Skip to content

Story Structure

Genre Rules

Apply genre-specific structural checks — auto-detected from keyword clusters.

What It Does

Auto-detects the genre of your text from keyword clusters, then applies genre-specific structural rules:

Genre Detection Rules applied
Mystery murder, detective, suspect, clue Clue density in Act I; clue trail continuity
Romance kiss, love, passion, embrace Relationship arc markers; HEA/HFN ending
Thriller danger, escape, chase, bomb Deadline presence; pacing acceleration in climax
Fantasy magic, spell, wizard, dragon Info-dump risk in Act I (low dialogue + worldbuilding)
Sci-Fi spaceship, galaxy, android, quantum Tech exposition tracking in Act I

Why It Matters

Each genre has reader expectations baked into its conventions. Mystery readers expect clues distributed throughout the narrative. Romance readers expect a Happily Ever After (HEA) or Happy For Now (HFN). Thriller readers expect accelerating pace in the final act. Violating these conventions isn't always wrong — but it should be intentional.

What Gets Flagged

Mystery: Missing Clues

Severity: Information / Warning

  • Few clues in the first third → readers expect early hooks
  • Clue trail goes cold after Act I → evidence mentions disappear

Romance: Weak Arc

Severity: Information / Hint

  • Limited relationship arc markers → strengthen the romantic progression
  • Ending lacks HEA/HFN signals → readers expect a satisfying resolution

Thriller: No Ticking Clock

Severity: Information

  • No deadline or temporal urgency detected
  • Pacing doesn't accelerate in the climax (final 20% should have shorter sentences)

Fantasy/Sci-Fi: Info-Dump Risk

Severity: Warning

  • Heavy worldbuilding (10+ genre keywords) with little dialogue (<3 instances) in Act I
  • Suggests grounding the reader in character before exposition

Configuration

No configuration options. Genre is auto-detected from keyword density (minimum 5 genre-keyword matches).

Technical Details

  • Source: prose-craft
  • Scope: Document-level (position-based sections)
  • Minimum text: 20 lines and 500 words
  • Method: Genre classification via keyword frequency; genre-specific rules activated per detected genre