What it reads

The Corpus

Most AI products keep what shapes their answers hidden. Davar publishes its entire study corpus in the open, so any pastor, scholar, or sceptic can read exactly what the Scholar stands on.

What's in it

The Scriptures

The full King James Bible, with cross-references and word-level study data for the original Greek and Hebrew.

The Creeds

Apostles' · Nicene · Chalcedonian, the church's shared confession of who God is.

The Confessions

Historic Reformed confessions, giving the Scholar doctrinal guardrails older than any AI model.

Why this corpus exists

In plain terms: this corpus exists to point people to Jesus and to keep the Scholar's answers faithful to what Scripture and the historic church actually teach, not to any single person's opinion. It's built from one tradition today (evangelical), stated plainly so there's no confusion about what it does and doesn't cover yet.

Browse the corpus

Live from the repository's main branch. Pick a file on the left to read it.

README.md

Davar Corpus

A public, source-attributed collection of scripture and historic Christian doctrinal texts, structured for retrieval-augmented generation (RAG) and any other application that needs grounded, citable Christian source material — not just Davar.

Why this exists

AI tools answering questions about scripture and doctrine need to ground their answers in real, attributable sources rather than improvising from model weights. This repo is that grounding layer: public domain texts, one JSON file per source, each entry carrying a precise reference so any answer built from it can be traced back to something a human can go read and verify.

This repo is intentionally separate from any single application. The goal is for it to become something other projects — not just Davar — can depend on directly.

What's here

File	Contents	Entries
`data/apostles.json`	Apostles' Creed	5
`data/nicene.json`	Nicene Creed	6
`data/chalcedon.json`	Chalcedonian Definition	2
`data/heidelberg.json`	Heidelberg Catechism (1563)	129
`data/belgic.json`	Belgic Confession (1561/1619)	37
`data/dort.json`	Canons of Dort (1618–19)	59
`data/wsc.json`	Westminster Shorter Catechism (1647)	107
`data/wcf.json`	Westminster Confession of Faith (1647, American revision)	171

Scripture text (KJV) is not yet vendored here — see SOURCES.md for why, and the carry-forward at the bottom of this README.

Every source is public domain. See SOURCES.md for the full ledger — what each source is, where its text came from, and why it was chosen. See schema.md for the JSON shape each file follows.

Governance

This repo currently has a single maintainer and an open, transparent review process — not yet a formal council. See CONTRIBUTING.md for how to propose a source or flag an error, and the long-term plan for how this moves toward broader, cross-tradition governance as the project grows.

Status

Early. Built alongside Davar's retrieval pipeline but designed to outlive being tied to any one app.

Carry-forwards

Vendor a verified-license copy of the KJV text here (currently fetched live by consuming apps from a third-party mirror at ingest time — that mirror's license hasn't been independently confirmed).
Add a second commentary/dictionary source once one is chosen (see Davar's feature-roadmap.md for candidates under consideration).
Decide how consuming apps pull from this repo — git submodule, versioned npm/PyPI package, or plain fetch-by-URL.

README.md · 2.6 KB

How it's governed

Today

Curated, but transparent

Sources are chosen deliberately: Scripture first, then the historic creeds and confessions the church has tested for centuries. The review conversation happens in the open, in the repository above.

As we grow

Community-governed

A formal doctrine council, accountable to the body of Christ and spanning traditions, not just one, takes over curation once real usage makes that conversion meaningful.

Get involved

Help maintain the corpus

We're building a small group of reviewers, people who help check that what the Scholar reads is accurate and fairly represents the breadth of the church, not just one tradition. This isn't a queue to process, it's the start of a working relationship.

Already an active contributor and just want the repository directly? Open it on GitHub

The corpus is what it reads

VAIL, our own AI engine, is what reasons over it.

How VAIL works

Your own data is different

The corpus is public. Your conversations aren't, see how we treat them.

Read the AI policy

Ready to start studying?

Create a free account and see what the Augmented Scholar can show you.

Create free account