The goal of a Study Bible project built with LLM-MAR is to create a collaborative AI system that helps users understand the Bible through multiple perspectives by using specialized agents that analyze scripture, explain historical and linguistic context, explore interpretations, and provide practical applications. By coordinating these agents through structured workflows, teams, and debates, the system aims to deepen understanding,and support more reflective, educational, and accessible engagement with biblical texts.
This project leverages LLM-MAR (Multi Agent Reasoning), a compact CLI that creates LLM agents, lets them debate, answers questions, and builds workflows, to facilitate multi-perspective biblical study.
The project currently focuses on the Greek New Testament and plans to expand to the Hebrew Old Testament in the future.
Repository: github.com/llm-mar/study-bible
The project follows a hierarchical structure organized by biblical books, chapters, and verses:
study-bible/
├── {book}/ # Book directory (e.g., matthew/, mark/)
│ ├── {chapter}/ # Chapter directory (e.g., 1/, 2/)
│ │ ├── {verse}/ # Verse directory (e.g., 1/, 2/, 3/)
│ │ │ ├── README.md # Formatted analysis with tables and sections
│ │ │ ├── raw.json # Original Greek text from OGNT
│ │ │ ├── teaching.json # Teaching and doctrinal analysis
│ │ │ ├── question.json # Questions and interpretive issues
│ │ │ ├── author.json # Author's perspective and intent
│ │ │ ├── audience.json # Original audience context
│ │ │ ├── noun.json # Greek noun analysis
│ │ │ ├── verb.json # Greek verb analysis
│ │ │ ├── phrase.json # Key phrase analysis
│ │ │ └── translation.json # Translation notes and variants
│ │ └── README.md # Chapter overview (when available)
│ └── README.md # Book overview (when available)
└── abbreviation.json # Biblical book abbreviations reference
Each verse directory contains comprehensive analysis generated by specialized LLM agents, providing multi-perspective insights into the biblical text.
The project uses specialized LLM agents to analyze different aspects of biblical texts. Each agent focuses on extracting specific types of information from the scripture.
For detailed information about each agent, their goals, instructions, and what they extract, see the Agents README.
The Greek New Testament text used in this project comes from the Open Greek New Testament (OGNT). You can view the original Greek text for any verse at:
https://opengnt.com/
The Hebrew Old Testament text used in this project comes from the Open Hebrew Bible Project. You can view the original Hebrew text and resources at:
https://marvel.bible/
Use the analyze.sh script to run the complete analysis pipeline for any verse:
./scripts/analyze.sh BOOK CHAPTER VERSE
Example:
./scripts/analyze.sh MT 1 1
This will:
This project uses the Open Greek New Testament text licensed under Creative Commons Attribution-ShareAlike 4.0 International License, the Open Hebrew Bible Project licensed under Creative Commons Attribution-NonCommercial 4.0 International License, and LLM-MAR licensed under MIT License. By accessing, using, downloading, or interacting with this project, you agree to comply with these licenses.
For full license details, credits, and attributions, see LICENSE.md.
The Open Greek New Testament project: https://github.com/eliranwong/OpenGNT
The Open Hebrew Bible project: https://github.com/eliranwong/OpenHebrewBible
LLM-MAR project: https://github.com/llm-mar/llm-mar