About this project
DeepTutor is an open-source project from HKUDS that describes itself as a lifelong personalized tutoring system. It is a full-stack application built with Python 3.11+ on the backend and Next.js 16 on the frontend, released under the Apache 2.0 license, with a companion documentation site at deeptutor.info and an arXiv paper reference.
The README presents a long, fast-moving release history (roughly v0.2.0 through v1.6.8) that documents the project's evolving feature set. Recurring themes include:
- Agentic chat and tutoring: a single agent loop powers chat, guided learning, deep research, problem solving and question generation, with tool/capability plugins and configurable per-model capabilities.
- Knowledge management: knowledge bases with multiple retrieval engines (LightRAG, GraphRAG, PageIndex, FAISS, LlamaIndex-based ingestion), document parsing engines (MinerU, Docling, PyMuPDF4LLM, LiteParse, Apache Tika), versioned indexes and re-indexing workflows.
- Learning features: Guided Learning with mastery gates, a Question Bank, quizzes, Book Engine for compiling "living books" with progress tracking and Markdown export, Co-Writer for multi-document writing, and interactive visualizations.
- Memory: a three-layer memory workbench (L1/L2/L3) and persistent per-user memory.
- Integrations: many LLM and embedding providers, web search providers, MCP services, and IM channels (the README mentions 15 channels) for "Partners" — configurable tutor personas with their own libraries, skills and private memory.
- Deployment: Docker and Compose support, rootless Podman notes, multi-user deployments with isolated workspaces, admin roles, and a CLI (`deeptutor`) with commands such as start, doctor and skill install.
The repository is clearly aimed at learners and educators who want a self-hostable, extensible AI tutoring environment rather than a single-purpose chatbot. The README is written in English with translations linked for Chinese, Japanese, Spanish, French, Arabic, Russian, Hindi, Portuguese, Thai and Polish.
Comments
0 Rating appears after 10 ratings
Sign in to join the discussion.