building · project

Olive & Flame

A free library and a quiet place to pray, built and run as Olive + Flame Ministries. Two applications, one corpus, no analytics script, and nothing about a reader sold or tracked.

Filed with the Iowa Secretary of State under Iowa Code ch. 501B — Business No. 826891, effective 3 May 2025.

Two products under one roof

The Lantern is an open library and reading room for people who read primary texts the way theologians, historians and linguists do — the original languages within reach, the commentators at hand, the whole tradition on one shelf. Twenty-three Bibles across English, Hebrew, Greek and Latin; the Church Fathers; the Beit Midrash; philosophy from Plato to Wittgenstein. Every word is free to read and no account is required to read it.

The Chapel is a virtual candlelit chapel, offered freely to all Christians.

Three decisions that shaped it

01

An openly licensed corpus, with no exceptions

Every source is openly licensed — Berean Standard, scrollmapper, Open Scriptures Hebrew Bible, STEPBible, Sefaria, CCEL, Standard Ebooks, Divinum Officium. Commentary is filtered to writers who died in or before 1900, so everything reprinted is public domain. No closed dependencies and no vendor lock-in, which means the library cannot be taken away by a licence renewal.

02

The text is frozen in version control

The house editions are built by a deterministic rule engine with every departure from the source disclosed on the page. Lantern Direct is vendored book by book, so any change to a single verse is a reviewable diff rather than a silent edit to a database. A word-level audit checks each rendering against its lemma’s own lexicon entry, and one signed ledger entry closes each book.

03

The Chapel is MIT so it can be copied

The Lantern is Apache-2.0; the Chapel is deliberately MIT, chosen so it can be mirrored freely by anyone wishing to serve persecuted Christians. That is a licence picked for what happens if the original goes away — the software equivalent of leaving the door unlocked.

A press that reads the database

The Lantern typesets its own books straight from the corpus: no manuscript, no layout program, just scripts that read the database and emit print-ready PDFs at KDP trim with matching full-wrap covers. Spine geometry is computed from the page count, so a regenerated interior always gets a cover that fits. Twelve titles ship this way, including the nine-volume Lantern Library and the four-volume Catena that sets each Gospel with the Fathers beneath every verse.

It also sets things that are not books — a pocket deck of thirty-six holy cards, imposed nine to a sheet with the backs mirrored for a long-edge duplex flip. Feast days are checked against the same calendar the daily Lectern reads, by the test suite rather than by eye.

How it is built

A pnpm-workspaces monorepo of two Next.js applications over shared core, corpus and modules packages — around 527 TypeScript files — on managed Postgres with pgvector, deployed to Render from a single render.yaml. The corpus syncs incrementally on deploy.

What is honestly not built yet

Cross-domain semantic memory is the design goal — every note embedded, so opening a passage surfaces your own prior thinking on it. The embeddings table and pgvector are provisioned and nothing is embedded yet; notes today are verse-anchored and searchable, not semantic. The module architecture is likewise the intended shape rather than the current one: packages/modules/ holds Theology alone, and the Mythology, Philosophy and History collections live as corpus shelves rather than separable modules.

Figures above are the project’s own, as recorded in its README on 24 August 2026.

Visit Olive & Flame ↗