Guides/Feature Workflows

The Story Bible

What the analysis finds in your story world, and where it goes

What the Story Bible Is ↓Running a Gap Check ↓What It Reports ↓The Check Over-Reaches, Predictably ↓Where Your Records Live ↓

What the Story Bible Is

A story bible, in the traditional sense, is the reference document a writer keeps beside the manuscript: who everyone is, where things are, how the world works. Most writers who keep one keep it by hand, which means it is accurate right up until the draft moves and nobody updates it.

Bookshaper keeps yours honest against the manuscript itself. The gap check reads your prose for the characters, locations, and relationships in it, compares them against the record you keep, and reports the difference — who is on the page and not on your list. Your records — the ones you edit, rename, and rely on — live in the Characters and Locations views under Structure, which is where a reference you own belongs.

That split is deliberate, and it is the same rule the rest of the app follows: analysis grades your manuscript, it does not write your project for you.

Running a Gap Check

A gap check spends credits, so it sits with the rest of the analysis tools under Analysis → Record Gaps. Nothing in Structure ever starts a paid run.

The first check needs something to read. Bookshaper asks for at least two thousand words of prose — roughly one full chapter — before offering it, because below that the pass reliably comes back thin or empty and the run is wasted. Structure alone does not help here: one rich chapter yields more than ten empty stub scenes.

It is an AI operation, so its credit cost is shown for confirmation before anything runs. Once a check has run you can re-run it at any time, and the button tells you when your manuscript has changed since the last one — a report built from a draft you have since rewritten is describing a book that no longer exists.

What It Reports

Only the difference. The page lists the characters, places and connections it found in the prose that are not on your record, grouped by kind, and each one comes with three answers: add it to your record, dismiss it if it is a passing mention you do not need to track, or defer it for a later pass.

Nothing is added unless you say so. Add to record is a button you press, on one item or a whole section, and it is the only thing in the app that writes to your cast and places — no run does, ever. That is the whole difference between a reference somebody else generated and one you keep.

Adding one also tags the scenes it appears in, using what the last analysis already wrote down. No re-run, no credits: the pass recorded every name it could not place against your record, so saying yes to a name is enough to put it back where it belongs.

It also distinguishes three outcomes a lesser page would collapse into one. A manuscript nothing has read yet gets no verdict. A run that found nothing to compare says so. And a genuinely complete record is only ever reported alongside the number of entities it was checked against, so an all-clear over nothing is impossible to mistake for good news.

The first whole-manuscript analysis on a brand-new project ends by offering to add everything it found in one action. It is offered once, never pre-selected, and declining it is a decision that sticks.
Until you add them, the views that draw on your cast — scene tags, the Timeline's POV threads, the relationship map, the Story Bible PDF, the marketing tools — say plainly that the check found people your list does not have yet, rather than reading as though your book has none.
Adding an entity — from the report, or by hand in Structure — closes its gap immediately. You never have to pay for another run to see that.
Scenes you have tagged yourself are left exactly as they are, and the confirmation tells you how many were skipped for that reason.
Analyzing the entire manuscript runs the same pass as part of a wider sweep. It reports what your record is missing; it does not fill it in.

The Check Over-Reaches, Predictably

It is reading prose without knowing which names matter to you, so anything that behaves like a person or a place on the page is a candidate — and a candidate you have deliberately left off your record comes back as a gap. Knowing the shapes it over-reaches on is what lets you read the report quickly, and dismissing one is a complete answer.

An over-reach has still told you something — that a name appears often enough in your prose to read as significant. Occasionally that is worth a second look at the passage rather than the name.

A name mentioned once in passing. A waiter with a line of dialogue is a character on the page and not a character in your book.
People inside in-world texts. A novel your protagonist is reading, a legend a character recounts, a play staged in chapter nine — their casts look exactly like yours to a reader working sentence by sentence.
Places that are figures of speech. Somewhere a character says they would rather be is not a location in your story.
The same person under two names. A character who goes by a nickname in dialogue can arrive twice. On your roster, keep the one you use and add the other spelling as an alias.
Historical or real-world references. A city named only as a comparison, or a public figure invoked in an argument, is background rather than setting.

Where Your Records Live

Characters and locations are yours, in the Characters and Locations views under Structure, editable like any record you created yourself — description, role, aliases, and the rest. Nothing the gap report says overwrites them.

The relationships an extraction finds live there too. Characters is tabbed — the roster itself, then Presence, then Relationships — and the Relationships tab lists every connection your project holds, alongside the characters no connection was recorded for at all. Connections run between two characters or between a character and a place, and you can add, edit and remove them yourself: an extraction proposes, you decide, and anything you write down is taken as read.

Those records feed the surfaces that read your roster. Both maps on the Presence tab — the one built from your own scene tags and the analysis's mention density — track the characters your project knows about, and the Timeline's point-of-view lens is built from them too.

For point questions about your story world — what colour were her eyes in chapter three, who was at the funeral — the in-app assistant answers from the manuscript directly and is usually faster than browsing any list.

← All guides Documentation FAQ

Character Presence & Arcs

Continuity & Timeline

Reading Analyze