The continuity review is a 20-point analysis that reads your entire series against one shared canon at once, runs in dependency-based waves, and hands back a severity-ranked verdict tied to the exact book and chapter where each issue lives. It only ever reads your prose; you write every word.
Most analysis tools read one book in isolation. That is fine for a standalone novel, but a series works differently. The slip that matters most is rarely inside book three; it sits between book one and book three, separated by a couple hundred thousand words and a year and a half of your life. To catch a drifted eye color or a resurrected minor character before a reader does, the analysis has to hold the whole shelf in its head at the same time.
What follows is what the twenty analyses look at, the order they run in, and what lands on your desk at the end.
A single-book tool builds a fresh picture of your world every time it opens a file. Book three knows nothing about book one. So when your hero's eye color shifts, when a character who died in book two answers the door in book four, or when a sword you established as one-of-a-kind shows up in two hands at once, a single-book read simply cannot see it. There is no shared memory to compare against.
StoryHelm starts from the opposite assumption. Every book points at one Canon, a structured source of truth for your characters, places, factions, items, and events. The analysis reads each manuscript against that shared Canon and against every other book in the Series Atlas, so a detail established in chapter two of book one is still on the table when the review reaches chapter forty of book five.
The contradiction that matters most is rarely inside one book. It is between books.
The review is twenty distinct analyses. Each is a focused pass with a single job, and each writes into the same shared picture so the next pass can build on it. Here they are, grouped by what they examine.
The Structure Analyzer maps chapter and act boundaries across the full text; the Outline Extractor turns that into a clean scene-by-scene skeleton you can actually read.
The Manuscript Extractor pulls every character, place, faction, item, and event into your Canon, so the world is described once and checked everywhere.
Plot DNA reads intensity, stakes, beats, thread coverage, and filler ratio per scene; the Scene Content pass gives every analysis after it real scene text to work from.
The Narrative Analyzer and the Coherence Guardian run live continuity checks for name, timeline, and trait contradictions, the heart of the series read.
The Foreshadowing pass traces every promise to its payoff across books; the Dialogue extractor captures who says what, the raw material for the voice checks.
The Cross-Validator reconciles findings against each other; the Timeline extractor orders events chronologically so age, season, and sequence line up across the series.
Voice Consistency tracks whether a character still sounds like themselves from book one to book six; the Style Coach flags overused words and clunky sentence patterns, and the Sensory Audit checks which of the five senses each scene leans on.
The World Rules extractor captures the rules of magic, tech, and society; the Validator confirms the gathered findings are sound, with an Optimizer pass if quality runs low.
Narrative Synthesis pulls the separate findings into a single read on the story; Diagnostic Synthesis assembles the severity-ranked report you actually open, with book and chapter references.
A Supervisor coordinates the whole run, deciding which pass goes when and what each one depends on, so the analyses arrive in the order they can actually use.
This continuity review is the part of StoryHelm most authors meet first, because it answers the question behind every long series: by book five, does the world still agree with itself?
The analyses are not independent. A timeline check is meaningless until events have been extracted; a foreshadowing trace needs scenes in hand before it can follow a promise to its payoff. So the review runs in dependency-based waves: each wave produces facts the next wave depends on.
Structure first, then the outline, the manuscript extraction, and Plot DNA. The skeleton and the cast have to exist before anything can be checked against them.
Scene content feeds the Narrative Analyzer, the Coherence Guardian, foreshadowing, and dialogue. This is where most cross-book contradictions surface.
Cross-validation, timeline ordering, and the voice, style, and sensory passes reconcile and deepen the picture, then world rules and validation confirm it.
Narrative and diagnostic synthesis assemble everything into one severity-ranked report. An optimizer pass runs only when the review's own quality score comes in low.
Because the work is ordered by dependency rather than crammed into a single prompt, each pass examines one thing closely with everything found before it already in hand.
The output is not a wall of raw data. It is a severity-ranked verdict on your series, anchored to the exact place each issue lives.
You can take that verdict deeper through the craft lenses, or run it as a one-time check. The same engine powers the one-time $99 Manuscript Audit if you want a single severity-ranked read without a subscription, and it runs continuously on the Series tier as you write.
StoryHelm keeps your whole series in one place. You write and edit your books in its built-in Scene Editor, or import finished manuscripts written elsewhere, with your Canon and Series Atlas alongside them. From there the continuity review, one workflow in a 41-agent engine powered by Claude, reads your entire series against that shared canon and returns a severity-ranked verdict with book and chapter references. You write every word. StoryHelm makes sure the words agree with each other.
People get one thing wrong about this review: it is not a tool you run a finished book through once and then put away. It lives inside the workspace where you write. Draft a new chapter in the Scene Editor and the Coherence Guardian is already checking it against the rest of the series. The full twenty-analysis review is the deep, deliberate read; the live checks are the day-to-day. Same canon, same engine, one place.
Books written in another tool import cleanly. Once a book is in StoryHelm, it joins the shared Canon and the series read sees it.
No. It reads, checks, and analyzes. It never drafts, rewrites, or generates a sentence of your story. You write every word.
Twenty distinct analyses, each with a single job, running in dependency-based waves so later passes can build on earlier findings. They share one canon and assemble into one report.
Every book, against one shared Canon and the Series Atlas. Because the Canon persists from one manuscript to the next, a detail established in book one is still in memory when the review reaches book five, so a finding can name both the book where a fact was set and the book where it was broken.
Because the analyses depend on each other. A timeline check needs extracted events; a foreshadowing trace needs scenes. Running in waves lets each pass work with everything found before it.
Bring in book one or all six, and watch twenty analyses read the series against one canon and hand back the exact book and chapter where it stops agreeing with itself. Find the slip between book one and book five now, not in a one-star review after the box set ships.
Request a founding seat