higgsfield

Installation
SKILL.md

Higgsfield AI Prompt Skill

Language rule: Reply in whatever language the user writes in.


MANDATORY WORKFLOW — do these in order, every time

For ANY Higgsfield-related request, follow this contract:

  1. Route the request. Match the user's ask to the routing table in the "Route to the Right Skill" section below, and open the matching sub-skill file(s) with the read tool BEFORE writing any prompt or advice. Do not rely on prior knowledge — platform vocabulary, preset names, and model parameters must be read fresh from the skill files, because this platform's lineup changes between releases.
  2. Apply MCSLA (Model · Camera · Subject · Look · Action) to every video prompt unless the user explicitly opts out. Full definition lives in skills/higgsfield-prompt/SKILL.md.
  3. Use named platform vocabulary only. Camera names, motion presets, and style tokens must come from vocab.md or the relevant sub-skill. Do not invent terms the platform doesn't recognize — invented names silently degrade generations without warning.
  4. Append shared negative constraints from skills/shared/negative-constraints.md before delivering any prompt.
  5. On the first Higgsfield response in a conversation, state ONE short line naming which sub-skill you're routing to. Example: "Routing to higgsfield-prompt + higgsfield-camera for a cinematic chase." One line only, then proceed with the work.

HARD RULES (do not skip under any circumstances)

  • NEVER write a Higgsfield prompt without reading at least skills/higgsfield-prompt/SKILL.md first in the current conversation.
Related skills

More from osidemedia/higgsfield-ai-prompt-skill

Installs
48
GitHub Stars
64
First Seen
Apr 14, 2026