Read contract 1.0.0 · prepared release v5
Built to be read.
By people and machines.
Explore the same records through the visual encyclopedia or a structured, source-linked collection. A machine-readable claim is still a claim—not automatically a fact.
474 stable entry identities
313 historical records, 150 theory/guide entries and 11 creator-linked videos. Namespaced IDs survive title and category changes.
Download the structured collection →Sources with roles
449 distinct source URLs and 56 directed editorial connections. Each connection explains relevance; none is encoded as proof.
Read the evidence and provenance contract →Inspectable structure
Versioned schema, original field provenance, per-record hashes and release checksums. Missing review dates stay unknown.
JSON Schema → · Release manifest →Update without erasing history
This is the first normalized snapshot. Future exports can report added, changed and removed IDs against an earlier snapshot.
Initial change baseline → · Offline export script →Human review remains the gate.
No AI is automatically changing or publishing these records. No continuous freshness service is running. The export time is not a source-review date. Private discussion drafts, personal information and credentials are excluded.
A reusable foundation—not a finished universal protocol.
The structured export is the read contract. The current editorial JSON and pre-rendered pages still need a coordinated build; they are not yet a single authoring database. Source checks, rights decisions, revisions and publication need human review. Sparse related-entry links are not a complete knowledge graph.
Suggested update flow: propose a change → inspect the original → review rights and evidence role → regenerate and compare → validate human and machine views → approve → publish and check.
Browse historical originals · Inspect claims · Watch and take private notes