# Antagonist Design

The antagonist is the story's embodied **counter-argument**: the character who
believes and acts on the counter-premise (see `controlling-idea.md`). Scenes
trade wins between the two positions until the climax delivers the verdict.
Design the antagonist to deserve that role.

## The worthy opponent

The antagonist must be stronger than the protagonist in at least one
dimension — intelligence, resources, position, ruthlessness, or conviction. A
weaker antagonist makes the protagonist's victory feel unearned and the theme's
verdict feel rigged. Record the asymmetry explicitly in the character file:

```yaml
antagonist-edge: "{Where the antagonist outmatches the protagonist}"
```

## The antagonist believes they are right

No one is the villain of their own story. The antagonist acts on a coherent
moral logic — the counter-premise stated as something a person could defend:

- Not "I want power because I am evil" but "Order requires a strong hand;
  mercy is weakness that gets people killed."
- Their flaw mirrors the protagonist's lie: both may share a wound, but the
  antagonist chose the lie permanently (a negative-arc character; see
  `lie-truth.md`).

If you cannot write a paragraph defending the antagonist's position, the
antagonist is a plot device, not a character — deepen or replace them.

## Plan the antagonist as a protagonist

Build the antagonist's through-line as if they were the hero of their own
story:

- Their **want** (external goal), their **need** (what the truth would give
  them), their **ghost wound** (why they believe the lie).
- Their **plan**: a sequence of moves against the protagonist, each adapting
  to the protagonist's resistance. The antagonist should *react* — a static
  obstacle is scenery, not opposition.
- Their **turning points**: moments the protagonist's truth nearly reaches
  them, refused (negative arc), or moments their lie's cost becomes visible
  to the reader even if not to them.

Record this as a `## Antagonist Through-Line` section in the character file
or as beats in the relevant arc file.

## Personify institutional opposition

When the real opposition is a system (a bureaucracy, a corporation, a
tradition, an empire), give the force a face:

- One character who *embodies* the system and can be confronted in scenes.
- The face character should genuinely believe in the system — a true
  believer is scarier and thematically richer than a corrupt functionary.
- The system itself keeps its own pressure (rules, consequences, inertia)
  that persists even when the face character loses a scene; note the
  institutional pressure separately from the character's actions.

## Antagonist checklist

- [ ] Has an edge over the protagonist in at least one dimension.
- [ ] Believes the counter-premise; their position is defensible in a
      paragraph.
- [ ] Has a want, a wound, and a plan that adapts across the story.
- [ ] Wins scenes — the protagonist's victories are contested, not gifted.
- [ ] Their defeat at the climax is a *thematic* defeat: the counter-premise
      is disproven by the protagonist acting on the truth, not merely a
      physical defeat.

# The Controlling Idea

Robert McKee's *Story* frames a story's theme as a **controlling idea**: a
value plus its cause, stated as one sentence — e.g. "Justice triumphs because
the hero is smarter than the criminals" (paraphrased concept, not McKee's
words). Theme *words* ("justice", "redemption") can't guide a scene decision;
a controlling idea can, because it names both the value at stake and what
causes it to triumph or fail.

## The sentence

A controlling idea has this shape:

> **{Value} {triumphs | fails} because {cause}.**

- **Value** — the story's core moral value, phrased as a continuum with a
  positive and negative pole: justice/injustice, love/hate, courage/cowardice,
  honesty/deception.
- **Triumphs or fails** — the ending's verdict. A positive ending argues the
  value triumphs; a negative ending argues it fails (or a counter-value wins).
- **Cause** — the protagonist's action or quality that *produces* the verdict.
  This is what makes the idea usable: every scene can be checked against
  whether the protagonist is moving toward or away from the cause.

Examples (original constructions in the form, not quotations):

- "Mercy triumphs over vengeance because the protagonist refuses the killing blow."
- "Honesty fails because the protagonist protects the lie that keeps them safe."

## The counter-premise

Record the **counter-premise** alongside the premise: the antagonist's (or
the world's) opposing claim, in the same value+cause form:

- Counter-premise: "Vengeance triumphs because the world punishes mercy."

The counter-premise is what the antagonist believes and acts on (see
`antagonist-design.md`). The story's climax stages the two premises against
each other and delivers a verdict — that verdict is the controlling idea's
"triumphs/fails" pole.

## Working-premise workflow

Craft opinion is genuinely split on planning theme up front (King warns
against starting from theme; McKee and Truby plan it). The repo's stance: use
a **working premise**, revisable.

1. **Plan:** at story-init (or when picking up an existing project), draft a
   working premise and counter-premise. Store in `story.md` frontmatter:

   ```yaml
   premise: "{Value} triumphs because {cause}."
   counter-premise: "{Opposing value} triumphs because {opposing cause}."
   ```

   If the user resists planning theme, record `premise: tbd-discovery` and
   let the revision audit (see `theme-audit.md`) build it from what the draft
   actually argues.

2. **Use during drafting:** the premise answers scene-level questions —
   "which choice tests the value?" A scene that never touches the value
   continuum is structurally decorative; flag it for the scene-craft sequel
   (the sequel is where the character's lie confronts the value).
3. **Audit in revision:** run `theme-audit.md`. The working premise is not
   sacred — if the draft argues something truer, rewrite the premise to match
   the draft, then revise the ending so it *delivers* the verdict cleanly.

## Premise quality check

- [ ] One sentence, with a named value and a named cause.
- [ ] The cause is something the *protagonist does*, not something that
      happens to them.
- [ ] The counter-premise is a genuine belief someone could hold (test: can
      you imagine the antagonist defending it?).
- [ ] The climax can plausibly stage both premises and pick a winner.

# Lie, Truth, Ghost Wound, and Arc Types

The planning machinery behind character arcs: the false belief that drives
the character, the truth that resolves it, the wound that planted the lie,
and which of the three canonical arc types the character follows.

## The lie and the truth

- **The lie** is the character's false belief about themselves, other people,
  or the world — e.g. "Trusting people gets you killed," "I am only valuable
  when I win," "Love is control." The lie generates:
  - **Flaw:** the behavior the lie produces (withdrawal, cruelty, control).
  - **Fear:** what the lie says will happen if they act differently.
  - **Want:** the external goal the character pursues *because* of the lie
    (the wrong thing they think will fix their life).
- **The truth** is the belief that replaces the lie — e.g. "Trust is a risk
  worth taking." The truth produces the character's **need**: the internal
  change the story actually demands.
- **The ghost wound** is the backstory event that planted the lie — the
  formative betrayal, loss, or failure. One specific scene, not a vague
  childhood. Record it concretely: who, what happened, what the character
  concluded.

## Arc types

The character template's "starting state → turning points → ending state"
field assumes a positive change arc. Three canonical types exist; record
which one each significant character follows, because the arc type dictates
what the turning points must contain.

**`change-positive`** — the character starts believing the lie and ends
believing the truth. Turning points are the moments the lie costs them, the
crisis where the lie fails completely, and the climax where they act on the
truth. This is the default; if unsure, use this.

**`change-negative`** — the character clings to the lie and it destroys them
(or leaves them diminished). Turning points: opportunities to choose the
truth, each refused; the lie's victories getting darker; the climax where
they double down on the lie and pay the full price. The negative arc still
needs the truth clearly visible — the reader must see the door the character
refuses to walk through.

**`flat`** — the character already holds the truth; the *world* changes
around them. Turning points are **tests of steadfastness**, not growth beats:
pressure to abandon the truth, temptations to adopt the lie, costs paid for
holding firm. The climax proves the truth by the character's refusal to
break, and the world (or the supporting cast) changes in response. Flat arcs
without escalating tests read as static — each test must cost more than the
last.

## Frontmatter fields

The character template (`character-management/references/character-template.md`)
carries these fields — fill them in for significant characters:

```yaml
arc-type: change-positive   # change-positive | change-negative | flat
lie: "{The false belief}"
truth: "{The belief that resolves it}"
ghost-wound: "{The specific backstory event that planted the lie}"
```

The protagonist must have all four fields. Supporting characters get them
when their arc intersects the theme; minor characters don't need them.

## Guidance for turning points

When drafting or revising a character's turning points, check them against
the arc type:

- `change-positive`: does each turning point make the lie more expensive?
  Does the climax require acting on the truth at real cost?
- `change-negative`: is the truth visible and reachable at each turning
  point? Does the character *choose* the lie each time (not merely fail to
  see the truth)?
- `flat`: is each turning point a genuine test — could the character
  plausibly break? Do the tests escalate? Does someone or something else
  change *because* they held firm?

# Motif and Symbolism

A motif is a recurring concrete element — an object, image, phrase, or action —
that accrues meaning across the story. Unlike theme (an idea), a motif is a
*thing the reader can see*. This reference covers the structural mechanics:
planting, varying, and paying off.

## Plant-and-vary

1. **Plant** the motif early in a literal, unremarkable form: a cracked
   teacup on a shelf, a phrase the mentor repeats, a door that never opens.
   At planting, it should read as set dressing or a character quirk — not as
   a symbol with a neon sign.
2. **Vary** it across the story: the motif reappears in changed contexts —
   the teacup gets used, breaks, gets glued, gets thrown. Each appearance
   should feel *earned by the scene*, not scheduled by the author. If a
   motif appears only on a fixed schedule, it reads as mechanical.
3. **Vary with the character's arc:** the motif's state should rhyme with the
   protagonist's relationship to the lie/truth. The motif is strongest when
   its transformations track the character's (see `lie-truth.md`).

## Object-symbol resonance with the ending

The motif's final appearance should resonate with the story's resolution:

- The cracked teacup, glued, holds tea at the celebration — the lie repaired,
  not erased.
- Or it shatters beyond repair — the cost of the victory made concrete.

The ending appearance is the **payoff**: the reader should feel the earlier
plantings click into place. If the motif's final appearance could be deleted
without loss, it was decoration, not structure — either give it an ending
resonance or cut the earlier appearances.

## Tracking motifs

Maintain a motif table in the relevant arc file or a `continuity/motifs.md`
ledger (one file per project is enough for most stories):

| Motif | Planted (Ch) | Variations (Ch) | Ending resonance (Ch) | Status |
|-------|--------------|-----------------|------------------------|--------|
| {kebab-id} | Ch 2 | Ch 7, Ch 14 | Ch 21 | planned / planted / paid-off |

- Motif ids are kebab-case; link the motif to the theme(s) it serves.
- The theme audit (see `theme-audit.md`) checks that every planted motif is
  paid off or deliberately left open with a recorded reason.

## Restraint rules

- **Two or three motifs per novel** is plenty. More dilutes all of them.
- A motif the narrative *explains* dies. Never have a character say what the
  teacup "means." Meaning lives in recurrence and context, not in
  commentary (see `theme-audit.md`: theme through consequence, not
  commentary).
- Motifs must be **concrete and perceivable** — an object, an image, an
  action. "The idea of freedom" is a theme, not a motif. "The unopened
  letter" is a motif.

# Theme Audit

A revision-time audit that checks whether the story *dramatizes* its theme —
through choices, consequences, and the ending's verdict — rather than merely
declaring it. Run this after a complete draft exists, as part of a
developmental revision pass (see the `revision-continuity` skill).

## Audit procedure

1. **Recover the working premise.** Read `premise:` and `counter-premise:`
   from `story.md` frontmatter. If `premise: tbd-discovery`, draft the
   premise the manuscript *actually argues* (value + cause, one sentence —
   see `controlling-idea.md`) before auditing. Record the recovered premise
   in `story.md`.
2. **Trace the value through the draft.** For each act (or each third of the
   chapters), note one scene where the protagonist's choice engages the
   value continuum, and whether the scene *cost* the protagonist something.
   Theme without cost is theme stated, not dramatized.
3. **Check the ending against the opening.** The opening poses the
   value-question (usually unknowingly: the protagonist acts on the lie).
   The ending must *engage* that question — not necessarily answer it
   cleanly, but the climax must stage premise vs. counter-premise and deliver
   a verdict the preceding story earned.
4. **Check the antagonist.** Re-read `antagonist-design.md`'s checklist
   against the draft: did the antagonist's counter-premise get a fair
   hearing? Did they win scenes? Is their climax defeat a *thematic* defeat?
5. **Check the motifs.** Every row in the motif ledger (see
   `motif-symbolism.md`) must be `paid-off`, or have a recorded reason for
   being left open.
6. **Check the arcs.** For each significant character with an `arc-type`
   (see `lie-truth.md`), verify the turning points contain what the arc type
   requires — growth beats for change-positive, refused truths for
   change-negative, escalating tests of steadfastness for flat.

## The five audit questions

Answer each in one or two sentences, with chapter references:

1. **Does the ending engage the opening's value-question?** Quote the
   opening's implicit question and the ending's verdict. If the ending
   resolves a different question, either the premise or the ending is wrong.
2. **Is theme dramatized through consequence, not commentary?** List the
   three most expensive choices the protagonist makes. If the theme is
   mostly delivered in dialogue speeches, narration summaries, or a mentor's
   lecture, flag it: move the theme into choices with costs.
3. **Is the motif paid off?** Every planted motif needs its ending
   resonance; see the motif ledger.
4. **Does the climax stage premise vs. counter-premise?** The final
   confrontation must be *about* the value, not adjacent to it. A climax
   won by superior firepower after a thematic story is a broken promise.
5. **Would the story collapse without the theme?** If you can remove the
   premise and the plot still works, the theme is decoration — wire the
   value into the protagonist's decisions or cut the pretense.

## Recording findings

Write the audit as `continuity/theme-audit.md` (one file per project,
updated each revision round) with the premise, the five answers, and a
**verdict**: `theme-holds` (minor fixes) or `theme-broken` (structural
revision needed, with the specific scenes to rework listed). If the audit
changes the premise, update `story.md`'s `premise:` field and the theme
tracking in `plot/_index.md`, then run `story reindex .` and
`story validate .`.