Purpose
DeepTutor is an agent-native lifelong personalized learning workbench developed by HKUDS, with the project page at https://github.com/HKUDS/DeepTutor , and the official website at https://deeptutor.info/ . It unifies AI tutoring, problem solving, quiz generation, research, visualization, and mastery practice in a single learning workspace, with the goal of providing learners with a sustainable learning environment that includes personalized memory.
Main Capabilities
According to the repository materials, DeepTutor includes the following capability modules:
- Multi-engine RAG (Retrieval-Augmented Generation) and knowledge base management, involving retrieval and indexing solutions such as GraphRAG, PageIndex, LightRAG, and FAISS, and supporting linked knowledge bases, Obsidian, and GitHub repositories as knowledge sources.
- A persistent memory system, including a three-layer memory (L1/L2/L3) workbench and configurable Memory.
- The Partners / My Agents agent system, supporting connection to local tools such as Claude Code / Codex and real-time consultation during conversations; each Partner has an independent persona, knowledge base, skills, and private memory.
- Real-time coding command-line integration, providing CLI entry points (such as commands like deeptutor start, deeptutor doctor, deeptutor skill install, etc.).
- Learning-related features: Guided Learning, Mastery Path, Question Bank and @-mention, Question Notebook, and learning logs.
- Content generation and reading: Book Engine book compiler, multi-document Co-Writer, immersive Reading and EPUB reading annotation, visualization (Chart.js/SVG/Mermaid), and Math Animator.
- Research-related features: Deep Research, question answering and problem solving, and quiz generation.
- Other engineering capabilities: multi-user isolated deployment, in-browser preview of Word documents DOCX/XLSX, MCP services, third-party tool/capability plugins, multilingual interfaces (French, Ukrainian, German, Polish, etc.), and multi-provider support for web search.
How to Use
The information provided in the materials includes: it can be accessed through the official website, the code repository is on GitHub, it can be self-deployed (involving Docker and Compose deployment-related content), and it provides CLI commands, such as deeptutor start, deeptutor doctor, deeptutor skill install. The repository includes collaboration entry points such as the Contributing Guide and Roadmap. Regarding the complete installation steps and configuration details, the materials do not provide the full command text, and it is not yet provided.
Applicable Scenarios
Suitable for learners who need long-term, personalized AI tutoring, and for users who want to integrate knowledge base retrieval, research, quizzes, reading, and visualization into the same workbench; it is also suitable for deploying learning platforms in multi-user environments.
Limitations and Missing Information
The materials do not provide specific pricing and fee information, platform coverage, or performance results data; they do not provide complete installation and deployment steps, system requirements, or a dependency list; the specific usage steps for some features are only briefly described in the materials, and the missing parts are explicitly stated as not yet provided.
