Mindory harnesses large language models to make reading interactive and delightful.
Ask questions to your books, explore characters, or even converse with the story—turning reading into a vivid, AI-powered journey.
- Chinese-friendly: Full support for Chinese text processing from segmentation to embedding
- Order-based authoritative RAG: Accurate recall, avoiding bias from implicit rearrangement
- Observability-first: Built-in RAG Trace for debugging, evaluation, and productization
- Engineering-oriented: Clear crate boundaries, error layering, and state management
| Scene | Mindory Does |
|---|---|
| Question | Ask questions to the book in natural language |
| Recall | Retrieve the most relevant text snippets |
| Understand | Completes context automatically to avoid misinterpretation |
| Generate | Generates answers based on real book content |
| Explain | Shows cited content in the answer |
- Multi-document support (TXT, PDF, EPUB)
- Multi Vector Store / Multi Embedding model
- RAG Trace visualization (Timeline / Rank / Context Diff)
This is the public-facing repository:
- Browse technical docs and examples
- Submit issues or feature requests
- Follow the roadmap for upcoming enhancements