Dracula
Fear assembled from letters and journals. A multi-document mystery in which every recorder sees a different part of the truth.
Alice's Adventures in Wonderland
Into the logic of a dream. Track identity through a world where body, scale, and name refuse to stay still.
The Time Machine
A leap across 800,000 years. The dinner-table frame and the Traveler's testimony run on two overlapping timelines.
The Adventures of Sherlock Holmes
A theater of deduction in fogbound London. Twelve cases trace concealed knowledge as it becomes visible truth.
The Iliad
An epic that begins in rage. Human vows and divine interventions collide across the battlefield of Troy.
The Measure of an Ego
From the moment Nova Nivalis begins to understand the world — the price of a world where loss is impossible.
Dracula Wiki (Curated)
Built by hand against the source text: the first public ontology, with search, graph views, and multiple timelines.
How it was built
- Blind extraction — AI sees only the current chapter text and extracts events, records, and states. Memory and summaries are forbidden.
- Anchor validation — Every event must carry a quoted anchor that resolves exactly once in the source. All 771 events pass.
- Precision filter — A claim that fails validation cannot enter canon. Missing information is acceptable; incorrect information is not.
- Incremental linker — The ontology is merged and compared chapter by chapter to find foreshadowing, reveals, and contradiction candidates. Only two-stage adjudicated results are published.
- As-of contract — Anything beyond the reader's current position remains locked. Spoiler prevention is an ontology principle, not a cosmetic filter.
Public read-only MCP
AI clients can query entities, event bundles, viewpoint records, reader snapshots, three timelines, and provenance layers directly. No write or delete tools are exposed.
https://ontology.onlynudles.com/mcp
● Live · Streamable HTTP · no authentication · read-only