A second reader who keeps the books on who knows what.
Mysteries fail on bookkeeping: a detective who acts on information she hasn't learned yet, a clue planted after the scene that pays it off, a timeline that can't survive a reader with a notepad. Scribegrove's canon tracks knowledge-state across your manuscript, and the whole-book Story Doctor surfaces contradictions anchored to the exact chapter and scene — without ever touching your misdirection.
1 million tokens on sign-up · 150,000 more every month · no credit card · You never have to subscribe — buy tokens and pay for what you need. · Never trained on your manuscript
Actual product — captured from the live web app.
What this tool does for you specifically.
Canon tracks who knows what — and when
Characters, facts, and rules live as structured canon. Record what each character knows and when they learned it, and Grove reads that record when reviewing your manuscript — so a detective acting on chapter-20 knowledge in chapter 12 gets flagged, not published.
Whole-book contradiction scans with anchors
The Story Doctor reads the entire manuscript and surfaces knowledge and timeline contradictions: an alibi that conflicts with an earlier scene, two events that can't both fit the same evening, a reveal that depends on a clue you cut in revision. Every finding links to the exact chapter and scene.
Outline beats hold the clue trail
Plot in beats before you draft: where each clue is planted, where the red herring lands, where the reveal pays off. Your beats stay attached to the book as you write, so the fair-play ledger — clue in, payoff out — is a structure you can see, not a memory you have to trust.
Preview-then-apply — misdirection stays yours
Grove never rewrites without permission. An AI that "fixes" your unreliable narrator or smooths out a deliberately misleading scene has ruined your book. Every suggestion is a margin note you can preview and decline; the deliberate lie you planted in chapter 3 is safe.
One plan — or pay as you go.
Same product on both paths. One $14.99/mo plan — The Writer — covers the whole book, or buy token packs from $5 and never subscribe.
ONE PLAN
The Writer
One plan for the whole book — write, edit, export, publish.
- 3M weighted tokens every month
- Exports and publishing free
- Images 20k tokens each
- Unused tokens: 25% carries into next month
- Cancel anytime
NO SUBSCRIPTION
Pay as you go
You never have to subscribe — every step just costs tokens.
- Token packs: $5 / 500k · $15 / 2M · $30 / 5M
- Purchased tokens never expire
- Any purchase unlocks exports + publishing (25k each)
- Images 20k tokens each
What mystery + thriller authors usually ask first.
How does the who-knows-what tracking actually work?
Your canon holds characters, facts, and rules as structured records — including what a character knows and when they learned it. Grove reads that canon when reviewing the manuscript, so knowledge-state violations surface as findings: a character referencing information they haven't been given, or two characters sharing something they had no scene to share. You maintain the ledger; Grove audits the prose against it.
Can it check my timeline?
The whole-book Story Doctor reads the full manuscript and surfaces timeline contradictions along with everything else: events that can't both have happened in the stated window, a character in two places at once, a wound that heals overnight. Every finding is anchored to the exact chapter and scene so you can judge it in context. It's a reader who takes notes, not an oracle — you make the call on each finding.
Will it "fix" my red herrings or unreliable narrator?
No — and this is a deliberate design line, not a limitation. Everything is preview-then-apply: Grove surfaces what looks like a contradiction, you decide whether it's a bug or your misdirection working as intended, and nothing changes until you apply it. A tool that auto-rewrites can't tell a plot hole from a deliberate lie. Scribegrove doesn't try; it asks.
I plot heavily before drafting. Does the tool support that?
Yes. You outline in beats — scene-level story units that hold your clue placements, red herrings, and reveals — before writing prose, and the beats stay attached to the book as you draft. If you're a discovery writer instead, you can import a finished draft (DOCX, TXT, MD, RTF) and blueprint review will extract characters, locations, and facts into an entity graph you preview before committing.
Is my manuscript safe? Thriller plots leak.
Your library is encrypted at rest and never used to train any model — your words stay safe and readable. Exports and publishing are free on The Writer plan, or 25k tokens each on pay as you go. When you're ready for beta readers, share links are watermarked and readers leave notes anchored to the chapter without needing an account — you control who sees the twist before publication.
What does it cost, and what's on which tier?
Two doors: The Writer at $14.99/mo with 3M tokens every month, exports and publishing free — or token packs from $5 and never subscribe. Every step (AI, images, exports) just costs tokens. Managed AI — no API keys, no model configuration. Free to start — 1M AI tokens, no card. 30-day money-back guarantee on unused subscriptions.