An OpenClaw skill for AI-powered tutoring of 6-12 year olds, combining Socratic questioning with adaptive difficulty.
一个面向6-12岁儿童的AI家教技能,部署在 OpenClaw 平台上。结合苏格拉底式引导和直接讲解两种教学法,以数学和科学为主。
- 🧠 AI主动出题 — 不是被动问答,而是AI根据孩子水平主动生成适合的题目
- 🎯 自适应难度 — 连续答对自动升级,遇到困难自动降级
- 💡 苏格拉底式引导 — 不直接给答案,用提问引导孩子自己思考
- 🎨 兴趣场景化 — 喜欢恐龙?数学题就用恐龙当场景
- 📊 学习报告 — 自动生成家长报告:正确率、薄弱点、进步趋势
- 📚 人教版对齐 — 1-6年级知识点完整覆盖,与教材同步
kid-tutor/
├── SKILL.md # 核心技能定义(OpenClaw读取)
├── references/
│ ├── curriculum.md # 1-6年级知识图谱(数学+科学)
│ ├── pedagogy.md # 苏格拉底教学法 + 适龄语言策略
│ └── question-templates.md # 题目模板库
├── scripts/
│ ├── manage_profile.py # 学生档案管理
│ └── generate_report.py # 家长报告生成
└── README.md
clawhub install kid-tutorcd ~/.openclaw/workspace/skills
git clone https://github.com/lixiang1076/kid-tutor.git安装后,在 OpenClaw 对话中即可触发。示例:
你: 帮我家孩子出几道数学题,他8岁,三年级
AI: 好呀!我先了解一下你家宝贝 🌟 他叫什么名字?平时喜欢什么?
# 初始化学生档案
python3 scripts/manage_profile.py data/kid-tutor/xiaoming init \
--name 小明 --age 8 --grade 3 --interests "恐龙,太空"
# 查看档案
python3 scripts/manage_profile.py data/kid-tutor/xiaoming show
# 生成最近7天学习报告
python3 scripts/generate_report.py data/kid-tutor/xiaoming --days 7| 场景 | 策略 |
|---|---|
| 孩子答对 | 热情鼓励 + "你是怎么想的?" |
| 孩子答错 | 不说"错了",追问引导:"有意思!如果按你说的算呢?" |
| 引导2-3轮仍困难 | 切换讲解模式,用生活类比讲清楚 |
| 连续3题正确 | 自动提升难度 |
| 孩子跑题 | 温和拉回:"哈哈,回头聊这个!先把这题搞定?" |
| 孩子沮丧 | 安慰 + 建议休息:"这题确实有点难,要不先休息一下?" |
- ✅ 只讨论学习相关内容
- ✅ 永远正面鼓励,不评判能力
- ✅ 保护儿童隐私
- ❌ 绝不说"你太笨了"或类似否定语
- ❌ 不收集敏感个人信息
An OpenClaw skill that turns your AI assistant into a personalized tutor for elementary school children (ages 6-12), focusing on math and science.
- Socratic Teaching — Guides children to discover answers through questioning, never gives answers directly
- Adaptive Difficulty — Auto-adjusts based on performance (3 correct → level up)
- Interest-based Contextualization — Loves dinosaurs? Math problems feature dinosaurs
- Parent Reports — Generates markdown reports with accuracy, weak spots, and progress trends
- PEP Curriculum Aligned — Covers grades 1-6 Chinese national curriculum (人教版)
- OpenClaw installed and configured
- Python 3.8+
- Any LLM backend (works with OpenAI, Claude, Gemini, local models, etc.)
MIT
Issues and PRs welcome! Ideas for improvement:
- Voice interaction (for younger children who can't type)
- Image generation for geometry/science diagrams
- English language support
- More subjects (Chinese language, English)
- Spaced repetition system
- Integration with school curriculum schedules