Clean

Shot lists that earn their place, scored rather than checked. Walter Murch ranked what a cut must honour and each item outweighs everything below it combined - emotion 51, story 23, rhythm 10, eye-trace 7, 2D plane 5, 3D space 4 - so a shot that respects screen direction and spatial geometry but misses the emotion is not two-thirds of the way there, it scores nine. That is why the free in-page tool returns a number and a ranking rather than a list of ticks: it reads every card in a fourteen-field shot list, scores it out of 100, and names the single criterion costing the most across the whole piece. It also catches what is simply wrong: the fifteen words that do not render, each with a concrete replacement; an emotion named without a body, which is the 51-point omission; three camera moves stacked in one clip; contradictions the model resolves by picking one at random; a shot carrying none of the three details; and a rhythm ladder with no pause before its biggest cut. Five lanes over one scene idea: treatment and five anchors, the shot cards, the rhythm ladder, a model prompt for Seedance, Kling or Veo, and a scoring pass that returns a repaired list. Lanes derived from the open-source skill smixs/visual-skills by Serge Shima, licensed CC BY 4.0. Not affiliated with or endorsed by Serge Shima, Walter Murch, or any director named in the reference library.

Share

Details

PricingUsage-based + 10% creator margin
Billed model rate$2.20 in / $13.20 out per 1M tokens
Creator margin+10%
Effective rate$2.40 in / $14.40 out per 1M tokens
Security scanClean — skill and frontend scanned
Model gpt-terra
Created2026-08-29
Updated2026-09-03

Every public app is built from a security-scanned skill and must pass a clean scan — skill and frontend — before it can be listed. Have a skill of your own? Turn it into an app — or read the step-by-step walkthrough.