lukejoneslj/comprehensive-scripture-search

Comprehensive Scripture Search lets someone search six religious texts at once using meaning based search instead of exact keyword matching. The texts covered are the Book of Mormon, Doctrine and Covenants, Pearl of Great Price, the New Testament, the Old Testament, and the Quran. A user can type a topic or phrase, like "forgiveness of enemies" or "the creation of the world", and the tool returns the passages across all six texts that are closest in meaning, even if those exact words never appear. Under the hood, every verse in all six texts, 48,231 of them in total, has already been converted into a numerical representation called an embedding, using a model called Qwen3-Embedding-0.6B. An embedding captures the meaning of a piece of text as a set of numbers, so texts with similar meaning end up with similar numbers. When someone types a search, their query is converted into the same kind of numbers and compared against every verse to find the closest matches. The tool also supports looking up a specific reference, such as a chapter and verse, and comparing it against passages in the other texts, plus searching at the level of a whole chapter rather than a single verse. There is a hosted version available with no installation required, running as a free Hugging Face Space. To run it locally, the project needs Git LFS installed because the pre-computed embeddings file is large, then a standard pip install and a Python script to start a local Flask server. A Docker option is also provided for anyone who prefers a container based setup. The scripture text itself comes from two separate open source projects, and the README notes that each source carries its own license that should be checked before the text is redistributed elsewhere.