After Forty Chapters, the King Has Had Four Names.
A reader emails. The King's name is spelled "Arlen" in chapter two and "Arnen" in chapter seven. You search the manuscript. You find four spellings, and you did it to yourself. Long projects drift, though — it's the specific failure mode a worldbuilding-heavy series gets into, and it's what Lorekeeper was built to catch. It reads your Legendry and your manuscript looking for contradictions, and tells you where both versions live.
Your Consistency Guardian
Lorekeeper reads your Legendry and manuscript and works through the world looking for contradictions, anachronisms, and logical inconsistencies. Results arrive in real time as it finds them, so you're not waiting on a batch job to finish before you can see what it caught.
- Consistency Scoring
- Every analysis produces a 0–100 consistency score. Tracking it over time is how you watch the world tighten as you resolve contradictions and fill in gaps.
- Contradiction Detection
- Cross-references your Legendry entries, manuscript text, and timeline events to find statements that don't agree with each other. "The capital fell in year 300" in one entry and "year 350" in another? Caught, with both locations cited.
- Issue Lifecycle
- Issues move through states: open, acknowledged, resolved, dismissed. Nothing gets lost in a wall of warnings you already know about, and you don't have to re-evaluate the same flagged passage three sessions in a row.
- Depth Levels
- Three analysis modes. Quick scan catches surface contradictions. Standard runs a full cross-reference check. Deep runs full semantic analysis with entity-level reasoning. Pick whichever depth fits the time you've got.
Pattern Analysis Across Your World
Machine learning algorithms that read your world at scale. Your judgment still decides what to act on — the analysis just gives you a lens that reveals patterns you can't see by reading a thousand lore entries one at a time.
Issues arrive in real time with severity badges. Cross-references your Legendry, manuscript, and timeline to find contradictions.
- Topic Clustering
- TF-IDF analysis groups your lore entries by topic similarity. You see which areas of your world are densely developed and which are thin, without having to read every entry yourself.
- Anomaly Detection
- Isolation Forest algorithm flags entries that don't fit their cluster. An entry about "economic trade policy" sitting inside your "combat techniques" category gets surfaced, which usually means it's miscategorized or the category drifted.
- Coverage Analysis
- Maps how thoroughly different aspects of your world are actually developed. Identifies the imbalances: fifty character entries and three location entries means your world might feel populated but placeless. Harder to notice without the count in front of you.
- Maturity Scoring
- Rates each Legendry entry on completeness: filled fields, relationship density, cross-references, voice profile depth. A ranked list of which entries need more development, so you know where to spend the next worldbuilding session.