media-forgemf-ugc-scriptlisted
Install: claude install-skill produtoramaxvision/maxvision-utilities
# media-forge:mf-ugc-script
Turn a brief into a timed script where every line has a picture attached. If
there is no brief yet, run `[skill:mf-ugc-brief]` or `[skill:mf-ugc-decode]`
first — scripting from nothing produces the generic thing both of those exist to
prevent.
## Intent
A short-form script is not prose with timestamps. It is a sequence of visual
states, and the words are what a person says while the picture changes. Written
the other way round — copy first, visuals bolted on — the result reads as an
advertisement being performed, which is the one thing UGC must not be.
## Shape
```
[0:00-0:03] HOOK
VISUAL: what is on screen, as a shot
AUDIO: what is said, if anything
[0:03-0:XX] BEAT n
VISUAL:
AUDIO:
```
Rules that make the shape hold:
- **One thing per beat.** If a beat needs "and", it is two beats.
- **VISUAL before AUDIO, always.** Writing the line first is how a script ends
up unshootable.
- **A beat with no visual change is a beat that gets cut.**
- **The product is on screen before the CTA**, not simultaneously with it.
Lengths that work: 15s = hook + 2 beats + CTA. 30s = hook + 4 beats + CTA.
60s only when there is a real demonstration to watch; otherwise it is a 30s
script with padding.
## Anti-slop — the part that decides whether it sounds human
`[skill:mf-antislop]` is the authority and this must not diverge from it. Load
it rather than re-deriving. What matters most here:
**Never:** "Le