Choose a study face below or use the tabs to jump in.
Scripture Lab is a study workspace for questions, comparisons, canon browsing, and timeline exploration. Choose a face below or use the tabs above (bottom bar on mobile) to jump in. Your question, database selection, and last Ask or Compare results carry across faces for the session.
View biblical journeys on a map and scan chronological events across Scripture’s span.
Filter by date range when the filter is enabled in Settings.
Toggle the journey map under Settings → Face extras if you prefer a simpler list.
Sources, answers, and verification
Scripture Lab reads verse text and book metadata from PostgreSQL Bible databases. Every loaded translation is a public-domain text. Optional AI commentary is composed only from the verses retrieved for your question — never from the model’s memory alone.
How Ask answers are produced
Your question is matched against verses in the databases using keywords, topics, and explicit references.
The returned answer, explanation, and cited verses all come from that retrieved evidence pack.
When an OpenAI key is set, commentary is generated from the pack; a citation checker strips invented Book chapter:verse references.
Without a key, or when a prior verified answer exists, the app serves local or cached answers built from the same database evidence.
Where Scripture and books come from
Full-text Bibles are ingested from public-domain sources — midvash bible-data, eBible USFM, JPS 1917, Brenton LXX, and the Ethiopian 81-book Ge'ez corpus — into versioned JSON and PostgreSQL schemas.
The catalog lists tradition, language, and license for every referenced translation; copyrighted modern versions stay catalog-only and are not loaded as study text.
Whole-book introductions and chapter guides are curated study content stored in the repository (`data/book_explanations.json`, `data/book_chapter_explanations.json`) and served with the Books face.
How content is verified
Verse text is always read from the database at request time — not paraphrased from model training data.
Shared Ask answers are stored only after citation verification against the evidence pack.
Chapter guides are checked for structure, uniqueness, and corpus grounding (`npm run guides:review`).
Corpus builds and migrations are versioned so operators can validate releases before publication.