Extract, review, and query the facts inside your manuscript. Track characters, relationships, and events across every chapter — then catch contradictions before your readers do.
What it does
Built for authors who want to know their manuscript as well as their readers will.
Every extracted fact carries a source excerpt and chapter provenance. Review, approve, or reject — nothing enters canon without your sign-off.
See how every character and location connects. Scrub the ordinal slider and watch the graph rebuild at any point in your story.
Chat with any character — they only know what they know at the chapter you set. Spoiler-safe by design, provenance-grounded by default.
Paste a passage and get a contradiction report against confirmed canon. Catches relationship conflicts even when the phrasing is different.
A zero-data-entry heat grid showing which entities appear in which chapters — built entirely from extracted facts, no manual tracking.
Share a spoiler-safe reader encyclopedia for any project. Visitors set their own bookmark; facts beyond it stay hidden.
How it works
No selling, no sharing, no training on your content — ever. Your manuscript is your livelihood, not our dataset.
Walk away whenever you like: one click exports everything you own — originals, prose, canon with provenance, chat transcripts, the audit log. And delete means delete: a purge, not a soft-hide, verified down to zero rows. Bring your own provider keys and your text never touches ours.
Three steps from raw manuscript to a queryable canon store.
Drop a docx, md, or txt file. Canon detects chapter breaks automatically — adjust the split preview before committing.
An LLM pass extracts entities, facts, relationships, and dialogue with source excerpts. You accept or reject — nothing is auto-confirmed.
Chat with characters, scrub the relationship graph, run contradiction checks, or export an Obsidian vault — all from confirmed canon.