SKULL → scene
premise, pov, the single weight the scene carries.
A multi-window cockpit for AI-assisted film & video production. Six rooms. One story format. Every prompt is a seed; every scene blooms from the same skeleton.
# Scene 12 — Foundry interior, night. Rhea finds the pattern. scene: int_foundry_night premise: "the wall remembers the song" cast: - id: rhea ref: studio://cast/rhea@v3 mood: [curious, afraid] world: location: studio://loc/foundry era: post-bloom light: single_sodium, falloff_fast timeline: - beat: enters # 00:00 - beat: wipes_ash # 00:08 - beat: hears_humming # 00:22 shots: - id: 12a framing: wide lens: 35 - id: 12b framing: ots lens: 50 - id: 12c framing: close lens: 85 render: engine: higgsfield bone: anamorphic.x@boneyard.v2 seed: 0247
Every cockpit speaks the same dialect: SKEL in, BONE out. The skill file makes that dialect legible to an LLM — so the model can open a scene, read its bones, write the next shot, and compile a prompt without ever leaving the project.
Drop it into Claude. Codex. A local agent. Anywhere a model can read text, SPORE can be operated by voice.