Create account

DEV Community

Gabriele Trovato

Posted on May 7

Building LoreKeeper: An Immersive 3D Library to Bridge EPUBs and AI

#ai #showdev #opensource #javascript

Introduction

Let's be honest: digital book management has felt "flat" for a long time. Most ebook managers look like Excel spreadsheets from the late 90s. As a developer who loves both 3D graphics and reading, I wanted something more.

I built LoreKeeper, an open-source, 3D digital library that transforms your EPUB collection into an interactive sanctuary.

But it's not just about the looks—it's about making your books "AI-ready."

The Vision: Beyond the List

The goal was to create a "Sanctuary" for digital lore. Using Three.js, I developed a 3D environment where books aren't just entries in a database; they are objects on a shelf. You can fly through the library, move books around, and interact with your collection in a way that feels organic.

The Tech Stack

Frontend: Three.js for the 3D engine, Vite for a lightning-fast build.
Backend: Node.js & Express for file handling and metadata extraction.
Containerization: Docker & Docker-Compose for seamless self-hosting.
APIs: Integration with Apple Books for automated metadata (covers, plots, authors).

The AI Challenge: From Local RAG to "AI-Bridge"

Initially, I experimented with local RAG (Retrieval-Augmented Generation) using local LLMs. However, I realized that for many users, running a heavy LLM alongside a 3D engine was a hardware nightmare.

I pivoted to an "AI-Bridge" philosophy:
LoreKeeper includes a specialized engine that extracts, de-clutters, and formats EPUB content into clean, structured Knowledge Bases. You can instantly export any book into a .txt file optimized for ChatGPT, Claude, or Gemini.

Local privacy + Global intelligence.

Key Features

Dynamic 3D Shelves: Smooth navigation and shelf management.
Automated Metadata: Just drop your EPUB; LoreKeeper finds the cover and the plot.
Docker-Ready: One command (docker-compose up -d) to get your library running locally.
Export for AI: Clean text extraction for deep analysis.

What's Next?

LoreKeeper is now at v1.0.0, but the roadmap is exciting:

RSVP Speed Reading: A "Rapid Serial Visual Presentation" mode to read at 500+ WPM.
Internationalization: Adding full English support.
E-Reader Sync: Bridging the gap between the 3D library and physical devices like Kindle/Kobo.

Conclusion

LoreKeeper is 100% open source. I’d love for the community to try it out, break it, and help me build the future of digital reading.

Check out the repository here:

GabrieleTrovato01 / LoreKeeper

EN: Immersive 3D EPUB manager with automated metadata extraction and clean text exporting for AI Knowledge Bases. 100% local and Docker-ready. IT: Gestore EPUB 3D immersivo con estrazione metadati e funzione di esportazione testi per Knowledge Base IA. 100% locale e pronto per Docker.

📚 LoreKeeper

Read this in: English | Italiano

🇬🇧 English

An interactive 3D digital library that allows you to upload, browse, and organize your EPUB files in an immersive graphical environment.

But that's not all: thanks to the Export for AI feature, you can extract the entire clean text of any book into a ready-to-use Knowledge Base (.txt file). You can easily upload this file to ChatGPT, Claude, or Gemini to literally "talk" to your books: ask questions about the plot, search for specific concepts, or ask for summaries, leaving the heavy computational lifting to cloud AIs without overloading your PC!