Summary

This project delivers an AI-driven personal knowledge management tool, functioning as an intelligent notebook assistant. It tackles the common issues of scattered knowledge and unreviewed notes by integrating robust note management, a Retrieval-Augmented Generation (RAG) knowledge base, and AI writing assistance. The solution benefits individuals seeking advanced knowledge organization and AI-powered content creation, as well as developers looking to integrate RAG capabilities.

How It Works

Built using FastAPI and LangChain, the system operates as a core RAG engine. It features a Markdown editor enhanced with AI completion, AI-generated note tagging for automatic classification, and semantic search powered by vector embeddings (ChromaDB). A key component is the Ebbinghaus-based spaced repetition system for effective knowledge recall. The RAG functionality supports multi-format document uploads (TXT, PDF, MD, PPTX, DOCX) for accurate question-answering, with results providing document source citations. The architecture includes MySQL for persistent chat history and JWT for user isolation, offering flexibility with Aliyun DashScope or local Ollama LLMs.

Quick Start & Requirements

• Installation: Clone the repository, install backend dependencies using uv sync, and frontend dependencies with npm install or pnpm install.
• Prerequisites: Python 3.12+, uv 0.11.9, Node.js 16+. Requires operational MySQL and Redis databases. LLM integration necessitates either Aliyun DashScope API keys or a local Ollama setup. A Qwen3-Reranker-0.6B model must be downloaded and its path configured. LangSmith API key is optional for tracing.
• Configuration: Comprehensive .env file setup for LLM, database, and service configurations is mandatory.
• Documentation: Interactive OpenAPI documentation is accessible at http://localhost:8000/docs.

Highlighted Details

• AI-powered asynchronous tagging and classification of notes.
• Full-text semantic search for notes and knowledge base content.
• Spaced repetition algorithm for effective knowledge recall.
• Real-time AI-assisted writing features, including inline completion and content generation.
• Cross-source relevance recommendations between notes and uploaded documents.
• RAG-based Q&A with verifiable document source citations.
• Support for diverse document formats: TXT, PDF, MD, PPTX, DOCX.
• Robust user-level data isolation for knowledge bases.

Maintenance & Community

Direct contact is available via email (n3032747608@163.com) and QQ (3032747608). Project issues can be submitted via GitHub.

Licensing & Compatibility

The project is released under the MIT License, permitting broad use, including commercial applications and integration into closed-source projects.

Limitations & Caveats

The project offers two distinct operational modes via branches: a full-featured "NoteBook" (master) and a basic "RAG Service" (base-rag). Setup is complex, requiring the deployment and configuration of multiple services (FastAPI, Django, databases) and specific LLM/embedding models. Manual model downloads and environment variable configuration are essential steps.