# Trope Table

> A periodic table of 48 common fan-fiction tropes. Paste a story premise and its tags: the table lights up with the
> tropes you use and shows your story's formula, free in the browser. A metered read keeps each trope's promise to
> readers and changes one axis of it, adds a catalyst, and writes a new summary. https://trope-table.skillsafe.ai/

## What is free (in the browser)

- The table: 48 tropes in six families - Pairing (Enemies to Lovers, Slow Burn, Fake Dating...), Plot devices (Only One
  Bed, Time Loop, Body Swap...), Settings + AUs (Coffee Shop AU, Royalty AU, Small Town...), Dynamics (Grumpy /
  Sunshine, Rivals, Found Family...), Feelings (Hurt / Comfort, Angst, Fluff...) and Structure (Five Times + One,
  Epistolary, Canon Divergence...). Each has a number and a two-letter symbol.
- Matching: word lists over the premise and the tags (a tag that is a trope's name always counts). These are
  estimates, and the page says so.
- The formula: the matched symbols in order, such as El + Sb + Ca + Gs.
- The page's verdict, by ordered rules: needs a premise (under 25 words, tags aside) > trope pile-up (6 or more tropes)
  > already fresh (every trope already twisted - only the read can see this) > twist the execution (otherwise).
- Exports: Markdown, plain text (the new summary and its tag line) and JSON. The summary and the tags also copy
  separately, one per box of a posting form; the tags keep the writer's own (fandom, characters) minus the ones the read cut.

## The metered read (task "twist")

Input: premise, tags, mood (either / fluff / angst), plus the page's matches and verdict. Output: verdict, a working
title, a summary, and every trope in play with what readers expect, what it gives them (the function that must
survive), the axis a twist moves (form, knowledge, goal or role - the Orthogonality Principle) and the twist; a
catalyst (one person, rule or thing from outside the tropes' world with a goal of its own); a 40-90 word summary
that uses only the premise's names; or, for a bare premise, 2-4 questions. The read never adds a named character,
a canon fact or a gender the premise did not give, and the page re-checks every reply for that.

## Sources

- Built on the SkillSafe skill @jwynia/cliche-transcendence (https://skillsafe.ai/skill/@jwynia/cliche-transcendence):
  its eight-step process (enumerate the defaults, extract the functions, generate alternatives, invert perspective,
  import from other domains, test specificity, trace consequences) and the Orthogonality Principle.
- Model: gpt-terra, metered per run. The samples are original characters made up for this page.

## Pages

- https://trope-table.skillsafe.ai/ - the app
- https://trope-table.skillsafe.ai/api.html - the API, with a worked example per verdict in eight languages