Identifies the specific ways AI strips characters of psychological complexity, turning them into functional story-delivery devices instead of believable humans.

## Key Points

- Characters feel "fine" but forgettable
- Every character seems reasonable and emotionally articulate
- Antagonists are cartoonishly evil or turn good too easily
- Reader feedback says "I didn't care about the characters"
- The protagonist never makes a genuinely bad decision
- Supporting characters feel like they exist only when the protagonist is in the room
- Character faces conflict → reflects for one paragraph → "realizes" the truth → changes behavior immediately
- Character is told something painful → initial resistance of 2-3 sentences → acceptance and growth
- Two characters argue → one says something wise → both understand each other perfectly
- Face conflict → react badly → double down on the wrong approach → face consequences → partially understand → backslide → eventually learn (or don't)
- Get told something painful → reject it → prove the other person right through their actions → maybe eventually acknowledge it chapters later
- Argue → both walk away misunderstanding each other → the argument festers and affects unrelated scenes

Character Flattening Detector

Identifies the specific ways AI strips characters of psychological complexity, turning them into functional story-delivery devices instead of believable humans.

When to Use This Skill

• Characters feel "fine" but forgettable
• Every character seems reasonable and emotionally articulate
• Antagonists are cartoonishly evil or turn good too easily
• Reader feedback says "I didn't care about the characters"
• The protagonist never makes a genuinely bad decision
• Supporting characters feel like they exist only when the protagonist is in the room

---

The AI Character Flattening Patterns

Pattern 1 — The Instant Epiphany

AI characters understand their problems the moment they encounter them. Real people resist insight, misinterpret their feelings, solve the wrong problem, and need the lesson beaten into them over time.

What AI does:

• Character faces conflict → reflects for one paragraph → "realizes" the truth → changes behavior immediately
• Character is told something painful → initial resistance of 2-3 sentences → acceptance and growth
• Two characters argue → one says something wise → both understand each other perfectly

What real characters do:

• Face conflict → react badly → double down on the wrong approach → face consequences → partially understand → backslide → eventually learn (or don't)
• Get told something painful → reject it → prove the other person right through their actions → maybe eventually acknowledge it chapters later
• Argue → both walk away misunderstanding each other → the argument festers and affects unrelated scenes

Diagnostic: Search for "realized," "understood," "saw clearly," "finally knew" in narration. Count how many of these lead to immediate behavioral change. If more than 50% do, the characters are flattened.

Pattern 2 — Universal Emotional Intelligence

AI gives every character the vocabulary and insight of a therapist. Every character can name their feelings, articulate their needs, and have productive emotional conversations.

What AI does:

• "I think I was projecting my fear of abandonment onto you"
• "What I need from you right now is space to process this"
• "I realize my anger isn't about the dishes — it's about feeling unseen"

What real characters do:

• Slam a door instead of saying what's wrong
• Pick a fight about something trivial when the real issue is too scary to name
• Say "I'm fine" when they're not, and genuinely believe it
• Express love through acts rather than words, clumsily
• Not know why they're upset until someone else points it out (and then deny it)

Diagnostic: Read each emotional conversation. Ask: could a real person in this situation, with this education level, in this emotional state, actually say these words? If the answer is "only if they'd been in therapy for five years," the dialogue is AI-flattened.

Pattern 3 — The Reasonable Protagonist

AI protagonists rarely make genuinely bad decisions. When they do, the narrative quickly validates them or the consequences are minimal. AI is trained to be helpful and non-offensive — this bleeds into its characters.

What AI does:

• Protagonist considers a bad option → internal voice warns them → they choose the wise path
• Protagonist makes a mistake → it turns out to be the right call in disguise
• Protagonist is angry → but channels it "constructively"
• Protagonist faces a moral dilemma → finds a third option that satisfies everyone

What real characters do:

• Make the stupid choice because they're tired, angry, drunk, in love, or just human
• Double down when caught rather than admitting fault
• Choose selfishly and rationalize it convincingly
• Hurt people they love and not immediately feel remorse
• Solve a problem in a way that creates a worse problem
• Win by doing something morally questionable and having to live with it

The bad decision audit: List every decision the protagonist makes. How many are wrong, selfish, shortsighted, or destructive in a way that has lasting consequences? If fewer than 25%, the protagonist is AI-sanitized.

Pattern 4 — Decorative Flaws

AI gives characters "flaws" that are actually charming quirks or that never cause real consequences.

AI flaws (decorative):

• "Stubborn" but only when it's admirable perseverance
• "Too honest" but it always endears them to others
• "Workaholic" but they always show up for the important personal moments
• "Has trust issues" but trusts the love interest by chapter 8
• "Sarcastic" but never in a way that actually hurts someone's feelings
• "Impulsive" but their impulses always lead to adventure, not disaster

Real flaws (consequential):

• Stubborn to the point of destroying a friendship and never apologizing
• Honest in a way that's actually cruel and makes people avoid them
• Workaholic who misses their kid's recital and tells themselves it doesn't matter
• Trust issues that sabotage the relationship and the love interest eventually gives up
• Sarcasm that masks contempt and poisons the room
• Impulsive in a way that lands them in jail/debt/a hospital

The flaw consequence test: For each stated character flaw, list every scene where it causes actual harm to the character or someone they care about. If a flaw never causes real harm, it's decoration.

Pattern 5 — Supporting Cast as Satellites

AI treats supporting characters as functions rather than people. They exist to serve the protagonist's arc and have no independent life, desires, or story.

Signs of satellite characters:

• They only appear when the protagonist needs something (advice, comfort, conflict)
• They have no goals that conflict with the protagonist's goals
• Their backstory only exists to create parallels with the protagonist's story
• They don't change over the course of the novel
• They could be replaced by a different character without the story changing
• They never do anything surprising — every action serves the plot
• When two supporting characters interact, they talk about the protagonist

The offscreen test: Does this character have a plausible life when they're not in a scene with the protagonist? Could you write a paragraph about what they did last Tuesday? If not, they're a satellite.

Pattern 6 — Emotional Convenience

Characters feel whatever the scene needs them to feel, regardless of their established psychology.

What AI does:

• A guarded character suddenly opens up because the scene needs vulnerability
• An angry character calms down because the scene needs resolution
• A character who's been betrayed forgives because the scene needs reconciliation
• A character who's afraid becomes brave because the plot needs them to act
• Characters grieve for exactly one scene, then function normally

What real characters do:

• Stay guarded long past when the reader wants them to open up
• Stay angry for chapters, taking it out on unrelated people
• Say they forgive but act cold, or forgive sincerely but then remember at the worst moment
• Stay afraid and find a workaround, or act despite fear in a sloppy, desperate way
• Grieve in waves that hit unexpectedly, months after the loss, in the middle of something ordinary

Pattern 7 — The Clean Arc

AI gives every character a tidy growth arc: flaw → challenge → growth → resolution. Real character development is messier.

AI arcs:

• Character starts closed off → meets love interest → opens up → happy ending
• Character starts ambitious → faces setback → learns what matters → balanced ending
• Character starts angry → confronts source of anger → heals → peaceful ending

Real arcs:

• Character starts closed off → opens up → gets hurt → closes off harder → slowly, reluctantly opens one crack → story ends with them still mostly closed but trying
• Character starts ambitious → faces setback → becomes more ambitious in a darker way → achieves the goal but it's hollow → no clean lesson
• Character starts angry → tries to deal with it → fails → the anger becomes part of them → they learn to be angry and still functional, never "healed"

---

Character Depth Scoring

For each named character with 5+ scenes, score on these dimensions:

Dimension	Score 1-5	What 1 Means	What 5 Means
Contradiction		Always consistent	Holds genuinely opposing beliefs/desires
Bad decisions		Never wrong	Makes costly mistakes driven by character
Flaw consequence		Flaws are charming	Flaws cause real damage
Emotional realism		Therapist-articulate	Messy, inarticulate, misguided
Independence		Exists for protagonist	Has own offscreen life and agenda
Growth messiness		Clean arc, clear lesson	Growth is partial, backsliding, hard-won
Surprise factor		Every action predictable	Does things that make sense in retrospect but weren't expected

Total: /35

• 28-35: Rich, complex character
• 21-27: Functional but could be deeper
• 14-20: AI-flattened — needs significant work
• 7-13: Plot device, not a character

---

Output Format

# Character Flattening Report
**Title**: [Novel title]

## Character Depth Scores

| Character | Contradiction | Bad Decisions | Flaw Consequence | Emotional Realism | Independence | Growth Mess | Surprise | Total |
|-----------|--------------|---------------|-----------------|-------------------|-------------|------------|---------|-------|
| ... | ... | ... | ... | ... | ... | ... | ... | .../35 |

## Detailed Analysis Per Character

### [Character Name] — Score: [N]/35
**Primary flattening pattern**: [which pattern is most dominant]
**Specific evidence**: [scene references]
**Recommendation**: [what to add, change, or deepen]

## Manuscript-Level Patterns
**Dominant flattening mode**: [which pattern appears most across characters]
**Characters most in need of work**: [ranked list]
**Quick wins**: [small changes that add the most depth]

---

Anti-Patterns

• Demanding that every character be morally grey. Some characters are genuinely kind. The test isn't moral ambiguity — it's psychological complexity. A kind person can be complex.
• Confusing "unlikeable" with "deep." Making a character a jerk doesn't make them complex. Complexity comes from contradictions the reader can understand, not from randomly negative traits.
• Ignoring genre conventions. Romance protagonists need to be somewhat likeable. Thriller protagonists need to be somewhat competent. Work within genre while adding depth.
• Rewriting characters instead of auditing them. This skill diagnoses. Rewriting is a separate step that requires the author's creative vision.
• Applying film/TV standards to novels. Novels have space for internal complexity that film can't show. Use that space — but don't mistake internal monologue for depth.