中小学各学科教学设计助手 — 基于 ICAP / 布鲁姆分类学 / 扶放有度框架，通过多智能体协作生成系统化教案。
An AI-powered instructional design assistant for K-12 education. Multi-agent collaboration for systematic lesson plan generation, powered by ICAP, Bloom's Taxonomy, and Gradual Release of Responsibility frameworks.
| Capability | Description |
|---|---|
| Lite Mode 🚀 | Quick lesson plan generation with 3 agents, 5 steps — ideal for daily teaching |
| Pro Mode 🏗️ | Deep instructional design with 8 agents, 9 steps — for demo/open class competitions |
| 5 Lesson Types 📚 | New lesson / Review / Practice / Inquiry / Assessment review |
| Multi-Agent Architecture 🤖 | Student analyst, Content architect, Goal designer, Activity designer, Assessment designer, Motivation designer, Metacognition coach, QA reviewer |
| ICAP Framework 🧠 | Interactive-Constructive-Active-Passive cognitive engagement optimization |
| Bloom's Taxonomy 📊 | 6×4 matrix for precise learning objective mapping |
| 7-D Quality Check 🔍 | Goals ↔ activities ↔ assessment alignment + readability audit (v3.0: plain-language output guaranteed) |
| Word Export 📄 | Generates downloadable .docx lesson plans |
- Bilingual Expression Mechanism — The skill thinks in professional theory internally, but outputs in plain, teacher-friendly language. Jargon is moved to an appendix with first-use explanations and "because…so…" rationale phrasing. New:
references/plain-language.md. - Knowledge Base Now Optional — KB integration is an enhancement layer, not a hard dependency. The skill works out of the box for anyone you share it with (all theory is distilled into
references/). - Edge Cases Handbook — New:
references/edge-cases.mdcovering special class types, missing information, and fallback strategies. - Leaner Core —
SKILL.mdstreamlined from 1059 → 443 lines; theory distilled into references, duplication removed.
# Clone the repo
git clone https://github.com/tanghua-git/skill-instructional-design.git \
~/.openclaw/skills/instructional-design
# Or install directly via OpenClaw:
openclaw skill install tanghua-git/skill-instructional-designCopy the SKILL.md and references/ folder to your AI agent's skills directory:
cp -r skill-instructional-design ~/my-agent-skills/Input: "设计八年级物理《压强》第一课时教案" Output: Full lesson plan with objectives, activities, assessment, and board design
Lite Mode Workflow:
Step 0: Requirements Analysis
Step 1: Goal Design (with embedded student/content/motivation constraints)
Step 2: Activity Design (with embedded metacognitive prompts)
Step 3: Assessment Design
Step 4: Lesson Plan Assembly → Word Export
Input: "高一语文《念奴娇·赤壁怀古》新授课，公开课" Output: Comprehensive instructional design with rationale, differentiated paths, and quality audit
Pro Mode Workflow (9 steps):
Step 0 → Requirements Analysis
Step 0.5 → Unit Positioning + Lesson Type Decision
Step 1 → Student + Content + Motivation Analysis (parallel)
Step 1.5 → Goal Design (Bloom's 6×4 matrix)
Step 2 → Activity + Assessment + Metacognition (parallel)
Step 2.5 → 5-D Consistency Review + Confidence Self-Assessment
Step 3 → Lesson Plan Assembly + Differentiated Paths
Step 4 → QA Feedback
Step 5 → Implementation Issue Prediction
| Framework | Application | Source |
|---|---|---|
| ICAP (Chi & Wylie) | Cognitive engagement optimization across lesson types | references/icap-scaffold.md |
| Bloom's Taxonomy (revised) | 6 cognitive processes × 4 knowledge types | references/bloom-matrix.md |
| Gradual Release of Responsibility | "I do → We do → You do" scaffolding | Embedded in activity design |
| Self-Determination Theory (Deci & Ryan) | Autonomy/Competence/Relatedness motivation design | references/sdt-motivation.md |
| Backward Design (UbD, Wiggins & McTighe) | Goals → Assessment → Activities | Embedded in workflow |
| AI in Education Framework | AI integration levels (No AI / AI-assisted / AI-integrated) | references/ai-era-framework.md |
skill-instructional-design/
├── SKILL.md # Main skill file (v3.0 — bilingual plain-language refactor)
├── README.md # This file
└── references/
├── bloom-matrix.md # Bloom's 6×4 matrix + ABCD objectives
├── icap-scaffold.md # ICAP framework + 50+ teaching strategies
├── assessment-design.md # GRASPS + SOLO + tiered rubrics
├── learner-analysis.md # Misconception prediction + 4-layer diagnosis
├── big-ideas.md # Big ideas extraction methodology
├── sdt-motivation.md # Self-determination theory design patterns
├── metacognition.md # Reflection card templates + Think Aloud
├── create-checklist.md # CREATE 6-dimension QA checklist
├── plain-language.md # v3.0: plain-language expression rules (jargon → appendix)
├── edge-cases.md # v3.0: edge cases & fallback strategies
└── ai-era-framework.md # AI in education: assessment taxonomy + L0-L5 teacher spectrum
| Dimension | Score | Note |
|---|---|---|
| Frontmatter | ✅ 10/10 | Clear name, description, and trigger keywords |
| Workflow Clarity | ✅ 10/10 | Numbered steps with explicit inputs/outputs |
| Failure Mode Encoding | ✅ 10/10 | If-fail-then branches throughout |
| Checkpoint Design | ⏳ 6/10 | Logic exists, visual markers pending |
| Actionable Specificity | ✅ 10/10 | Concrete parameters, templates, and examples |
| Resource Integration | ✅ 10/10 | All reference paths correct and accessible |
| Architecture | ✅ 10/10 | Clean hierarchy, no redundancy |
| Empirical Testing | ✅ 10/10 | Test prompts validated |
| Anti-pattern Blacklist | ✅ 10/10 | Explicit "don't do" list |
Overall: 92.4/100 — v2.1 Darwin baseline. v3.0 keeps this architecture and adds the bilingual plain-language layer on top (see "What's New in v3.0").
- OpenClaw — AI agent platform
- Darwin Skill — Skill optimization framework
- Academic Research Skills — Academic workflow skills for OpenClaw
MIT © 2026 Tanghua (唐华)
Found an issue or have a suggestion? Feel free to open an issue or submit a PR.
If this skill is useful to you, consider giving it a ⭐ — it helps others discover it!
Built with ❤️ for K-12 educators. 教学法第一，工具第二。