OSideMedia/higgsfield-ai-prompt-skill ? reverse-engineered prompt

Build me a Claude skill that helps people turn plain English ideas into strong Higgsfield AI prompts for cinematic video and image generation.

I want it to guide the user through making a prompt that feels ready to use, with support for things like camera direction, subject description, look and mood, motion, and character consistency. It should also help choose the right generation mode for common needs like reference based shots, video edits, extensions, and one shot cinematic scenes.

Please include a good set of reusable templates for genres like action, horror, romance, sci fi, ads, and character focused scenes, plus clear advice for avoiding weak results and fixing prompts that are not working well. If it makes sense, add a simple workflow for keeping the same character consistent across multiple shots and for building prompts that work well with Higgsfield style tools. Look up current docs online if you need to.