Verified project record

HKUDS/DeepTutor

DeepTutor is an agent-native learning and research workspace that unifies tutoring, problem solving, quiz generation, and document analysis in a single local runtime. It is intended for users who want to connect retrieval-augmented generation libraries with multiple knowledge formats without switching between isolated tools.

Project overview

Project type: AI Agent
Use cases: Search & Research · Learning & Education
Deployment: Refer to project documentation
License: Apache-2.0

Best for

General users and educators seeking a local, unified workspace that connects tutoring, quizzes, and research across their own documents.
Users who rely on existing Obsidian vaults or specific RAG frameworks like LlamaIndex, GraphRAG, or LightRAG.

Key capabilities

Provides one runtime where Chat, Quiz, Research, Visualize, Solve, and Mastery Path run on the same agent loop.
Provides versioned RAG libraries across LlamaIndex, PageIndex, GraphRAG, LightRAG, or a linked Obsidian vault.
Provides a 'deeptutor' binary with an interactive REPL and structured JSON output for other agents.
Provides L1 traces, L2 surface summaries, and L3 synthesis with a Memory Graph to trace claims back to evidence.
Built-in docx / pdf / pptx / xlsx skills that execute model-generated code to produce files.

Limitations and risks

Hardware graphics processing unit requirements are not documented in the available materials.
Requires an external or local LLM provider configuration, meaning users must supply or connect their own language model.

Getting started

Create a project directory, install the package using pip, initialize the configuration, and start the tool using the provided command-line interface.

Evidence and sources

GitHub project description: DeepTutor: Lifelong Personalized Tutoring.
README: DeepTutor is an agent-native learning workspace that connects tutoring, problem solving, quiz generation, research, visualization, and mastery practice in one extensible system.
README: 2026-04-19 🎉 20k stars in 111 days! Thank you for the support toward truly personalized, intelligent tutoring.
README: GA cut of v1.4: Auto Mode, three-layer Memory, agentic Deep Research / Solve / Question, LlamaIndex RAG refactor, Visualize/Animator merge, and restart-safe turn runtime.
README: One `deeptutor` binary, two ways in: an interactive **REPL** for people who live in the terminal, and structured **JSON** for other agents that drive DeepTutor as a tool.

Reviewed at: 2026-07-16T10:18:28Z