How PlotAudit Works

Every audit runs your manuscript through a series of specialized passes. Each pass has a different job, and later passes build on the findings of earlier ones.

1. Topology — Maps the narrative structure (linear, flashback, multi-timeline) so downstream passes know how to track knowledge and time.
2. Extractions — Five parallel passes pull out characters, timeline events, spatial data, promise/payoff pairs, and curiosity gaps.
3. Story Bible — Synthesizes all extractions into a single ground-truth document: who, what, where, when, and what was promised.
4. Chapter Audits — Each chapter is individually audited against the Story Bible for factual consistency, continuity errors, and drafting residue.
5. Literal Logic — A cold arithmetic pass: do the distances, times, ages, and object counts actually add up?
6. Cross-Reference — Checks for systemic contradictions across the full manuscript, not just chapter-by-chapter.
7. Adversarial — A hostile reader pass that hunts for deus ex machina, unresolved hooks, convenience logistics, and emotional shortcuts. This pass runs twice per chunk and is genre-calibrated (see below).
8. Meta-Audit — A Senior Executive Editor synthesizes all findings, arbitrates conflicts between passes, and delivers the editorial verdict.
9. Manifest & Author Report — All findings are deduplicated, severity-ranked, and compiled into the structured report you see.

PlotAudit doesn't audit your book against the conventions of a different genre. When you set your genre, the entire pipeline calibrates to it.

How it works: Your genre label is injected into the adversarial pass, meta-audit, and final manifest generation. When the Story Bible classifies a pattern as genre-appropriate (e.g., a meet-cute in romance, a red herring in mystery, a serum confession as an inciting incident in speculative romance), the adversarial pass downgrades that finding to a NOTE instead of flagging it as HARD or CRITICAL.

The genre label is a hint. The Story Bible is the authority. If your Bible says "serum confessions are inciting incidents, not resolutions," the adversarial pass respects that regardless of the genre label.

Common effective labels: Contemporary Romance, Romantic Suspense, Cozy Mystery, Urban Fantasy, Epic Fantasy, Sci-Fi Thriller, Literary Fiction, YA Fantasy, Women's Fiction, Historical Romance, Paranormal Romance, Dark Romance, Psychological Thriller.

When your book belongs to a series, PlotAudit can pull continuity context from previous books. Mark a completed audit as "Series Canon" to make its Story Bible available to future audits in the same series.

The pipeline uses vector-similarity search to pull in only the relevant canon context for each pass — not the entire previous Bible. This keeps the audits focused and avoids overwhelming the model with irrelevant detail.

When you re-audit a book after making revisions, PlotAudit compares the new manuscript against the previous version chapter-by-chapter. Unchanged chapters reuse their previous audit results — only modified or new chapters are re-analyzed.

This makes re-audits faster and cheaper, and lets you verify that your fixes landed without re-running the entire pipeline from scratch.

Use the Re-Audit This Book button on any completed report to start an incremental re-audit with the previous audit pre-linked.

PlotAudit runs on audit tokens. Every account starts with a free trial token so you can run your first audit immediately. After that, choose the plan that fits your workflow.

Pro subscribers ($20/mo) can connect their own OpenRouter API key. When BYOK is active, audits bill directly to your OpenRouter account instead of consuming PlotAudit tokens.

How to set it up: Go to Settings, paste your OpenRouter API key (starts with sk-or-), and PlotAudit validates it against OpenRouter before saving. The key is encrypted at rest and never exposed to the browser after initial entry.

BYOK users can also select custom models for each pipeline pass — useful if you want to use a specific model for the adversarial or meta-audit pass.

The final report organizes findings by severity:

• Critical — Must-fix issues that break the story's logic or ending. Capped at 5 maximum.
• Moderate — Significant friction points that readers will likely notice.
• Minor / Note — Small observations, genre-acknowledged patterns, or polish items.

The verdict is one of: Structurally Sound (0 critical, ≤3 moderate), Needs Revision, or Significant Problems (≥3 critical or a fundamental logic failure).