🎓 EduBot — AI-Powered Education Assistant

An intelligent teaching assistant built on nanobot that helps educators design lesson plans, manage course materials, and interact with students via a modern web interface.

---

✨ Features

Feature	Description
💬 AI Chat	Conversational AI assistant powered by any LLM (OpenAI, Claude, Gemini, DeepSeek, etc.)
📝 Lesson Plan Generator	Auto-generates structured, pedagogically sound lesson plans in Chinese or English
🎬 Video Prompt Generator	Converts lesson plans into storyboard-style prompts for educational video production
📚 Document Library	Import, index, and semantically search course materials (PDF, DOCX, TXT, Markdown)
🧩 Skills & MCP	Extend agent capabilities via custom skills or Model Context Protocol (MCP) servers
📁 File Manager	Browse and edit workspace files directly from the web UI
⚙️ Config Panel	Manage API keys, LLM providers, and agent settings through the browser

---

Demo

---

🏗️ Architecture

EduBot
├── nanobot/              # Core agent framework (based on nanobot)
│   ├── agent/
│   │   ├── tools/
│   │   │   ├── education_lesson.py    # Lesson plan & video prompt tools
│   │   │   └── education_document.py  # Document import / vector search
│   │   └── loop.py                    # Agent reasoning loop
│   ├── api/
│   │   └── app.py        # FastAPI backend (REST API)
│   ├── channels/         # Chat platform integrations
│   ├── providers/        # LLM provider adapters
│   └── frontend/         # Next.js web console
│       └── app/page.tsx  # Main UI (chat, config, skills, files, docs)
└── workspace/            # Agent workspace (memory, skills, sessions)

---

📦 Installation

Prerequisites

• Python ≥ 3.11
• Node.js ≥ 18 (for the web frontend)

1. Clone the repository

git clone https://github.com/10and01/EduBot.git
cd EduBot

2. Install Python dependencies

pip install -e ./nanobot

Or with uv (faster):

uv pip install -e ./nanobot

3. Install frontend dependencies

cd nanobot/frontend
npm install

---

🚀 Quick Start

Configure your LLM provider

Create or edit ~/.nanobot/config.json:

{
  "providers": {
    "openrouter": {
      "apiKey": "sk-or-v1-YOUR_KEY"
    }
  },
  "agents": {
    "defaults": {
      "model": "anthropic/claude-opus-4-5",
      "provider": "openrouter"
    }
  }
}

Supported providers: OpenAI, Anthropic, Azure OpenAI, Gemini, DeepSeek, Qwen, Moonshot, VolcEngine, vLLM (local), OpenRouter, and more.

Start the backend

python -m uvicorn nanobot.api.app:app --host 127.0.0.1 --port 8000

Start the frontend

cd nanobot/frontend
npm run dev

Open http://localhost:3000 to access the EduBot web console.

---

🎓 Education Features

Lesson Plan Generation

EduBot can produce fully structured lesson plans conforming to standard pedagogical frameworks. A lesson plan includes:

• 学习目标 / Learning Objectives — scaffolded knowledge, skill, and attitude goals
• 学情分析 / Learner Analysis — prerequisite knowledge assessment
• 教学重点与难点 / Key Points & Challenges — core concepts and common misconceptions
• 教学流程 / Teaching Process — timed phases (warm-up → instruction → practice → assessment)
• 评价与作业 / Assessment & Homework — formative and summative strategies, tiered homework
• 板书与资源 / Board Plan & Resources — visual layout and material checklist

Example prompt:

帮我设计一份七年级数学"一次函数"的45分钟教案

Document Library (RAG)

Import your teaching materials and ask questions grounded in your own content:

1. Upload a PDF, DOCX, TXT, or Markdown file via the 文档库 panel
2. Specify subject and grade for metadata filtering
3. Chat with the AI — it will retrieve and cite relevant passages from your documents

Example:

根据上传的课文，为五年级语文设计三道阅读理解题

Video Prompt Generation

Convert a lesson plan into a storyboard prompt ready for AI video tools (e.g., Sora, Kling):

把刚才的教案转成适合拍摄教学视频的分镜脚本

---

💬 Chat Platform Integrations

EduBot inherits all chat channels from nanobot:

Platform	Notes
Web Console	Built-in — no extra setup needed
Telegram	Bot token from @BotFather
DingTalk	App Key + App Secret
Feishu / Lark	App ID + App Secret
Slack	Bot token + App-Level token
Discord	Bot token + Message Content intent
WeChat Work	Bot ID + Bot Secret
Email	IMAP/SMTP credentials
QQ	App ID + App Secret

Configure any channel in ~/.nanobot/config.json, then run:

nanobot gateway

---

🐳 Docker

docker compose up -d

The docker-compose.yml in the nanobot/ directory starts the agent and API server together.

---

🧩 Extending with Skills & MCP

Place custom skill files in workspace/skills/ or configure MCP servers in the Skills / MCP panel of the web console.

A skill is a Markdown file that instructs the agent on specialized behavior. See workspace/TOOLS.md for built-in tool constraints and the workspace/skills/weather/ example for a simple reference implementation.

---

🗂️ Project Structure

workspace/
├── SOUL.md        # Agent personality and values
├── AGENTS.md      # Agent behavioral instructions
├── TOOLS.md       # Tool usage notes
├── USER.md        # User profile (personalize interactions)
├── HEARTBEAT.md   # Periodic background tasks
├── memory/        # Persistent agent memory
├── sessions/      # Conversation history
└── skills/        # Custom agent skills

---

🌟 Quick Tips

For Educators

1. Upload course materials → Ask questions → AI cites relevant content
2. Generate lesson plans → Convert to video prompts → Use with video AI tools
3. Extend with custom skills → Add domain-specific tools for your subject area

For Developers

• Backend: FastAPI + Pydantic (async-first Python)
• Frontend: Next.js 14 + Ant Design (React Server Components)
• Vector Database: Chroma for semantic search (RAG)
• Extensibility: MCP servers, custom skills, multiple LLM providers

---

🤝 Contributing

1. Fork the repository
2. Create a feature branch: git checkout -b feature/my-feature
3. Commit your changes: git commit -m "feat: add my feature"
4. Push and open a pull request

---

📄 License

This project is licensed under the MIT License.

---

🙏 Acknowledgements

EduBot is built on top of nanobot by HKUDS, an ultra-lightweight personal AI assistant framework.