# Same Bones

> Remake a story you love in a new setting without making a reskin: X-ray what makes it work, sort every part into
> transfer, translate or replace, and check whether the remake pitch is a real transformation.

https://same-bones.skillsafe.ai/

## What it does

You describe an original (a book, film, series, game, play or myth) in your own notes - one line per character, turn or
moment, plus what you love about it - and name a new setting and the form the remake will take.

Free, in the browser:

- The notes are split into numbered parts ("bones") and each gets a first keyword sort: transfer (human nature, works
  anywhere), translate (how the original world works - authority, resources, messages, enforcement) or replace (bound
  to one period or culture). You can change any sort.
- The setting is read for the pressures it names (authority, resources, communication, enforcement, environment) and
  for a tone clash (tragic notes inside a light comedy or a show for small children): rich / thin / clash.
- A pitch is read for what it carries over word for word: names from the notes, one-for-one swap phrasings, "but in"
  framing, period words the new setting would not have: transformed / reskin / fidelity trap / genre collision.
- The story is drawn as an X-ray film on a lightbox - skull = core appeal, spine = central tension, one rib per part,
  coloured by kind - with a "new skin" overlay of the setting.

Two metered lanes (a model call each, priced before you run):

- `xray` - verdict `rich`, `thin` or `clash`; the core appeal (journey, what is underneath, the tension, peak moments),
  who holds power, every bone's job and proposed new form, the pressures the setting adds, a logline, risks
  (reskin, fidelity trap, genre collision, surface diversity) and questions to answer before writing.
- `verdict` - verdict `transformed`, `reskin`, `fidelity_trap` or `genre_collision` (checked in that reverse order:
  genre collision first, then fidelity trap, then reskin); four checks (systemic consistency, emotional core,
  originality, cultural care) as pass / mixed / fail; bone-by-bone swaps marked rebuilt, surface or forced; fixes;
  whether it stands alone; a revised logline.

The page re-checks every reply: one bone per part of the notes, swaps that point at words the pitch really uses,
verdict rules that hold (a fidelity trap has a forced swap, a reskin fails originality), no internal field names, and
no gendered pronoun for a character the notes never gendered.

## Grounding

The model works only from your notes about the original - it does not add characters, scenes or facts from its own
knowledge of the work. New-setting material is offered as proposals. Everything you mark as a must-keep survives.
The samples use public-domain works (Frankenstein, Pride and Prejudice, Romeo and Juliet, A Christmas Carol) with
made-up pitches.

## Sources

- Built on the SkillSafe skill @jwynia/media-adaptation: https://skillsafe.ai/skill/@jwynia/media-adaptation
- API tutorial: https://same-bones.skillsafe.ai/api.html