开源 DeepTutor:把笔记、测验、解题与学习路径收进同一套可自托管辅导工作区
HKUDS/DeepTutor(Apache-2.0)是面向个性化辅导的开源学习工作区:上传资料建知识库、生成测验、分步解题、把视频变成可检索学习会话,并支持多模型与自托管;站点 deeptutor.info。
开源项目 DeepTutor(GitHub:HKUDS/DeepTutor,协议 Apache-2.0,站点 deeptutor.info)把自己定位为「终身个性化辅导」工作区:不是单一聊天框,而是把辅导、测验、解题、研究、可视化与掌握路径放在同一运行时与会话上下文里。
据本窗介绍帖与官方文档,使用者可以上传文档构建知识库、生成测验与练习题、用分步推理解题、把 YouTube 等视频变成可检索学习会话、做带引用的主题研究,并生成个性化学习路径与进度追踪。系统支持多种模型供应商与知识库引擎,可自托管;安装路径包括 PyPI(pip install -U deeptutor 后 deeptutor init && deeptutor start)、Docker 与源码。文档还提到可与本地 Claude Code / Codex 等智能体对接(「My Agents」一类能力)。
对教育培训智能化,DeepTutor 的重点是「学习材料—练习—记忆」闭环留在同一工作区,减少在多个通用聊天工具间搬运笔记。仓库星标与版本以 GitHub 为准(近期有 v1.6.x 发布说明)。采用时需自备模型密钥,并按文档配置解析与检索引擎。
和只把教材贴进通用聊天窗口不同,DeepTutor 强调同一工作区里复用知识库、题库与记忆,让测验、解题与阅读进度互相可见。自托管意味着数据与模型密钥留在自己的环境,但也要求维护解析引擎、向量检索与版本升级。对学校或培训团队,可先用公开文档跑通「上传讲义→生成测验→分步解题」闭环,再评估是否对接本地编程智能体作为「My Agents」。
DeepTutor open-source AI learning workspace
FREE AI tutor that can turn your study materials into a complete learning system 😳
DeepTutor is an open-source AI learning workspace that goes beyond asking ChatGPT questions.
You can use it to: → Upload documents and build your own knowledge base → Generate quizzes and practice questions → Solve problems with step-by-step reasoning → Turn YouTube videos into searchable learning sessions → Research topics and create cited reports → Build personalized learning paths and track your progress
The interesting part?
Your books, notes, conversations, and learning progress can work together instead of being scattered across different AI tools.
It also supports multiple AI model providers and knowledge-base engines, so you can customize the setup around your workflow.
Open source Apache 2.0 license Self-hostable Built for personalized learning
AI tutors are becoming more than chatbots.
Repo: https://t.co/dasKZnlwcI
The real opportunity is building a system that helps you actually understand and remember what you learn.