficial add-on that let you run either cloud or local diffusion models. (i use the latter).
Also, you need to draw, if you have an idea in mind it will be more easily to draw what you want and then use the 'refiner' option of the plugin. (The more clearly/better you draw, the better results you will have).
I'd recommend you use a checkpoint-specific-style per art style instead of a 'all in one'. Also, if you already draw, you can train a Lora in your own style.
Broken hands, is like weird faces or weird eyes. Using some text inversions can help. But working over the generated images is still mandatory. There is no such thing as 'create everything with a prompt' That's more a marketing stuff.
Extra fingers are done normally by 'cutting' the finger, editing the wrong hand, selecting that area and refine with 40 or 50% strength (less may be needed sometimes).
Faces sometimes is as easy as selecting the face, use: "High quality,HD,hd,high quality" as prompt with an intensity of 50% or less. If the problem is you don't have consistent faces between images (like if they belong to different people and not your characters) then tool like FaceFusion helps (Basically you do a DeepFake of your characters).
Ai is a tool. Is a matter in learning how to use it. Like i said before, many people was caught up with the advertising that they could to anything with a simple prompt. But is like redonihunter just said. As an artist/musician/programer (the creator) you have to put some effort in it.