Repolode
Submit
HKUDS

HKUDS/DeepTutor

An agent-native, self-hostable AI learning workspace for personalized tutoring, Q&A, quizzes, research and visualization, built on Python and Next.js.

Open-source alternative to:KhanmigoQuizlet

View on GitHubWebsite
Stars
41,795
Forks
5,286
Open issues
152
Last commit
10 hours ago

What is it?

DeepTutor is a self-hosted learning workspace that puts tutoring, problem solving, quiz generation, deep research, visualization and mastery practice into one system. Chat, Ask Questions, Quiz, Research, Solve, Course Study, Mastery Path and Immersive Reading share a common capability runtime and session context, so learners can move between modes without losing their place.

Learning material is handled through versioned knowledge bases built on several retrieval engines, including LlamaIndex, PageIndex, GraphRAG and LightRAG, with pluggable document parsing. Users can also connect Obsidian vaults, Kiwix archives, and external libraries, and add YouTube or Bilibili videos to Reading with transcript search. A Memory layer records traces and summaries of the learner's progress, and the Memory Graph links summaries back to their evidence.

The project ships as a pip package, a source checkout and a Docker image. It includes a Next.js web app, a `deeptutor` CLI, and support for MCP servers, installable skills and multiple LLM providers.

Key features

  • 01Chat, quiz, research, solve and visualization modes share one runtime and session context.
  • 02Versioned knowledge bases built on LlamaIndex, PageIndex, GraphRAG, LightRAG and other engines.
  • 03Video learning from YouTube and Bilibili links with transcript search and saved notes.
  • 04Memory layers with traces and summaries that link back to their evidence.
  • 05Installable via pip, from source, or as a Docker image, with a CLI and web app.
  • 06Extensible through MCP servers, CLI apps and community skills.

Use cases

  • Self-hosting a personal AI tutor that works from your own course notes and documents.
  • Preparing for exams with generated quizzes and mastery-based practice paths.
  • Running deep research on a topic and turning the results into study notes or visualizations.
  • Building a knowledge base from an Obsidian vault and studying it with an AI tutor.

Things to know

  • The README excerpt shows a breaking front/back-end refactor in v1.6.3, so upgrading across that version may require changes.
  • Self-hosting requires a Python 3.11+ backend and a Next.js frontend, so setup is more involved than a single binary.

Topics

#ai-tutor#rag#llm#learning#multi-agent#python#nextjs#self-hosted#ai-agents#clawbot

FAQ

What is DeepTutor?
An agent-native, self-hostable AI learning workspace for personalized tutoring, Q&A, quizzes, research and visualization, built on Python and Next.js. DeepTutor is a self-hosted learning workspace that puts tutoring, problem solving, quiz generation, deep research, visualization and mastery practice into one system. Chat, Ask Questions, Quiz, Research, Solve, Course Study, Mastery Path and Immersive Reading share a common capability runtime and session context, so learners can move between modes without losing their place.
Is DeepTutor free to use? What license does it have?
DeepTutor is open source under the Apache-2.0 license. Check the license terms before using it in commercial products.
Is DeepTutor still maintained?
Actively maintained. The last commit to DeepTutor was on Oct 19, 2026.
What can I use DeepTutor for?
Self-hosting a personal AI tutor that works from your own course notes and documents. Preparing for exams with generated quizzes and mastery-based practice paths. Running deep research on a topic and turning the results into study notes or visualizations. Building a knowledge base from an Obsidian vault and studying it with an AI tutor.
Where can I get DeepTutor?
The source code and installation instructions are on GitHub: https://github.com/HKUDS/DeepTutor

Summary generated by AI from the project's README. GitHub stats refresh daily.

Related projects