{"id":"0ccac157-3c1c-4337-a3c8-4898bb44eea5","entityType":"agent","slug":"clawhub-jacky-wzj-master-teacher","name":"Master Teacher","canonicalUrl":"https://www.xpersona.co/agent/clawhub-jacky-wzj-master-teacher","canonicalPath":"/agent/clawhub-jacky-wzj-master-teacher","generatedAt":"2026-10-12T17:51:26.655Z","source":"CLAWHUB","claimStatus":"UNCLAIMED","verificationTier":"NONE","summary":{"evidence":{"source":"editorial-content","verified":true,"confidence":"high","updatedAt":"2026-10-11T21:55:35.084Z","emptyReason":null},"description":"Systematic teaching skill for AI agents. Transforms the agent into a master-level instructor using mastery learning, Socratic questioning, and structured les... Skill: Master Teacher Owner: jacky-wzj Summary: Systematic teaching skill for AI agents. Transforms the agent into a master-level instructor using mastery learning, Socratic questioning, and structured les... Tags: latest:2.8.0 Version history: v2.8.0 | 2026-04-21T03:19:20.362Z | user v2.8: Add teaching principles #11 (state facts before comparing) and #12 (state facts before questioning) v2.7.0 | 2026-04-20T22:44:45","descriptionLabel":"Technical summary","evidenceSummary":"Capability contract not published. No trust telemetry is available yet. 988 downloads reported by the source. Last updated 10/11/2026.","installCommand":"clawhub skill install s1736s06248q8gm2sshtab5kgx853tmt:master-teacher","sourceUrl":"https://clawhub.ai/jacky-wzj/master-teacher","homepage":"https://clawhub.ai/jacky-wzj/skills/master-teacher","primaryLinks":[{"label":"View on ClawHub","url":"https://clawhub.ai/jacky-wzj/master-teacher","kind":"source"},{"label":"Homepage","url":"https://clawhub.ai/jacky-wzj/skills/master-teacher","kind":"homepage"}],"safetyScore":84,"overallRank":62,"popularityScore":60,"trustScore":null,"claimedByName":null,"isOwner":false,"seoDescription":"Systematic teaching skill for AI agents. Transforms the agent into a master-level instructor using mastery learning, Socratic questioning, and structured les..."},"coverage":{"evidence":{"source":"public-profile","verified":false,"confidence":"medium","updatedAt":"2026-10-11T21:55:35.084Z","emptyReason":null},"protocols":[{"protocol":"OPENCLEW","label":"OpenClaw","status":"self-declared","notes":"Declared in the public agent profile."}],"capabilities":[],"verifiedCount":0,"selfDeclaredCount":1,"capabilityMatrix":{"rows":[{"key":"OPENCLEW","type":"protocol","support":"unknown","confidenceSource":"profile","notes":"Listed on profile"}],"flattenedTokens":"protocol:OPENCLEW|unknown|profile"}},"adoption":{"evidence":{"source":"CLAWHUB","verified":false,"confidence":"medium","updatedAt":"2026-10-11T21:55:35.084Z","emptyReason":null},"stars":null,"forks":null,"downloads":988,"likes":null,"task":null,"library":null,"packageName":null,"latestVersion":"2.8.0","tractionLabel":"988 downloads"},"release":{"evidence":{"source":"CLAWHUB","verified":false,"confidence":"medium","updatedAt":"2026-10-11T21:55:35.071Z","emptyReason":null},"lastUpdatedAt":"2026-10-11T21:55:35.084Z","lastCrawledAt":"2026-10-11T21:55:35.071Z","lastIndexedAt":null,"nextCrawlAt":"2026-10-12T21:55:35.071Z","lastVerifiedAt":null,"highlights":[{"version":"2.8.0","createdAt":"2026-04-21T03:19:20.362Z","changelog":"v2.8: Add teaching principles #11 (state facts before comparing) and #12 (state facts before questioning)","fileCount":11,"zipByteSize":17139},{"version":"2.7.0","createdAt":"2026-04-20T22:44:45.109Z","changelog":"v2.7: Phase 2.5 transition guidance, quality checklist conditional checks, teaching principle #10 (never generate content during class), fix step script table","fileCount":10,"zipByteSize":15745},{"version":"2.6.0","createdAt":"2026-04-18T18:10:32.363Z","changelog":"v2.6: Phase 2.5 content authoring overhaul, flexible lesson structure (按需拆分), pre-generated multimedia, remove hardcoded constraints, overview.md blueprint, save-as-you-go, conditional step scripts, student interaction options, Phase 3 pre-check","fileCount":10,"zipByteSize":15408},{"version":"2.1.0","createdAt":"2026-04-18T15:46:36.669Z","changelog":"Bind scripts to phases: every Phase now says which script to call and when. Removed redundant markdown templates. Scripts are single source of truth for state.","fileCount":10,"zipByteSize":12786},{"version":"2.0.0","createdAt":"2026-04-18T15:34:45.803Z","changelog":"Major: extract deterministic operations into Python scripts (init_course, track_progress, show_progress, lesson_report, resume). Model handles teaching and judgment; scripts handle state management. Works reliably regardless of base model capability.","fileCount":10,"zipByteSize":14509},{"version":"1.6.0","createdAt":"2026-04-18T15:27:46.695Z","changelog":"Lesson report after each completion (score, strengths, weak points, key takeaway). Visual progress overview map — image when available, ASCII fallback. Seeing progress motivates.","fileCount":5,"zipByteSize":7370},{"version":"1.5.0","createdAt":"2026-04-18T15:22:19.933Z","changelog":"Intra-lesson progress tracking (exact breakpoint within a lesson), Phase 5 Resume with recall questions, spaced review at lessons 3/6/midpoint","fileCount":5,"zipByteSize":6848},{"version":"1.4.0","createdAt":"2026-04-18T15:13:33.336Z","changelog":"Unify language to English throughout SKILL.md — no more mixed Chinese/English","fileCount":5,"zipByteSize":6167}]},"execution":{"evidence":{"source":"CLAWHUB","verified":false,"confidence":"low","updatedAt":null,"emptyReason":"No published capability contract is available yet."},"installCommand":"clawhub skill install s1736s06248q8gm2sshtab5kgx853tmt:master-teacher","setupComplexity":"low","setupSteps":["Setup complexity is classified as HIGH. You must provision dedicated cloud infrastructure or an isolated VM. Do not run this directly on your local workstation.","Final validation: Expose the agent to a mock request payload inside a sandbox and trace the network egress before allowing access to real customer data."],"contract":{"contractStatus":"missing","authModes":[],"requires":[],"forbidden":[],"supportsMcp":false,"supportsA2a":false,"supportsStreaming":false,"inputSchemaRef":null,"outputSchemaRef":null,"dataRegion":null,"contractUpdatedAt":null,"sourceUpdatedAt":null,"freshnessSeconds":null},"invocationGuide":{"preferredApi":{"snapshotUrl":"https://www.xpersona.co/api/v1/agents/clawhub-jacky-wzj-master-teacher/snapshot","contractUrl":"https://www.xpersona.co/api/v1/agents/clawhub-jacky-wzj-master-teacher/contract","trustUrl":"https://www.xpersona.co/api/v1/agents/clawhub-jacky-wzj-master-teacher/trust"},"curlExamples":["curl -s \"https://www.xpersona.co/api/v1/agents/clawhub-jacky-wzj-master-teacher/snapshot\"","curl -s \"https://www.xpersona.co/api/v1/agents/clawhub-jacky-wzj-master-teacher/contract\"","curl -s \"https://www.xpersona.co/api/v1/agents/clawhub-jacky-wzj-master-teacher/trust\""],"jsonRequestTemplate":{"query":"summarize this repo","constraints":{"maxLatencyMs":2000,"protocolPreference":["OPENCLEW"]}},"jsonResponseTemplate":{"ok":true,"result":{"summary":"...","confidence":0.9},"meta":{"source":"CLAWHUB","generatedAt":"2026-10-12T17:51:26.651Z"}},"retryPolicy":{"maxAttempts":3,"backoffMs":[500,1500,3500],"retryableConditions":["HTTP_429","HTTP_503","NETWORK_TIMEOUT"]}},"endpoints":{"dossierUrl":"https://www.xpersona.co/api/v1/agents/clawhub-jacky-wzj-master-teacher/dossier","snapshotUrl":"https://www.xpersona.co/api/v1/agents/clawhub-jacky-wzj-master-teacher/snapshot","contractUrl":"https://www.xpersona.co/api/v1/agents/clawhub-jacky-wzj-master-teacher/contract","trustUrl":"https://www.xpersona.co/api/v1/agents/clawhub-jacky-wzj-master-teacher/trust"}},"reliability":{"evidence":{"source":"runtime-metrics","verified":false,"confidence":"low","updatedAt":null,"emptyReason":"No trust, reliability, or runtime telemetry is available."},"trust":{"status":"unavailable","handshakeStatus":"UNKNOWN","verificationFreshnessHours":null,"reputationScore":null,"p95LatencyMs":null,"successRate30d":null,"fallbackRate":null,"attempts30d":null,"trustUpdatedAt":null,"trustConfidence":"unknown","sourceUpdatedAt":null,"freshnessSeconds":null},"decisionGuardrails":{"doNotUseIf":["Contract metadata is missing or unavailable for deterministic execution."],"safeUseWhen":[],"riskFlags":["missing_or_unavailable_contract","trust_data_unavailable","schema_references_missing"],"operationalConfidence":"low"},"executionMetrics":{"observedLatencyMsP50":null,"observedLatencyMsP95":null,"estimatedCostUsd":null,"uptime30d":null,"rateLimitRpm":null,"rateLimitBurst":null,"lastVerifiedAt":null,"verificationSource":null},"runtimeMetrics":{"successRate":null,"avgLatencyMs":null,"avgCostUsd":null,"hallucinationRate":null,"retryRate":null,"disputeRate":null,"p50Latency":null,"p95Latency":null,"lastUpdated":null}},"benchmarks":{"evidence":{"source":"no-benchmark-data","verified":false,"confidence":"low","updatedAt":null,"emptyReason":"No benchmark suites or observed failure patterns are available."},"suites":[],"failurePatterns":[]},"artifacts":{"evidence":{"source":"CLAWHUB","verified":false,"confidence":"high","updatedAt":"2026-10-11T21:55:35.084Z","emptyReason":null},"readme":"Skill: Master Teacher\n\nOwner: jacky-wzj\n\nSummary: Systematic teaching skill for AI agents. Transforms the agent into a master-level instructor using mastery learning, Socratic questioning, and structured les...\n\nTags: latest:2.8.0\n\nVersion history:\n\nv2.8.0 | 2026-04-21T03:19:20.362Z | user\n\nv2.8: Add teaching principles #11 (state facts before comparing) and #12 (state facts before questioning)\n\nv2.7.0 | 2026-04-20T22:44:45.109Z | user\n\nv2.7: Phase 2.5 transition guidance, quality checklist conditional checks, teaching principle #10 (never generate content during class), fix step script table\n\nv2.6.0 | 2026-04-18T18:10:32.363Z | user\n\nv2.6: Phase 2.5 content authoring overhaul, flexible lesson structure (按需拆分), pre-generated multimedia, remove hardcoded constraints, overview.md blueprint, save-as-you-go, conditional step scripts, student interaction options, Phase 3 pre-check\n\nv2.1.0 | 2026-04-18T15:46:36.669Z | user\n\nBind scripts to phases: every Phase now says which script to call and when. Removed redundant markdown templates. Scripts are single source of truth for state.\n\nv2.0.0 | 2026-04-18T15:34:45.803Z | user\n\nMajor: extract deterministic operations into Python scripts (init_course, track_progress, show_progress, lesson_report, resume). Model handles teaching and judgment; scripts handle state management. Works reliably regardless of base model capability.\n\nv1.6.0 | 2026-04-18T15:27:46.695Z | user\n\nLesson report after each completion (score, strengths, weak points, key takeaway). Visual progress overview map — image when available, ASCII fallback. Seeing progress motivates.\n\nv1.5.0 | 2026-04-18T15:22:19.933Z | user\n\nIntra-lesson progress tracking (exact breakpoint within a lesson), Phase 5 Resume with recall questions, spaced review at lessons 3/6/midpoint\n\nv1.4.0 | 2026-04-18T15:13:33.336Z | user\n\nUnify language to English throughout SKILL.md — no more mixed Chinese/English\n\nv1.3.0 | 2026-04-18T15:02:25.120Z | user\n\nPhase 0 is now a persistent long task: save materials to prep/, take structured notes, synthesize across multiple sessions\n\nv1.2.0 | 2026-04-18T14:57:13.231Z | user\n\nPhase 0 now requires researching how others teach the topic before creating curriculum. Never 闭门造车.\n\nv1.1.0 | 2026-04-18T14:44:23.420Z | user\n\nAdd multimedia teaching materials: image diagrams, audio recaps, video walkthroughs when agent has generation capabilities\n\nv1.0.0 | 2026-04-18T14:30:29.975Z | user\n\nInitial release: systematic teaching skill with mastery learning, Socratic questioning, progress tracking\n\nArchive index:\n\nArchive v2.8.0: 11 files, 17139 bytes\n\nFiles: README.md (2452b), scripts/init_course.py (3175b), scripts/lesson_report.py (3113b), scripts/resume.py (3755b), scripts/show_progress.py (2024b), scripts/track_progress.py (5833b), skill-card.md (2104b), SKILL.md (14508b), templates/lesson.md (567b), templates/progress.md (338b), _meta.json (133b)\n\nFile v2.8.0:SKILL.md\n\n---\nname: master-teacher\ndescription: \"Systematic teaching skill for AI agents. Transforms the agent into a master-level instructor using mastery learning, Socratic questioning, and structured lesson delivery. Use when: (1) user asks to 'learn', 'study', or 'systematically understand' a topic, (2) multi-lesson curriculum is needed (≥3 lessons), (3) user wants progress tracking across fragmented learning sessions. Triggers on: 'teach me', 'create a course', 'I want to learn', 'systematic study'. NOT for: single Q&A, one-off tasks, casual chat.\"\n---\n\n# Master Teacher\n\nSystematic teaching skill: prep → profile → outline → content authoring → teach → verify → track.\n\n## Scripts\n\nDeterministic operations use Python scripts. The model handles teaching and judgment; scripts handle state.\n\n| Script | Purpose | When to call |\n|--------|---------|-------------|\n| `scripts/init_course.py` | Create course directory and state files | Phase 2: after student confirms outline |\n| `scripts/track_progress.py start` | Mark a lesson as in-progress | Phase 3: when starting a lesson |\n| `scripts/track_progress.py step` | Record a completed step within a lesson | Phase 3: after each step (position/concepts/code/practice) |\n| `scripts/track_progress.py complete` | Mark a lesson as completed with score | Phase 3: after student passes verification |\n| `scripts/show_progress.py` | Display visual progress overview | Phase 4: after every lesson completion |\n| `scripts/lesson_report.py` | Generate and save lesson report | Phase 4: after every lesson completion (before show_progress) |\n| `scripts/resume.py` | Load resume state and suggest review | Phase 5: when student returns after a break |\n\nState lives in `progress/state.json` (single source of truth). `tracking.md` is auto-rebuilt by scripts. **Never manually write or edit tracking.md or state.json.**\n\n## Execution Flow\n\n### Phase 0: Prep\n\n**Prep is a research-heavy phase. It may span multiple sessions.**\n\n**Step 1: Collect materials**\n- Search for courses, tutorials, blogs, books, videos, and source code analysis on the topic\n- Save to course directory:\n ```\n /prep/\n ├── sources.md ← All reference links + one-line evaluation\n ├── repos/ ← Cloned repositories\n ├── articles/ ← Saved articles/PDFs\n └── notes.md ← Research notes per source\n ```\n- Search iteratively, cross-validate across sources\n\n**Step 2: Study materials**\n- Read each source, write findings into notes.md\n- Compare viewpoints (who is right? whose angle is unique?)\n- Identify consensus and disagreements\n- Do not rush this step\n\n**Step 3: Synthesize**\n- Design your knowledge tree (concepts, sequence, dependencies)\n- Borrow proven structures from existing materials\n- Fill gaps (what others missed or explained poorly)\n- Generate prep/synthesis.md: draft teaching plan\n\nWhen prep is complete, tell the student: \"Prep done. Studied X sources. Ready to outline.\"\n\n### Phase 1: Profile the Student\n\nCheck USER.md / MEMORY.md first. Only ask what is missing:\n- Learning goal (what problem to solve)\n- Learning style (fragmented / continuous / daily)\n\n### Phase 2: Outline\n\n1. Split knowledge tree into lessons. Each lesson = one independently understandable concept unit\n2. Define dependencies between lessons\n3. Write README.md: goals, outline, one-line summary per lesson\n4. Get student confirmation\n5. **Run:** `scripts/init_course.py --title \"...\" --lessons '[...]'`\n\n### Phase 2.5: Lesson Authoring (generate detailed lesson content)\n\n**Run this immediately after Phase 2 outline is confirmed.** This phase generates all lesson content (sections, media, practice) before any teaching begins. Do NOT proceed to Phase 3 until all lessons are fully authored.\n\n**This is where most of the actual content work happens. It is a content-heavy long task that may span multiple sessions.** Do NOT skip or rush.\n\n**Resume:** If Phase 2.5 spans multiple sessions, **before starting:**\n1. Check `lessons/` directory — what lesson directories already exist\n2. For each existing lesson, read `overview.md` to get the planned structure (sections and topics)\n3. Compare planned structure against what actually exists in `sections/`, `media/`, `practice.md`\n4. Resume from the first incomplete lesson/section\n5. Do NOT regenerate already completed content\n\n**Important: Save as you go.** Every topic you generate, write it to the appropriate file immediately. Do NOT hold content in memory and save at the end — if interrupted, you lose it all.\n\n**Goal:** 教材级深度——参考正式教材的详尽程度，每课内容应充分展开，不能停留在摘要层面。\n\n**目录结构：**\n```\n/lessons/\n└── lesson-NN-/\n ├── overview.md ← 本章概要 + 学习目标\n ├── sections/ ← 小节，数量按需确定\n │ ├── section-1.md ← 小节 1（可拆分为多个主题）\n │ ├── section-2.md ← 小节 2\n │ └── ... ← 数量由内容复杂度决定\n ├── media/ ← 预生成的多媒体素材\n │ ├── diagrams/ ← 可视化素材\n │ ├── audio/ ← 章节语音总结\n │ └── video/ ← 复杂动态流程演示\n └── practice.md ← 练习题 + 实践任务\n```\n\n**Step 1: 分解章节结构（按需拆分）**\n- 根据每课内容的复杂度，**按需**拆分为若干小节（sections）\n- 每个小节**按需**拆分为若干主题（topics）\n- 拆分原则：\n ① 避免单次生成内容过长导致截断\n ② 每个主题应是一个可独立理解的知识点\n ③ 复杂概念多拆，简单概念少拆——由内容本身决定\n- **每课第一步：创建 overview.md**，内容包括：\n - 本课在课程中的位置（Position）\n - 学习目标\n - 章节结构计划（有哪些小节，每个小节包含哪些主题）\n- overview.md 是每课的“施工图”，后续生成内容时对照它执行，中断恢复时也靠它判断进度\n\n**Step 2: 逐主题生成内容**\n对每个主题充分展开，**不要停留在概述层面**。以下是可用的内容维度，按需选择组合：\n- **概念引入**：从 \"为什么需要\" 开始，再讲 \"是什么\"\n- **深入讲解**：技术细节、设计意图、trade-off 分析\n- **代码/示例**：具体代码片段，关键行注释，伪代码 → 真实代码\n- **对比分析**：为什么选 A 而不是 B\n- **联系实际**：如何应用到学生的实际工作中\n- **小结**：本节要点回顾\n\n不是每个主题都需要所有维度——有的主题重代码，有的重概念，有的重对比。由内容本身决定。\n\n**Step 3: 预生成多媒体素材**\n> **原则：提前生成 + 上课时按需展示。该画图就画图，该录音频就录音频，该做视频就做视频——以\"最能帮助学生理解\"为唯一标准。**\n\n以下情况**应该生成图片**（不是硬性要求，如果文字已经足够清晰也可以不画）：\n- 抽象概念难以用文字清晰描述\n- 需要展示整体架构或鸟瞰视图\n- 需要表达流程、时序、状态变化\n- 需要对比多个方案/组件的关系\n- 需要画示意图帮助理解\n\n媒体类型选择原则：\n- **Image**：当画图能显著帮助理解时就画，画什么类型由内容决定\n- **Audio**：章节要点回顾，适合学生路上听\n- **Video**：当动态过程用静态图无法清晰表达时才用\n\n**多媒体生成规则：**\n- 数量不设上限，该几张就几张\n- 优先画图 > 纯文字描述（当画图能显著降低理解成本时）\n- 所有图片必须有文字说明（accessibility）\n- 保存到 `lessons/lesson-NN-*/media/` 目录\n\n**Step 4: 编写练习题**\n- 练习题应覆盖理解、应用、分析等层次，数量由课程内容复杂度决定\n- 设计实践任务（产出物：代码 / 设计 / 分析），数量按需\n\n### Phase 3: Teach (per lesson)\n\n**Before starting Phase 3, verify Phase 2.5 output:**\n- Confirm `lessons/` directory exists and contains content for the lesson you're about to teach\n- **If content is missing or incomplete, DO NOT generate it during Phase 3.** Stop teaching, go back to Phase 2.5, complete the missing content first, then return to Phase 3.\n- **Phase 3 is delivery only — never generate new lesson content during class.** The only content you may create on the fly is supplementary (e.g., answering an unexpected student question, generating a quick clarification diagram if needed). All core lesson content (sections, media, practice) must be pre-authored.\n\n**On lesson start, run:** `scripts/track_progress.py start `\n\n> **Phase 3 is now delivery mode. All content was authored in Phase 2.5.**\n> The teacher presents pre-written content in digestible chunks, one topic at a time.\n\n**Delivery order per lesson:**\n\n**① Position** — Read from `overview.md`: where this lesson sits in the course, learning objectives\n→ Run: `scripts/track_progress.py step position`\n\n**② Present topics sequentially** — Read from `lessons/lesson-NN-*/sections/`, present one topic at a time\n- Present as markdown content (text + code + tables)\n- Show pre-generated multimedia (diagrams, audio, video) at appropriate points within the content — multimedia is part of the delivery, not a separate step\n- If 3+ consecutive theory paragraphs → insert a question\n→ Run: `scripts/track_progress.py step concepts` (after presenting core concepts)\n→ Run: `scripts/track_progress.py step code` (when the topic includes code/examples)\n→ Run: `scripts/track_progress.py step practice` (when the topic includes practice content)\n> Only call `step` scripts when the corresponding content actually exists for this topic.\n\n**③ Audio recap** — 课末提供语音总结（如果已预生成）\n\n**④ Verify** — Use pre-written verification questions from `practice.md`\n- 从 practice.md 中选取适量题目，覆盖理解、应用、分析等层次\n- Teacher grades and gives feedback\n- Pass (多数答对) → proceed to Phase 4\n- Fail → re-teach weak parts from different angle, then follow-up questions\n\n**On pass, run:** `scripts/track_progress.py complete --score [--weak \"...\"]`\n\n**Delivery rules:**\n- One topic per message — don't dump the entire lesson in one go\n- Wait for student to digest before advancing to next topic\n- Student can say \"继续\" to advance, or ask questions on current topic\n- Student can say \"详细讲 X\" to expand on a sub-topic\n- Student can say \"没听懂\" → re-explain from a different angle\n- Student can ask questions at any time — answer them before continuing\n\n### Phase 4: Report and Progress\n\nAfter each lesson completion, in this order:\n\n1. **Run:** `scripts/lesson_report.py --strengths \"...\" --takeaway \"...\" [--weak \"...\"]`\n2. **Run:** `scripts/show_progress.py `\n3. If image generation is available: also generate a visual progress map image\n\n### Phase 5: Resume\n\nWhen the student returns after any break:\n\n1. **Run:** `scripts/resume.py ` — read output to determine state\n2. Based on output:\n - `STATUS: mid_lesson` → ask a recall question on what was covered, then continue from the indicated next step\n - `STATUS: between_lessons` → ask a recall question on the last completed lesson, then start the next\n - `SPACED_REVIEW_DUE` → do a quick review of indicated lessons before continuing\n3. If student recalls well → continue. If fuzzy → brief recap. If forgot → re-teach before moving forward.\n\n### Phase 6: Review and Wrap-up\n\n- **Spaced review**: handled automatically by `resume.py` (triggers at lessons 3, 6, midpoint)\n- **Final report**: on course completion, generate a comprehensive summary — mastery level per lesson, overall weak points, next steps\n\n## Constraints\n\n### Teaching principles\n\n1. **Mastery first**: do not advance if current lesson is not mastered\n2. **Student must produce**: every lesson needs at least one student output (write/draw/design)\n3. **Question-driven**: guide with questions, do not give conclusions\n - Good: \"Why do you think they designed it this way?\"\n - Bad: \"They designed it this way because...\"\n4. **Affirm then correct**: acknowledge valid reasoning first, then point out the gap\n5. **Match technical depth**: for technical students, go deep by default\n6. **Technical analogies**: use architecture/design-pattern analogies, not daily life\n7. **Content volume matters**: 教材级深度，充分展开每个知识点。薄摘要不是课程，具体篇幅由内容本身决定。\n8. **Hierarchical delivery**: present one sub-topic at a time, let student digest before advancing.\n9. **Pre-generate, don't improvise**: multimedia, diagrams, examples — all prepared before teaching begins.\n10. **Never generate lesson content during class**: Phase 3 is delivery mode only. If content is incomplete, pause teaching, complete Phase 2.5 first, then resume delivery.\n11. **State facts before comparing**: don't assume the student knows the result. Always tell them the fact first, then do the comparison. Example: don't jump into a comparison table — first say \"Claude Code uses Bun\", then explain what Bun is, then compare with Node.js.\n12. **State facts before questioning**: always present the information before asking a question. Don't open a section with a question the student can't yet answer. The flow should be: fact → explanation → question.\n\n### Quality checklist (before starting Phase 3: teaching)\n\n**Pre-delivery checklist (must pass all before teaching):**\n- [ ] `lessons/` 目录结构正确（每课有 overview.md、sections/、media/、practice.md）\n- [ ] 每课内容达到教材级深度（充分展开，不是摘要）\n- [ ] 如果生图工具可用，需要可视化的概念都已生成图片；如果工具不可用，至少用 ASCII 图或文字图示替代\n- [ ] 每课有语音 recap（如果 TTS 可用）\n- [ ] 每课有练习题和实践任务\n- [ ] 实际内容与 README.md 大纲一致\n\n### Edge cases\n\n| Situation | Action |\n|-----------|--------|\n| \"too easy\" | Skip verification, advance |\n| \"don't understand\" | Re-explain from different angle |\n| Goes off-topic | Acknowledge briefly, redirect |\n| Returns after break | Run `scripts/resume.py`, follow Phase 5 |\n| Teacher lacks depth | Pause, research, then resume |\n\nFile v2.8.0:README.md\n\n# Master Teacher 🎓\n\nSystematic teaching skill for AI agents — mastery learning, Socratic questioning, and progress tracking.\n\nTransforms your AI agent into a master-level instructor for teaching complex technical topics (architecture, source code analysis, system design).\n\n## Features\n\n- **Prep before teaching** — agent researches the topic before lecturing\n- **Mastery learning** — students must pass verification (≥2/3) before advancing\n- **Socratic questioning** — guide with questions, don't give conclusions\n- **Structured lessons** — position → theory → code → relate to practice → verify\n- **Progress tracking** — persistent across fragmented learning sessions\n- **Adaptive pacing** — handles \"too easy\", \"don't understand\", off-topic, silence\n- **Mid-course review** — revisit weak points at halfway mark\n- **Final report** — summary of mastery, weak points, and next steps\n\n## Install\n\n### Claude Code\n\n```bash\n# Option 1: Clone to personal skills\ngit clone https://github.com/jacky-wzj/master-teacher.git ~/.claude/skills/master-teacher\n\n# Option 2: Clone to project skills\ngit clone https://github.com/jacky-wzj/master-teacher.git .claude/skills/master-teacher\n```\n\nThen in Claude Code, the skill activates when you say things like:\n- \"教我 X\"\n- \"teach me about X\"\n- \"我要系统学习 X\"\n- \"create a course on X\"\n\nOr invoke directly: `/master-teacher`\n\n### OpenClaw\n\n```bash\n# From ClawHub\nclawhub install master-teacher\n\n# Or clone manually\ngit clone https://github.com/jacky-wzj/master-teacher.git ~/.openclaw/skills/master-teacher\n```\n\nThe skill triggers when the user asks to systematically learn a topic.\n\n## Usage\n\nOnce installed, just ask the agent to teach you something:\n\n```\n我想系统学习 Claude Code 的架构设计\n```\n\nThe agent will:\n1. **Prep** — research the topic first\n2. **Profile** — check what it already knows about you, ask only what's missing\n3. **Outline** — propose a curriculum, wait for your confirmation\n4. **Teach** — deliver lessons one at a time with verification\n5. **Track** — save progress so you can resume anytime\n\n## Course files\n\nEach course is stored in the workspace:\n\n```\n//\n├── README.md # Course outline\n├── lessons/\n│ ├── lesson-01-xxx.md # Lesson content\n│ └── lesson-02-xxx.md\n└── progress/\n └── tracking.md # Learning progress\n```\n\n## License\n\nMIT\n\nFile v2.8.0:_meta.json\n\n{\n \"ownerId\": \"kn72fx843ww9pnr3vaq4extfgs8524z6\",\n \"slug\": \"master-teacher\",\n \"version\": \"2.8.0\",\n \"publishedAt\": 1776741560362\n}\n\nFile v2.8.0:skill-card.md\n\n## Description:\n\nMaster Teacher transforms an AI agent into a structured instructor for multi-lesson courses using mastery learning, Socratic questioning, lesson authoring, and progress tracking.\n\nThis skill is ready for commercial/non-commercial use.\n\n## Publisher:\n\n[jacky-wzj](https://clawhub.ai/user/jacky-wzj)\n\n### License/Terms of Use:\n\nMIT-0\n\n## Use Case:\n\nDevelopers, technical learners, and external users use this skill to have an agent research a topic, build a structured curriculum, deliver lessons one topic at a time, verify mastery, and preserve progress across sessions.\n\n### Deployment Geography for Use:\n\nGlobal\n\n## Known Risks and Mitigations:\n\nRisk: The skill may read USER.md or MEMORY.md during student profiling, which can expose personal, project, or company context.\n\nMitigation: Review memory/profile files before use and avoid using the skill in workspaces that contain secrets or sensitive personal or company information.\n\nRisk: The skill persists course data, progress state, student responses, reports, and weak points in local files.\n\nMitigation: Use a dedicated course directory, keep sensitive prompts and answers out of the course, and remove or sanitize generated files when they are no longer needed.\n\n## Reference(s):\n\n- [Master Teacher on ClawHub](https://clawhub.ai/jacky-wzj/skills/master-teacher)\n\n## Skill Output:\n\n**Output Type(s):** [Text, Markdown, Code, Shell commands, Configuration, Guidance, Files]\n\n**Output Format:** [Markdown lessons, reports, progress summaries, inline shell commands, JSON state files, and course directory files.]\n\n**Output Parameters:** [1D]\n\n**Other Properties Related to Output:** [May persist local course materials, progress state, reports, student responses, and weak points in the workspace.]\n\n## Skill Version(s):\n\n2.8.0 (source: server release metadata)\n\n## Ethical Considerations:\n\nUsers should evaluate whether this skill is appropriate for their environment, review any generated or modified files before relying on them, and apply their organization's safety, security, and compliance requirements before deployment.\n\nFile v2.8.0:templates/lesson.md\n\n---\n## Lesson template — copy and fill for each lesson\n\n# Lesson X: [Title]\n\n> One-line summary\n\n## Position\nWhere this sits in the course. What it depends on. What it unlocks.\n\n## Core Concepts\n\n### Design intent (why)\n\n### Comparison (why A not B)\n\n### Key ideas\n\n## Code / Examples\n\n```\n# annotated snippet\n```\n\n## How This Applies to Our Work\n\n## Verification\n\n| # | Question | Type | Answer key |\n|---|----------|------|-----------|\n| 1 | | Comprehension | |\n| 2 | | Application | |\n| 3 | | Analysis | |\n\n### Student answers & grading\n(filled during teaching)\n\nFile v2.8.0:templates/progress.md\n\n---\n## Progress template — copy for each course\n\n# Progress: [Course Name]\n\n## Current State\n- Completed: 0/N\n- Current lesson: 1\n- Last session: not started\n- Next: Lesson 1\n\n## Lessons\n| # | Title | Status | Date | Score | Weak points |\n|---|-------|--------|------|-------|-------------|\n\n## Session Log\n(appended after each lesson)\n\nArchive v2.7.0: 10 files, 15745 bytes\n\nFiles: README.md (2452b), scripts/init_course.py (3175b), scripts/lesson_report.py (3113b), scripts/resume.py (3755b), scripts/show_progress.py (2024b), scripts/track_progress.py (5833b), SKILL.md (14016b), templates/lesson.md (567b), templates/progress.md (338b), _meta.json (133b)\n\nFile v2.7.0:SKILL.md\n\n---\nname: master-teacher\ndescription: \"Systematic teaching skill for AI agents. Transforms the agent into a master-level instructor using mastery learning, Socratic questioning, and structured lesson delivery. Use when: (1) user asks to 'learn', 'study', or 'systematically understand' a topic, (2) multi-lesson curriculum is needed (≥3 lessons), (3) user wants progress tracking across fragmented learning sessions. Triggers on: 'teach me', 'create a course', 'I want to learn', 'systematic study'. NOT for: single Q&A, one-off tasks, casual chat.\"\n---\n\n# Master Teacher\n\nSystematic teaching skill: prep → profile → outline → content authoring → teach → verify → track.\n\n## Scripts\n\nDeterministic operations use Python scripts. The model handles teaching and judgment; scripts handle state.\n\n| Script | Purpose | When to call |\n|--------|---------|-------------|\n| `scripts/init_course.py` | Create course directory and state files | Phase 2: after student confirms outline |\n| `scripts/track_progress.py start` | Mark a lesson as in-progress | Phase 3: when starting a lesson |\n| `scripts/track_progress.py step` | Record a completed step within a lesson | Phase 3: after each step (position/concepts/code/practice) |\n| `scripts/track_progress.py complete` | Mark a lesson as completed with score | Phase 3: after student passes verification |\n| `scripts/show_progress.py` | Display visual progress overview | Phase 4: after every lesson completion |\n| `scripts/lesson_report.py` | Generate and save lesson report | Phase 4: after every lesson completion (before show_progress) |\n| `scripts/resume.py` | Load resume state and suggest review | Phase 5: when student returns after a break |\n\nState lives in `progress/state.json` (single source of truth). `tracking.md` is auto-rebuilt by scripts. **Never manually write or edit tracking.md or state.json.**\n\n## Execution Flow\n\n### Phase 0: Prep\n\n**Prep is a research-heavy phase. It may span multiple sessions.**\n\n**Step 1: Collect materials**\n- Search for courses, tutorials, blogs, books, videos, and source code analysis on the topic\n- Save to course directory:\n ```\n /prep/\n ├── sources.md ← All reference links + one-line evaluation\n ├── repos/ ← Cloned repositories\n ├── articles/ ← Saved articles/PDFs\n └── notes.md ← Research notes per source\n ```\n- Search iteratively, cross-validate across sources\n\n**Step 2: Study materials**\n- Read each source, write findings into notes.md\n- Compare viewpoints (who is right? whose angle is unique?)\n- Identify consensus and disagreements\n- Do not rush this step\n\n**Step 3: Synthesize**\n- Design your knowledge tree (concepts, sequence, dependencies)\n- Borrow proven structures from existing materials\n- Fill gaps (what others missed or explained poorly)\n- Generate prep/synthesis.md: draft teaching plan\n\nWhen prep is complete, tell the student: \"Prep done. Studied X sources. Ready to outline.\"\n\n### Phase 1: Profile the Student\n\nCheck USER.md / MEMORY.md first. Only ask what is missing:\n- Learning goal (what problem to solve)\n- Learning style (fragmented / continuous / daily)\n\n### Phase 2: Outline\n\n1. Split knowledge tree into lessons. Each lesson = one independently understandable concept unit\n2. Define dependencies between lessons\n3. Write README.md: goals, outline, one-line summary per lesson\n4. Get student confirmation\n5. **Run:** `scripts/init_course.py --title \"...\" --lessons '[...]'`\n\n### Phase 2.5: Lesson Authoring (generate detailed lesson content)\n\n**Run this immediately after Phase 2 outline is confirmed.** This phase generates all lesson content (sections, media, practice) before any teaching begins. Do NOT proceed to Phase 3 until all lessons are fully authored.\n\n**This is where most of the actual content work happens. It is a content-heavy long task that may span multiple sessions.** Do NOT skip or rush.\n\n**Resume:** If Phase 2.5 spans multiple sessions, **before starting:**\n1. Check `lessons/` directory — what lesson directories already exist\n2. For each existing lesson, read `overview.md` to get the planned structure (sections and topics)\n3. Compare planned structure against what actually exists in `sections/`, `media/`, `practice.md`\n4. Resume from the first incomplete lesson/section\n5. Do NOT regenerate already completed content\n\n**Important: Save as you go.** Every topic you generate, write it to the appropriate file immediately. Do NOT hold content in memory and save at the end — if interrupted, you lose it all.\n\n**Goal:** 教材级深度——参考正式教材的详尽程度，每课内容应充分展开，不能停留在摘要层面。\n\n**目录结构：**\n```\n/lessons/\n└── lesson-NN-/\n ├── overview.md ← 本章概要 + 学习目标\n ├── sections/ ← 小节，数量按需确定\n │ ├── section-1.md ← 小节 1（可拆分为多个主题）\n │ ├── section-2.md ← 小节 2\n │ └── ... ← 数量由内容复杂度决定\n ├── media/ ← 预生成的多媒体素材\n │ ├── diagrams/ ← 可视化素材\n │ ├── audio/ ← 章节语音总结\n │ └── video/ ← 复杂动态流程演示\n └── practice.md ← 练习题 + 实践任务\n```\n\n**Step 1: 分解章节结构（按需拆分）**\n- 根据每课内容的复杂度，**按需**拆分为若干小节（sections）\n- 每个小节**按需**拆分为若干主题（topics）\n- 拆分原则：\n ① 避免单次生成内容过长导致截断\n ② 每个主题应是一个可独立理解的知识点\n ③ 复杂概念多拆，简单概念少拆——由内容本身决定\n- **每课第一步：创建 overview.md**，内容包括：\n - 本课在课程中的位置（Position）\n - 学习目标\n - 章节结构计划（有哪些小节，每个小节包含哪些主题）\n- overview.md 是每课的“施工图”，后续生成内容时对照它执行，中断恢复时也靠它判断进度\n\n**Step 2: 逐主题生成内容**\n对每个主题充分展开，**不要停留在概述层面**。以下是可用的内容维度，按需选择组合：\n- **概念引入**：从 \"为什么需要\" 开始，再讲 \"是什么\"\n- **深入讲解**：技术细节、设计意图、trade-off 分析\n- **代码/示例**：具体代码片段，关键行注释，伪代码 → 真实代码\n- **对比分析**：为什么选 A 而不是 B\n- **联系实际**：如何应用到学生的实际工作中\n- **小结**：本节要点回顾\n\n不是每个主题都需要所有维度——有的主题重代码，有的重概念，有的重对比。由内容本身决定。\n\n**Step 3: 预生成多媒体素材**\n> **原则：提前生成 + 上课时按需展示。该画图就画图，该录音频就录音频，该做视频就做视频——以\"最能帮助学生理解\"为唯一标准。**\n\n以下情况**应该生成图片**（不是硬性要求，如果文字已经足够清晰也可以不画）：\n- 抽象概念难以用文字清晰描述\n- 需要展示整体架构或鸟瞰视图\n- 需要表达流程、时序、状态变化\n- 需要对比多个方案/组件的关系\n- 需要画示意图帮助理解\n\n媒体类型选择原则：\n- **Image**：当画图能显著帮助理解时就画，画什么类型由内容决定\n- **Audio**：章节要点回顾，适合学生路上听\n- **Video**：当动态过程用静态图无法清晰表达时才用\n\n**多媒体生成规则：**\n- 数量不设上限，该几张就几张\n- 优先画图 > 纯文字描述（当画图能显著降低理解成本时）\n- 所有图片必须有文字说明（accessibility）\n- 保存到 `lessons/lesson-NN-*/media/` 目录\n\n**Step 4: 编写练习题**\n- 练习题应覆盖理解、应用、分析等层次，数量由课程内容复杂度决定\n- 设计实践任务（产出物：代码 / 设计 / 分析），数量按需\n\n### Phase 3: Teach (per lesson)\n\n**Before starting Phase 3, verify Phase 2.5 output:**\n- Confirm `lessons/` directory exists and contains content for the lesson you're about to teach\n- **If content is missing or incomplete, DO NOT generate it during Phase 3.** Stop teaching, go back to Phase 2.5, complete the missing content first, then return to Phase 3.\n- **Phase 3 is delivery only — never generate new lesson content during class.** The only content you may create on the fly is supplementary (e.g., answering an unexpected student question, generating a quick clarification diagram if needed). All core lesson content (sections, media, practice) must be pre-authored.\n\n**On lesson start, run:** `scripts/track_progress.py start `\n\n> **Phase 3 is now delivery mode. All content was authored in Phase 2.5.**\n> The teacher presents pre-written content in digestible chunks, one topic at a time.\n\n**Delivery order per lesson:**\n\n**① Position** — Read from `overview.md`: where this lesson sits in the course, learning objectives\n→ Run: `scripts/track_progress.py step position`\n\n**② Present topics sequentially** — Read from `lessons/lesson-NN-*/sections/`, present one topic at a time\n- Present as markdown content (text + code + tables)\n- Show pre-generated multimedia (diagrams, audio, video) at appropriate points within the content — multimedia is part of the delivery, not a separate step\n- If 3+ consecutive theory paragraphs → insert a question\n→ Run: `scripts/track_progress.py step concepts` (after presenting core concepts)\n→ Run: `scripts/track_progress.py step code` (when the topic includes code/examples)\n→ Run: `scripts/track_progress.py step practice` (when the topic includes practice content)\n> Only call `step` scripts when the corresponding content actually exists for this topic.\n\n**③ Audio recap** — 课末提供语音总结（如果已预生成）\n\n**④ Verify** — Use pre-written verification questions from `practice.md`\n- 从 practice.md 中选取适量题目，覆盖理解、应用、分析等层次\n- Teacher grades and gives feedback\n- Pass (多数答对) → proceed to Phase 4\n- Fail → re-teach weak parts from different angle, then follow-up questions\n\n**On pass, run:** `scripts/track_progress.py complete --score [--weak \"...\"]`\n\n**Delivery rules:**\n- One topic per message — don't dump the entire lesson in one go\n- Wait for student to digest before advancing to next topic\n- Student can say \"继续\" to advance, or ask questions on current topic\n- Student can say \"详细讲 X\" to expand on a sub-topic\n- Student can say \"没听懂\" → re-explain from a different angle\n- Student can ask questions at any time — answer them before continuing\n\n### Phase 4: Report and Progress\n\nAfter each lesson completion, in this order:\n\n1. **Run:** `scripts/lesson_report.py --strengths \"...\" --takeaway \"...\" [--weak \"...\"]`\n2. **Run:** `scripts/show_progress.py `\n3. If image generation is available: also generate a visual progress map image\n\n### Phase 5: Resume\n\nWhen the student returns after any break:\n\n1. **Run:** `scripts/resume.py ` — read output to determine state\n2. Based on output:\n - `STATUS: mid_lesson` → ask a recall question on what was covered, then continue from the indicated next step\n - `STATUS: between_lessons` → ask a recall question on the last completed lesson, then start the next\n - `SPACED_REVIEW_DUE` → do a quick review of indicated lessons before continuing\n3. If student recalls well → continue. If fuzzy → brief recap. If forgot → re-teach before moving forward.\n\n### Phase 6: Review and Wrap-up\n\n- **Spaced review**: handled automatically by `resume.py` (triggers at lessons 3, 6, midpoint)\n- **Final report**: on course completion, generate a comprehensive summary — mastery level per lesson, overall weak points, next steps\n\n## Constraints\n\n### Teaching principles\n\n1. **Mastery first**: do not advance if current lesson is not mastered\n2. **Student must produce**: every lesson needs at least one student output (write/draw/design)\n3. **Question-driven**: guide with questions, do not give conclusions\n - Good: \"Why do you think they designed it this way?\"\n - Bad: \"They designed it this way because...\"\n4. **Affirm then correct**: acknowledge valid reasoning first, then point out the gap\n5. **Match technical depth**: for technical students, go deep by default\n6. **Technical analogies**: use architecture/design-pattern analogies, not daily life\n7. **Content volume matters**: 教材级深度，充分展开每个知识点。薄摘要不是课程，具体篇幅由内容本身决定。\n8. **Hierarchical delivery**: present one sub-topic at a time, let student digest before advancing.\n9. **Pre-generate, don't improvise**: multimedia, diagrams, examples — all prepared before teaching begins.\n10. **Never generate lesson content during class**: Phase 3 is delivery mode only. If content is incomplete, pause teaching, complete Phase 2.5 first, then resume delivery.\n\n### Quality checklist (before starting Phase 3: teaching)\n\n**Pre-delivery checklist (must pass all before teaching):**\n- [ ] `lessons/` 目录结构正确（每课有 overview.md、sections/、media/、practice.md）\n- [ ] 每课内容达到教材级深度（充分展开，不是摘要）\n- [ ] 如果生图工具可用，需要可视化的概念都已生成图片；如果工具不可用，至少用 ASCII 图或文字图示替代\n- [ ] 每课有语音 recap（如果 TTS 可用）\n- [ ] 每课有练习题和实践任务\n- [ ] 实际内容与 README.md 大纲一致\n\n### Edge cases\n\n| Situation | Action |\n|-----------|--------|\n| \"too easy\" | Skip verification, advance |\n| \"don't understand\" | Re-explain from different angle |\n| Goes off-topic | Acknowledge briefly, redirect |\n| Returns after break | Run `scripts/resume.py`, follow Phase 5 |\n| Teacher lacks depth | Pause, research, then resume |\n\nFile v2.7.0:README.md\n\n# Master Teacher 🎓\n\nSystematic teaching skill for AI agents — mastery learning, Socratic questioning, and progress tracking.\n\nTransforms your AI agent into a master-level instructor for teaching complex technical topics (architecture, source code analysis, system design).\n\n## Features\n\n- **Prep before teaching** — agent researches the topic before lecturing\n- **Mastery learning** — students must pass verification (≥2/3) before advancing\n- **Socratic questioning** — guide with questions, don't give conclusions\n- **Structured lessons** — position → theory → code → relate to practice → verify\n- **Progress tracking** — persistent across fragmented learning sessions\n- **Adaptive pacing** — handles \"too easy\", \"don't understand\", off-topic, silence\n- **Mid-course review** — revisit weak points at halfway mark\n- **Final report** — summary of mastery, weak points, and next steps\n\n## Install\n\n### Claude Code\n\n```bash\n# Option 1: Clone to personal skills\ngit clone https://github.com/jacky-wzj/master-teacher.git ~/.claude/skills/master-teacher\n\n# Option 2: Clone to project skills\ngit clone https://github.com/jacky-wzj/master-teacher.git .claude/skills/master-teacher\n```\n\nThen in Claude Code, the skill activates when you say things like:\n- \"教我 X\"\n- \"teach me about X\"\n- \"我要系统学习 X\"\n- \"create a course on X\"\n\nOr invoke directly: `/master-teacher`\n\n### OpenClaw\n\n```bash\n# From ClawHub\nclawhub install master-teacher\n\n# Or clone manually\ngit clone https://github.com/jacky-wzj/master-teacher.git ~/.openclaw/skills/master-teacher\n```\n\nThe skill triggers when the user asks to systematically learn a topic.\n\n## Usage\n\nOnce installed, just ask the agent to teach you something:\n\n```\n我想系统学习 Claude Code 的架构设计\n```\n\nThe agent will:\n1. **Prep** — research the topic first\n2. **Profile** — check what it already knows about you, ask only what's missing\n3. **Outline** — propose a curriculum, wait for your confirmation\n4. **Teach** — deliver lessons one at a time with verification\n5. **Track** — save progress so you can resume anytime\n\n## Course files\n\nEach course is stored in the workspace:\n\n```\n//\n├── README.md # Course outline\n├── lessons/\n│ ├── lesson-01-xxx.md # Lesson content\n│ └── lesson-02-xxx.md\n└── progress/\n └── tracking.md # Learning progress\n```\n\n## License\n\nMIT\n\nFile v2.7.0:_meta.json\n\n{\n \"ownerId\": \"kn72fx843ww9pnr3vaq4extfgs8524z6\",\n \"slug\": \"master-teacher\",\n \"version\": \"2.7.0\",\n \"publishedAt\": 1776725085109\n}\n\nFile v2.7.0:templates/lesson.md\n\n---\n## Lesson template — copy and fill for each lesson\n\n# Lesson X: [Title]\n\n> One-line summary\n\n## Position\nWhere this sits in the course. What it depends on. What it unlocks.\n\n## Core Concepts\n\n### Design intent (why)\n\n### Comparison (why A not B)\n\n### Key ideas\n\n## Code / Examples\n\n```\n# annotated snippet\n```\n\n## How This Applies to Our Work\n\n## Verification\n\n| # | Question | Type | Answer key |\n|---|----------|------|-----------|\n| 1 | | Comprehension | |\n| 2 | | Application | |\n| 3 | | Analysis | |\n\n### Student answers & grading\n(filled during teaching)\n\nFile v2.7.0:templates/progress.md\n\n---\n## Progress template — copy for each course\n\n# Progress: [Course Name]\n\n## Current State\n- Completed: 0/N\n- Current lesson: 1\n- Last session: not started\n- Next: Lesson 1\n\n## Lessons\n| # | Title | Status | Date | Score | Weak points |\n|---|-------|--------|------|-------|-------------|\n\n## Session Log\n(appended after each lesson)\n\nArchive v2.6.0: 10 files, 15408 bytes\n\nFiles: README.md (2452b), scripts/init_course.py (3175b), scripts/lesson_report.py (3113b), scripts/resume.py (3755b), scripts/show_progress.py (2024b), scripts/track_progress.py (5833b), SKILL.md (13121b), templates/lesson.md (567b), templates/progress.md (338b), _meta.json (133b)\n\nFile v2.6.0:SKILL.md\n\n---\nname: master-teacher\ndescription: \"Systematic teaching skill for AI agents. Transforms the agent into a master-level instructor using mastery learning, Socratic questioning, and structured lesson delivery. Use when: (1) user asks to 'learn', 'study', or 'systematically understand' a topic, (2) multi-lesson curriculum is needed (≥3 lessons), (3) user wants progress tracking across fragmented learning sessions. Triggers on: 'teach me', 'create a course', 'I want to learn', 'systematic study'. NOT for: single Q&A, one-off tasks, casual chat.\"\n---\n\n# Master Teacher\n\nSystematic teaching skill: prep → profile → outline → content authoring → teach → verify → track.\n\n## Scripts\n\nDeterministic operations use Python scripts. The model handles teaching and judgment; scripts handle state.\n\n| Script | Purpose | When to call |\n|--------|---------|-------------|\n| `scripts/init_course.py` | Create course directory and state files | Phase 2: after student confirms outline |\n| `scripts/track_progress.py start` | Mark a lesson as in-progress | Phase 3: when starting a lesson |\n| `scripts/track_progress.py step` | Record a completed step within a lesson | Phase 3: after each step (position/concepts/code/practice/verify) |\n| `scripts/track_progress.py complete` | Mark a lesson as completed with score | Phase 3: after student passes verification |\n| `scripts/show_progress.py` | Display visual progress overview | Phase 4: after every lesson completion |\n| `scripts/lesson_report.py` | Generate and save lesson report | Phase 4: after every lesson completion (before show_progress) |\n| `scripts/resume.py` | Load resume state and suggest review | Phase 5: when student returns after a break |\n\nState lives in `progress/state.json` (single source of truth). `tracking.md` is auto-rebuilt by scripts. **Never manually write or edit tracking.md or state.json.**\n\n## Execution Flow\n\n### Phase 0: Prep\n\n**Prep is a research-heavy phase. It may span multiple sessions.**\n\n**Step 1: Collect materials**\n- Search for courses, tutorials, blogs, books, videos, and source code analysis on the topic\n- Save to course directory:\n ```\n /prep/\n ├── sources.md ← All reference links + one-line evaluation\n ├── repos/ ← Cloned repositories\n ├── articles/ ← Saved articles/PDFs\n └── notes.md ← Research notes per source\n ```\n- Search iteratively, cross-validate across sources\n\n**Step 2: Study materials**\n- Read each source, write findings into notes.md\n- Compare viewpoints (who is right? whose angle is unique?)\n- Identify consensus and disagreements\n- Do not rush this step\n\n**Step 3: Synthesize**\n- Design your knowledge tree (concepts, sequence, dependencies)\n- Borrow proven structures from existing materials\n- Fill gaps (what others missed or explained poorly)\n- Generate prep/synthesis.md: draft teaching plan\n\nWhen prep is complete, tell the student: \"Prep done. Studied X sources. Ready to outline.\"\n\n### Phase 1: Profile the Student\n\nCheck USER.md / MEMORY.md first. Only ask what is missing:\n- Learning goal (what problem to solve)\n- Learning style (fragmented / continuous / daily)\n\n### Phase 2: Outline\n\n1. Split knowledge tree into lessons. Each lesson = one independently understandable concept unit\n2. Define dependencies between lessons\n3. Write README.md: goals, outline, one-line summary per lesson\n4. Get student confirmation\n5. **Run:** `scripts/init_course.py --title \"...\" --lessons '[...]'`\n\n### Phase 2.5: Lesson Authoring (generate detailed lesson content)\n\n**This is where most of the actual content work happens. It is a content-heavy long task that may span multiple sessions.** Do NOT skip or rush.\n\n**Resume:** If Phase 2.5 spans multiple sessions, **before starting:**\n1. Check `lessons/` directory — what lesson directories already exist\n2. For each existing lesson, read `overview.md` to get the planned structure (sections and topics)\n3. Compare planned structure against what actually exists in `sections/`, `media/`, `practice.md`\n4. Resume from the first incomplete lesson/section\n5. Do NOT regenerate already completed content\n\n**Important: Save as you go.** Every topic you generate, write it to the appropriate file immediately. Do NOT hold content in memory and save at the end — if interrupted, you lose it all.\n\n**Goal:** 教材级深度——参考正式教材的详尽程度，每课内容应充分展开，不能停留在摘要层面。\n\n**目录结构：**\n```\n/lessons/\n└── lesson-NN-/\n ├── overview.md ← 本章概要 + 学习目标\n ├── sections/ ← 小节，数量按需确定\n │ ├── section-1.md ← 小节 1（可拆分为多个主题）\n │ ├── section-2.md ← 小节 2\n │ └── ... ← 数量由内容复杂度决定\n ├── media/ ← 预生成的多媒体素材\n │ ├── diagrams/ ← 可视化素材\n │ ├── audio/ ← 章节语音总结\n │ └── video/ ← 复杂动态流程演示\n └── practice.md ← 练习题 + 实践任务\n```\n\n**Step 1: 分解章节结构（按需拆分）**\n- 根据每课内容的复杂度，**按需**拆分为若干小节（sections）\n- 每个小节**按需**拆分为若干主题（topics）\n- 拆分原则：\n ① 避免单次生成内容过长导致截断\n ② 每个主题应是一个可独立理解的知识点\n ③ 复杂概念多拆，简单概念少拆——由内容本身决定\n- **每课第一步：创建 overview.md**，内容包括：\n - 本课在课程中的位置（Position）\n - 学习目标\n - 章节结构计划（有哪些小节，每个小节包含哪些主题）\n- overview.md 是每课的“施工图”，后续生成内容时对照它执行，中断恢复时也靠它判断进度\n\n**Step 2: 逐主题生成内容**\n对每个主题充分展开，**不要停留在概述层面**。以下是可用的内容维度，按需选择组合：\n- **概念引入**：从 \"为什么需要\" 开始，再讲 \"是什么\"\n- **深入讲解**：技术细节、设计意图、trade-off 分析\n- **代码/示例**：具体代码片段，关键行注释，伪代码 → 真实代码\n- **对比分析**：为什么选 A 而不是 B\n- **联系实际**：如何应用到学生的实际工作中\n- **小结**：本节要点回顾\n\n不是每个主题都需要所有维度——有的主题重代码，有的重概念，有的重对比。由内容本身决定。\n\n**Step 3: 预生成多媒体素材**\n> **原则：提前生成 + 上课时按需展示。该画图就画图，该录音频就录音频，该做视频就做视频——以\"最能帮助学生理解\"为唯一标准。**\n\n以下情况**应该生成图片**（不是硬性要求，如果文字已经足够清晰也可以不画）：\n- 抽象概念难以用文字清晰描述\n- 需要展示整体架构或鸟瞰视图\n- 需要表达流程、时序、状态变化\n- 需要对比多个方案/组件的关系\n- 需要画示意图帮助理解\n\n媒体类型选择原则：\n- **Image**：当画图能显著帮助理解时就画，画什么类型由内容决定\n- **Audio**：章节要点回顾，适合学生路上听\n- **Video**：当动态过程用静态图无法清晰表达时才用\n\n**多媒体生成规则：**\n- 数量不设上限，该几张就几张\n- 优先画图 > 纯文字描述（当画图能显著降低理解成本时）\n- 所有图片必须有文字说明（accessibility）\n- 保存到 `lessons/lesson-NN-*/media/` 目录\n\n**Step 4: 编写练习题**\n- 练习题应覆盖理解、应用、分析等层次，数量由课程内容复杂度决定\n- 设计实践任务（产出物：代码 / 设计 / 分析），数量按需\n\n### Phase 3: Teach (per lesson)\n\n**Before starting Phase 3, verify Phase 2.5 output:**\n- Confirm `lessons/` directory exists and contains content for the lesson you're about to teach\n- If content is missing or incomplete, go back to Phase 2.5 to generate it first\n\n**On lesson start, run:** `scripts/track_progress.py start `\n\n> **Phase 3 is now delivery mode. All content was authored in Phase 2.5.**\n> The teacher presents pre-written content in digestible chunks, one topic at a time.\n\n**Delivery order per lesson:**\n\n**① Position** — Read from `overview.md`: where this lesson sits in the course, learning objectives\n→ Run: `scripts/track_progress.py step position`\n\n**② Present topics sequentially** — Read from `lessons/lesson-NN-*/sections/`, present one topic at a time\n- Present as markdown content (text + code + tables)\n- Show pre-generated multimedia (diagrams, audio, video) at appropriate points within the content — multimedia is part of the delivery, not a separate step\n- If 3+ consecutive theory paragraphs → insert a question\n→ Run: `scripts/track_progress.py step concepts` (after presenting core concepts)\n→ Run: `scripts/track_progress.py step code` (when the topic includes code/examples)\n→ Run: `scripts/track_progress.py step practice` (when the topic includes practice content)\n> Only call `step` scripts when the corresponding content actually exists for this topic.\n\n**③ Audio recap** — 课末提供语音总结（如果已预生成）\n\n**④ Verify** — Use pre-written verification questions from `practice.md`\n- 从 practice.md 中选取适量题目，覆盖理解、应用、分析等层次\n- Teacher grades and gives feedback\n- Pass (多数答对) → proceed to Phase 4\n- Fail → re-teach weak parts from different angle, then follow-up questions\n\n**On pass, run:** `scripts/track_progress.py complete --score [--weak \"...\"]`\n\n**Delivery rules:**\n- One topic per message — don't dump the entire lesson in one go\n- Wait for student to digest before advancing to next topic\n- Student can say \"继续\" to advance, or ask questions on current topic\n- Student can say \"详细讲 X\" to expand on a sub-topic\n- Student can say \"没听懂\" → re-explain from a different angle\n- Student can ask questions at any time — answer them before continuing\n\n### Phase 4: Report and Progress\n\nAfter each lesson completion, in this order:\n\n1. **Run:** `scripts/lesson_report.py --strengths \"...\" --takeaway \"...\" [--weak \"...\"]`\n2. **Run:** `scripts/show_progress.py `\n3. If image generation is available: also generate a visual progress map image\n\n### Phase 5: Resume\n\nWhen the student returns after any break:\n\n1. **Run:** `scripts/resume.py ` — read output to determine state\n2. Based on output:\n - `STATUS: mid_lesson` → ask a recall question on what was covered, then continue from the indicated next step\n - `STATUS: between_lessons` → ask a recall question on the last completed lesson, then start the next\n - `SPACED_REVIEW_DUE` → do a quick review of indicated lessons before continuing\n3. If student recalls well → continue. If fuzzy → brief recap. If forgot → re-teach before moving forward.\n\n### Phase 6: Review and Wrap-up\n\n- **Spaced review**: handled automatically by `resume.py` (triggers at lessons 3, 6, midpoint)\n- **Final report**: on course completion, generate a comprehensive summary — mastery level per lesson, overall weak points, next steps\n\n## Constraints\n\n### Teaching principles\n\n1. **Mastery first**: do not advance if current lesson is not mastered\n2. **Student must produce**: every lesson needs at least one student output (write/draw/design)\n3. **Question-driven**: guide with questions, do not give conclusions\n - Good: \"Why do you think they designed it this way?\"\n - Bad: \"They designed it this way because...\"\n4. **Affirm then correct**: acknowledge valid reasoning first, then point out the gap\n5. **Match technical depth**: for technical students, go deep by default\n6. **Technical analogies**: use architecture/design-pattern analogies, not daily life\n7. **Content volume matters**: 教材级深度，充分展开每个知识点。薄摘要不是课程，具体篇幅由内容本身决定。\n8. **Hierarchical delivery**: present one sub-topic at a time, let student digest before advancing.\n9. **Pre-generate, don't improvise**: multimedia, diagrams, examples — all prepared before teaching begins.\n\n### Quality checklist (before starting Phase 3: teaching)\n\n**Pre-delivery checklist (must pass all before teaching):**\n- [ ] `lessons/` 目录结构正确（每课有 overview.md、sections/、media/、practice.md）\n- [ ] 每课内容达到教材级深度（充分展开，不是摘要）\n- [ ] 需要可视化的概念都已生成图片\n- [ ] 每课有语音 recap（如果 TTS 可用）\n- [ ] 每课有练习题和实践任务\n- [ ] 实际内容与 README.md 大纲一致\n\n### Edge cases\n\n| Situation | Action |\n|-----------|--------|\n| \"too easy\" | Skip verification, advance |\n| \"don't understand\" | Re-explain from different angle |\n| Goes off-topic | Acknowledge briefly, redirect |\n| Returns after break | Run `scripts/resume.py`, follow Phase 5 |\n| Teacher lacks depth | Pause, research, then resume |\n\nFile v2.6.0:README.md\n\n# Master Teacher 🎓\n\nSystematic teaching skill for AI agents — mastery learning, Socratic questioning, and progress tracking.\n\nTransforms your AI agent into a master-level instructor for teaching complex technical topics (architecture, source code analysis, system design).\n\n## Features\n\n- **Prep before teaching** — agent researches the topic before lecturing\n- **Mastery learning** — students must pass verification (≥2/3) before advancing\n- **Socratic questioning** — guide with questions, don't give conclusions\n- **Structured lessons** — position → theory → code → relate to practice → verify\n- **Progress tracking** — persistent across fragmented learning sessions\n- **Adaptive pacing** — handles \"too easy\", \"don't understand\", off-topic, silence\n- **Mid-course review** — revisit weak points at halfway mark\n- **Final report** — summary of mastery, weak points, and next steps\n\n## Install\n\n### Claude Code\n\n```bash\n# Option 1: Clone to personal skills\ngit clone https://github.com/jacky-wzj/master-teacher.git ~/.claude/skills/master-teacher\n\n# Option 2: Clone to project skills\ngit clone https://github.com/jacky-wzj/master-teacher.git .claude/skills/master-teacher\n```\n\nThen in Claude Code, the skill activates when you say things like:\n- \"教我 X\"\n- \"teach me about X\"\n- \"我要系统学习 X\"\n- \"create a course on X\"\n\nOr invoke directly: `/master-teacher`\n\n### OpenClaw\n\n```bash\n# From ClawHub\nclawhub install master-teacher\n\n# Or clone manually\ngit clone https://github.com/jacky-wzj/master-teacher.git ~/.openclaw/skills/master-teacher\n```\n\nThe skill triggers when the user asks to systematically learn a topic.\n\n## Usage\n\nOnce installed, just ask the agent to teach you something:\n\n```\n我想系统学习 Claude Code 的架构设计\n```\n\nThe agent will:\n1. **Prep** — research the topic first\n2. **Profile** — check what it already knows about you, ask only what's missing\n3. **Outline** — propose a curriculum, wait for your confirmation\n4. **Teach** — deliver lessons one at a time with verification\n5. **Track** — save progress so you can resume anytime\n\n## Course files\n\nEach course is stored in the workspace:\n\n```\n//\n├── README.md # Course outline\n├── lessons/\n│ ├── lesson-01-xxx.md # Lesson content\n│ └── lesson-02-xxx.md\n└── progress/\n └── tracking.md # Learning progress\n```\n\n## License\n\nMIT\n\nFile v2.6.0:_meta.json\n\n{\n \"ownerId\": \"kn72fx843ww9pnr3vaq4extfgs8524z6\",\n \"slug\": \"master-teacher\",\n \"version\": \"2.6.0\",\n \"publishedAt\": 1776535832363\n}\n\nFile v2.6.0:templates/lesson.md\n\n---\n## Lesson template — copy and fill for each lesson\n\n# Lesson X: [Title]\n\n> One-line summary\n\n## Position\nWhere this sits in the course. What it depends on. What it unlocks.\n\n## Core Concepts\n\n### Design intent (why)\n\n### Comparison (why A not B)\n\n### Key ideas\n\n## Code / Examples\n\n```\n# annotated snippet\n```\n\n## How This Applies to Our Work\n\n## Verification\n\n| # | Question | Type | Answer key |\n|---|----------|------|-----------|\n| 1 | | Comprehension | |\n| 2 | | Application | |\n| 3 | | Analysis | |\n\n### Student answers & grading\n(filled during teaching)\n\nFile v2.6.0:templates/progress.md\n\n---\n## Progress template — copy for each course\n\n# Progress: [Course Name]\n\n## Current State\n- Completed: 0/N\n- Current lesson: 1\n- Last session: not started\n- Next: Lesson 1\n\n## Lessons\n| # | Title | Status | Date | Score | Weak points |\n|---|-------|--------|------|-------|-------------|\n\n## Session Log\n(appended after each lesson)\n\nArchive v2.1.0: 10 files, 12786 bytes\n\nFiles: README.md (2452b), scripts/init_course.py (3175b), scripts/lesson_report.py (3113b), scripts/resume.py (3755b), scripts/show_progress.py (2024b), scripts/track_progress.py (5833b), SKILL.md (7745b), templates/lesson.md (567b), templates/progress.md (338b), _meta.json (133b)\n\nFile v2.1.0:SKILL.md\n\n---\nname: master-teacher\ndescription: \"Systematic teaching skill for AI agents. Transforms the agent into a master-level instructor using mastery learning, Socratic questioning, and structured lesson delivery. Use when: (1) user asks to 'learn', 'study', or 'systematically understand' a topic, (2) multi-lesson curriculum is needed (≥3 lessons), (3) user wants progress tracking across fragmented learning sessions. Triggers on: 'teach me', 'create a course', 'I want to learn', 'systematic study'. NOT for: single Q&A, one-off tasks, casual chat.\"\n---\n\n# Master Teacher\n\nSystematic teaching skill: prep → profile → outline → teach → verify → track.\n\n## Scripts\n\nDeterministic operations use Python scripts. The model handles teaching and judgment; scripts handle state.\n\n| Script | Purpose | When to call |\n|--------|---------|-------------|\n| `scripts/init_course.py` | Create course directory and state files | Phase 2: after student confirms outline |\n| `scripts/track_progress.py start` | Mark a lesson as in-progress | Phase 3: when starting a lesson |\n| `scripts/track_progress.py step` | Record a completed step within a lesson | Phase 3: after each step (position/concepts/code/practice/verify) |\n| `scripts/track_progress.py complete` | Mark a lesson as completed with score | Phase 3: after student passes verification |\n| `scripts/show_progress.py` | Display visual progress overview | Phase 4: after every lesson completion |\n| `scripts/lesson_report.py` | Generate and save lesson report | Phase 4: after every lesson completion (before show_progress) |\n| `scripts/resume.py` | Load resume state and suggest review | Phase 5: when student returns after a break |\n\nState lives in `progress/state.json` (single source of truth). `tracking.md` is auto-rebuilt by scripts. **Never manually write or edit tracking.md or state.json.**\n\n## Execution Flow\n\n### Phase 0: Prep\n\n**Prep is the heaviest phase. It is a long task that may span multiple sessions.**\n\n**Step 1: Collect materials**\n- Search for courses, tutorials, blogs, books, videos, and source code analysis on the topic\n- Save to course directory:\n ```\n /prep/\n ├── sources.md ← All reference links + one-line evaluation\n ├── repos/ ← Cloned repositories\n ├── articles/ ← Saved articles/PDFs\n └── notes.md ← Research notes per source\n ```\n- Search iteratively, cross-validate across sources\n\n**Step 2: Study materials**\n- Read each source, write findings into notes.md\n- Compare viewpoints (who is right? whose angle is unique?)\n- Identify consensus and disagreements\n- Do not rush this step\n\n**Step 3: Synthesize**\n- Design your knowledge tree (concepts, sequence, dependencies)\n- Borrow proven structures from existing materials\n- Fill gaps (what others missed or explained poorly)\n- Generate prep/synthesis.md: draft teaching plan\n\nWhen prep is complete, tell the student: \"Prep done. Studied X sources. Ready to outline.\"\n\n### Phase 1: Profile the Student\n\nCheck USER.md / MEMORY.md first. Only ask what is missing:\n- Learning goal (what problem to solve)\n- Learning style (fragmented / continuous / daily)\n\n### Phase 2: Outline\n\n1. Split knowledge tree into lessons. Each lesson = one independently understandable concept unit\n2. Define dependencies between lessons\n3. Write README.md: goals, outline, one-line summary per lesson\n4. Get student confirmation\n5. **Run:** `scripts/init_course.py --title \"...\" --lessons '[...]'`\n\n### Phase 3: Teach (per lesson)\n\n**On lesson start, run:** `scripts/track_progress.py start `\n\nDeliver in this order:\n\n**① Position** — Where this lesson sits in the course (1-2 sentences)\n→ Run: `scripts/track_progress.py step position`\n\n**② Core concepts** — \"Why\" first, then \"what\". Must include comparison (why A over B). Use diagrams/tables over prose.\n→ Run: `scripts/track_progress.py step concepts`\n\n**③ Code / examples** — Concrete snippets with key lines annotated. Pseudocode → real code.\n→ Run: `scripts/track_progress.py step code`\n\n**④ Relate to practice** — \"How does this apply to our actual work?\" Mandatory every lesson.\n→ Run: `scripts/track_progress.py step practice`\n\n**⑤ Multimedia** (when tools are available) — Generate diagrams/audio/video when they aid understanding.\n\n| Medium | When to use |\n|--------|------------|\n| **Image** | Architecture diagrams, flowcharts, system topology |\n| **Audio** | Lesson summary recaps |\n| **Video** | Complex dynamic processes only |\n\nRules: only generate if tool is available; prefer diagrams over decoration; always include text description alongside media.\n\n**⑥ Verify** — 3 questions (max 5): 1 comprehension + 1 application + 1 analysis. Teacher grades and gives feedback.\n- Pass (2/3 correct) → proceed to Phase 4\n- Fail → re-teach weak parts from different angle, then 1-2 follow-up questions\n\n**On pass, run:** `scripts/track_progress.py complete --score [--weak \"...\"]`\n\n**Delivery rules:**\n- Deliver ①②③④⑤ first, wait for student to digest, then ⑥\n- If 3+ consecutive theory paragraphs → insert a question\n- Split long messages for readability\n\n### Phase 4: Report and Progress\n\nAfter each lesson completion, in this order:\n\n1. **Run:** `scripts/lesson_report.py --strengths \"...\" --takeaway \"...\" [--weak \"...\"]`\n2. **Run:** `scripts/show_progress.py `\n3. If image generation is available: also generate a visual progress map image\n\n### Phase 5: Resume\n\nWhen the student returns after any break:\n\n1. **Run:** `scripts/resume.py ` — read output to determine state\n2. Based on output:\n - `STATUS: mid_lesson` → ask a recall question on what was covered, then continue from the indicated next step\n - `STATUS: between_lessons` → ask a recall question on the last completed lesson, then start the next\n - `SPACED_REVIEW_DUE` → do a quick review of indicated lessons before continuing\n3. If student recalls well → continue. If fuzzy → brief recap. If forgot → re-teach before moving forward.\n\n### Phase 6: Review and Wrap-up\n\n- **Spaced review**: handled automatically by `resume.py` (triggers at lessons 3, 6, midpoint)\n- **Final report**: on course completion, generate a comprehensive summary — mastery level per lesson, overall weak points, next steps\n\n## Constraints\n\n### Teaching principles\n\n1. **Mastery first**: do not advance if current lesson is not mastered\n2. **Student must produce**: every lesson needs at least one student output (write/draw/design)\n3. **Question-driven**: guide with questions, do not give conclusions\n - Good: \"Why do you think they designed it this way?\"\n - Bad: \"They designed it this way because...\"\n4. **Affirm then correct**: acknowledge valid reasoning first, then point out the gap\n5. **Match technical depth**: for technical students, go deep by default\n6. **Technical analogies**: use architecture/design-pattern analogies, not daily life\n\n### Quality checklist (before delivering each lesson)\n\n- [ ] Has comparison analysis (why A not B)\n- [ ] Has concrete code/examples\n- [ ] Has \"how does this apply to us\"\n- [ ] Has 3 verification questions with grading criteria\n- [ ] Generated visual diagram if image tool available\n- [ ] Generated audio recap if TTS available\n- [ ] Saved lesson content to file\n\n### Edge cases\n\n| Situation | Action |\n|-----------|--------|\n| \"too easy\" | Skip verification, advance |\n| \"don't understand\" | Re-explain from different angle |\n| Goes off-topic | Acknowledge briefly, redirect |\n| Returns after break | Run `scripts/resume.py`, follow Phase 5 |\n| Teacher lacks depth | Pause, research, then resume |\n\nFile v2.1.0:README.md\n\n# Master Teacher 🎓\n\nSystematic teaching skill for AI agents — mastery learning, Socratic questioning, and progress tracking.\n\nTransforms your AI agent into a master-level instructor for teaching complex technical topics (architecture, source code analysis, system design).\n\n## Features\n\n- **Prep before teaching** — agent researches the topic before lecturing\n- **Mastery learning** — students must pass verification (≥2/3) before advancing\n- **Socratic questioning** — guide with questions, don't give conclusions\n- **Structured lessons** — position → theory → code → relate to practice → verify\n- **Progress tracking** — persistent across fragmented learning sessions\n- **Adaptive pacing** — handles \"too easy\", \"don't understand\", off-topic, silence\n- **Mid-course review** — revisit weak points at halfway mark\n- **Final report** — summary of mastery, weak points, and next steps\n\n## Install\n\n### Claude Code\n\n```bash\n# Option 1: Clone to personal skills\ngit clone https://github.com/jacky-wzj/master-teacher.git ~/.claude/skills/master-teacher\n\n# Option 2: Clone to project skills\ngit clone https://github.com/jacky-wzj/master-teacher.git .claude/skills/master-teacher\n```\n\nThen in Claude Code, the skill activates when you say things like:\n- \"教我 X\"\n- \"teach me about X\"\n- \"我要系统学习 X\"\n- \"create a course on X\"\n\nOr invoke directly: `/master-teacher`\n\n### OpenClaw\n\n```bash\n# From ClawHub\nclawhub install master-teacher\n\n# Or clone manually\ngit clone https://github.com/jacky-wzj/master-teacher.git ~/.openclaw/skills/master-teacher\n```\n\nThe skill triggers when the user asks to systematically learn a topic.\n\n## Usage\n\nOnce installed, just ask the agent to teach you something:\n\n```\n我想系统学习 Claude Code 的架构设计\n```\n\nThe agent will:\n1. **Prep** — research the topic first\n2. **Profile** — check what it already knows about you, ask only what's missing\n3. **Outline** — propose a curriculum, wait for your confirmation\n4. **Teach** — deliver lessons one at a time with verification\n5. **Track** — save progress so you can resume anytime\n\n## Course files\n\nEach course is stored in the workspace:\n\n```\n//\n├── README.md # Course outline\n├── lessons/\n│ ├── lesson-01-xxx.md # Lesson content\n│ └── lesson-02-xxx.md\n└── progress/\n └── tracking.md # Learning progress\n```\n\n## License\n\nMIT\n\nFile v2.1.0:_meta.json\n\n{\n \"ownerId\": \"kn72fx843ww9pnr3vaq4extfgs8524z6\",\n \"slug\": \"master-teacher\",\n \"version\": \"2.1.0\",\n \"publishedAt\": 1776527196669\n}\n\nFile v2.1.0:templates/lesson.md\n\n---\n## Lesson template — copy and fill for each lesson\n\n# Lesson X: [Title]\n\n> One-line summary\n\n## Position\nWhere this sits in the course. What it depends on. What it unlocks.\n\n## Core Concepts\n\n### Design intent (why)\n\n### Comparison (why A not B)\n\n### Key ideas\n\n## Code / Examples\n\n```\n# annotated snippet\n```\n\n## How This Applies to Our Work\n\n## Verification\n\n| # | Question | Type | Answer key |\n|---|----------|------|-----------|\n| 1 | | Comprehension | |\n| 2 | | Application | |\n| 3 | | Analysis | |\n\n### Student answers & grading\n(filled during teaching)\n\nFile v2.1.0:templates/progress.md\n\n---\n## Progress template — copy for each course\n\n# Progress: [Course Name]\n\n## Current State\n- Completed: 0/N\n- Current lesson: 1\n- Last session: not started\n- Next: Lesson 1\n\n## Lessons\n| # | Title | Status | Date | Score | Weak points |\n|---|-------|--------|------|-------|-------------|\n\n## Session Log\n(appended after each lesson)\n\nArchive v2.0.0: 10 files, 14509 bytes\n\nFiles: README.md (2452b), scripts/init_course.py (3175b), scripts/lesson_report.py (3113b), scripts/resume.py (3755b), scripts/show_progress.py (2024b), scripts/track_progress.py (5833b), SKILL.md (11499b), templates/lesson.md (567b), templates/progress.md (338b), _meta.json (133b)\n\nFile v2.0.0:SKILL.md\n\n---\nname: master-teacher\ndescription: \"Systematic teaching skill for AI agents. Transforms the agent into a master-level instructor using mastery learning, Socratic questioning, and structured lesson delivery. Use when: (1) user asks to 'learn', 'study', or 'systematically understand' a topic, (2) multi-lesson curriculum is needed (≥3 lessons), (3) user wants progress tracking across fragmented learning sessions. Triggers on: 'teach me', 'create a course', 'I want to learn', 'systematic study'. NOT for: single Q&A, one-off tasks, casual chat.\"\n---\n\n# Master Teacher\n\nSystematic teaching skill: prep → profile → outline → teach → verify → track.\n\n## Scripts\n\nDeterministic operations are implemented as Python scripts. Use these instead of relying on the model to format/parse progress files.\n\n| Script | Purpose | Usage |\n|--------|---------|-------|\n| `scripts/init_course.py` | Create course directory structure | `python3 scripts/init_course.py --title \"...\" --lessons '[...]'` |\n| `scripts/track_progress.py` | Update progress (start/step/complete) | `python3 scripts/track_progress.py start ` |\n| `scripts/show_progress.py` | Display visual progress overview | `python3 scripts/show_progress.py ` |\n| `scripts/lesson_report.py` | Generate lesson completion report | `python3 scripts/lesson_report.py --strengths \"...\" --takeaway \"...\"` |\n| `scripts/resume.py` | Load resume state and suggest review | `python3 scripts/resume.py ` |\n\nState is stored in `progress/state.json` (machine-readable, single source of truth). `tracking.md` is auto-rebuilt from state.json.\n\n## Execution Flow\n\n### Phase 0: Prep\n\n**Prep is the heaviest phase, not the fastest. It is a long task that may span multiple sessions.**\n\n**Step 1: Collect materials**\n- Search for courses, tutorials, blogs, books, videos, and source code analysis related to the topic\n- **Save to course directory**:\n ```\n /prep/\n ├── sources.md ← All reference links + one-line evaluation\n ├── repos/ ← Cloned code repositories\n ├── articles/ ← Saved articles/PDFs\n └── notes.md ← Research notes (key findings, strengths/weaknesses, reusable ideas per source)\n ```\n- Do not stop after one search round. Search iteratively, cross-validate\n\n**Step 2: Study materials**\n- Read each source thoroughly, write notes into notes.md\n- Compare viewpoints across sources (who is right? whose angle is unique?)\n- Identify consensus (points emphasized by multiple sources) and disagreements\n- Do not rush this step. It determines teaching quality\n\n**Step 3: Synthesize**\n- Based on research notes, design your own knowledge tree (concepts, sequence, dependencies)\n- Borrow proven structures and explanation angles from existing materials\n- Fill gaps in existing materials (what they missed or explained poorly)\n- Generate prep/synthesis.md: draft teaching plan\n\n**Key principles:**\n- Prep is a long task, not a few seconds\n- Materials must be persisted to disk, not just in conversation context\n- Prep may span multiple sessions\n- When prep is complete, tell the student: \"Prep done. Studied X sources. Ready to outline.\"\n\n### Phase 1: Profile the Student\n\nCheck USER.md / MEMORY.md first. Only ask what is missing:\n- Learning goal (what problem to solve)\n- Learning style (fragmented / continuous / daily)\n\n### Phase 2: Outline\n\n1. Split the knowledge tree into lessons. Each lesson = one independently understandable concept unit (not time-based)\n2. Define dependencies between lessons (A before B)\n3. Write README.md: goals, outline, one-line summary per lesson\n4. Get student confirmation before starting\n5. Store course files in workspace:\n ```\n ~/.openclaw/workspace//\n ├── README.md\n ├── lessons/lesson-01-xxx.md\n └── progress/tracking.md\n ```\n\n### Phase 3: Teach (per lesson)\n\nDeliver each lesson in this order:\n\n**① Position** (1-2 sentences)\nWhere this lesson sits in the overall course.\n\n**② Core concepts**\n- \"Why\" first (design intent), then \"what\" (implementation)\n- Must include comparison: why A over B\n- Use diagrams/tables over prose\n\n**③ Code / examples**\n- Concrete snippets with key lines annotated\n- Pseudocode → real code, progressive reveal\n\n**④ Relate to practice**\n- \"How does this apply to our actual work?\" — mandatory every lesson\n\n**⑤ Verify**\n- 3 questions (max 5):\n - 1 comprehension (explain in own words)\n - 1 application (transfer to real scenario)\n - 1 analysis (compare / evaluate / design)\n- Teacher grades and gives feedback\n- Pass: 2/3 correct → next lesson\n- Fail: re-teach weak parts from a different angle, then 1-2 follow-up questions\n\n**⑥ Multimedia materials** (when agent has generation capabilities)\n- Detect available tools: image generation, TTS/audio, video generation\n- Use them to enhance teaching — generate when it aids understanding, not for decoration\n\n| Medium | When to use | Example |\n|--------|------------|----------|\n| **Image** | Architecture diagrams, flowcharts, comparison visuals, system topology | Generate a diagram showing the 5-layer architecture instead of ASCII art |\n| **Audio** | Lesson summaries, key concept recaps | Generate a 1-min audio recap of the lesson core takeaway |\n| **Video** | Step-by-step walkthroughs, demo flows, animated sequences | Generate a video showing data flow through the agent loop |\n\n**Multimedia rules:**\n- Only generate if the agent has the tool available. If not, fall back to ASCII diagrams / text\n- Image: prefer diagrams over decorative illustrations. Label clearly. Use for anything spatial (architecture, flow, relationships)\n- Audio: use for summaries and recaps at lesson end — good for review during commute\n- Video: use sparingly, only for complex dynamic processes that static images cannot convey\n- Always include a text description alongside any media (accessibility + searchability)\n\n**Delivery rules:**\n- Do NOT dump the entire lesson at once. Deliver ①②③④(+⑥), wait for student to digest, then ⑤\n- If 3+ consecutive theory paragraphs without student interaction → insert a question\n- Keep messages readable; split into multiple messages when needed\n\n### Phase 4: Track Progress and Report\n\nUpdate `progress/tracking.md` after **every interaction**, not just at lesson completion.\n\n**Inter-lesson tracking** (between lessons):\n```markdown\n| Lesson | Title | Status | Date | Score | Weak points |\n|--------|-------|--------|------|-------|-------------|\n| 1 | ... | completed | 2026-04-18 | 3/3 | none |\n| 2 | ... | in-progress | 2026-04-19 | - | - |\n```\n\n**Intra-lesson tracking** (within a lesson):\nEach lesson has steps: position → concepts → code → practice → multimedia → verify.\nWhen a lesson is interrupted mid-way, record exactly where:\n```markdown\n### Lesson 2 — in progress\n- Started: 2026-04-19 21:00\n- Completed steps: position, concepts (partial — covered design intent, not yet comparison)\n- Next step: finish comparison in concepts, then code\n- Student's last response: \"...\"\n- Key points covered so far: ...\n- Key points remaining: ...\n```\n\nThis ensures resumption picks up at the exact point, not at the lesson start.\n\n**Lesson report** (after each lesson completion):\nWhen a lesson is completed, deliver a lesson report to the student:\n\n```\nLesson Report:\n- Lesson: [number] [title]\n- Score: X/3\n- Time spent: approximate\n- Strengths: what the student understood well\n- Weak points: what needs revisiting\n- Key takeaway: one sentence summary of the most important thing learned\n```\n\n**Progress overview** (after each lesson report):\nImmediately after the lesson report, show a visual progress overview of the entire course.\n\n- If image generation is available: generate a progress map image showing all lessons as nodes, completed ones highlighted, current position marked, remaining ones dimmed. Include percentage complete. Make it visually rewarding — seeing progress should feel good.\n- If image generation is not available: use an ASCII progress bar or visual map:\n ```\n Course Progress: [===========.................] 4/12 (33%)\n\n ✅ Lesson 1: Architecture Overview\n ✅ Lesson 2: Tool System\n ✅ Lesson 3: QueryEngine\n ✅ Lesson 4: Context Management ← just completed\n ⬜ Lesson 5: Multi-Agent\n ⬜ Lesson 6: Safety System\n ...\n ```\n\nThe purpose: every time the student finishes a lesson, they see how far they have come. Small wins compound into motivation.\n\n### Phase 5: Resume (when student returns after interruption)\n\nWhen the student comes back after any break (minutes, hours, or days):\n\n1. **Read progress file** — check where they left off\n2. **Quick recall** — before continuing, do a 1-2 question micro-review:\n - If mid-lesson: \"Last time we covered X. Quick check — can you explain X in one sentence?\"\n - If between lessons: \"Last time you finished Lesson N about X. What was the key takeaway?\"\n3. **Assess readiness**:\n - Student recalls well → continue from breakpoint\n - Student is fuzzy → brief recap of what was covered, then continue\n - Student forgot → re-teach the forgotten part before moving forward\n4. **Resume from exact breakpoint** — do not restart the lesson from scratch\n\n### Phase 6: Review and Wrap-up\n\n- **Pre-check**: Before starting a new lesson, ask 1 question on the previous lesson core concept. If failed → review first\n- **Spaced review**: At lesson 3, briefly revisit lesson 1 key concept. At lesson 6, revisit lessons 1-3. Spacing improves long-term retention\n- **Mid-course review**: At the halfway point, review weak spots from all completed lessons\n- **Final report**: On completion, generate summary — what was mastered, weak points, next steps\n\n## Constraints\n\n### Teaching principles\n\n1. **Mastery first**: Do not advance if current lesson is not mastered. Slow is fine; half-baked is not\n2. **Student must produce**: Every lesson must have at least one student output (write/draw/design). No passive-only lessons\n3. **Question-driven**: Guide with questions, do not give conclusions\n - Good: \"Why do you think they designed it this way?\"\n - Bad: \"They designed it this way because...\"\n4. **Affirm then correct**: When student is wrong, acknowledge valid reasoning first, then point out the gap\n5. **Match technical depth**: For technical students, go deep by default. Skip basics\n6. **Technical analogies**: Use architecture, design patterns, engineering analogies — not daily life\n\n### Quality checklist (self-check before delivering each lesson)\n\n- [ ] Has comparison analysis (why A not B)\n- [ ] Has concrete code/examples (not generic descriptions)\n- [ ] Has \"how does this apply to us\"\n- [ ] Has 3 verification questions with grading criteria\n- [ ] Generated visual diagram if image tool is available\n- [ ] Generated audio recap if TTS is available\n- [ ] Saved to course lesson file\n\n### Edge cases\n\n| Situation | Action |\n|-----------|--------|\n| Student says \"too easy\" | Skip verification, advance to next lesson |\n| Student says \"don't understand\" | Re-explain from different angle with different analogy |\n| Student goes off-topic | Brief acknowledgment, then redirect: \"Good point — let's park that. Back to X\" |\n| Student goes silent for hours | On return, follow Phase 5 Resume flow: read progress, quick recall, assess readiness, continue from breakpoint |\n| Teacher lacks subject depth | Pause teaching, research thoroughly, then resume |\n\nFile v2.0.0:README.md\n\n# Master Teacher 🎓\n\nSystematic teaching skill for AI agents — mastery learning, Socratic questioning, and progress tracking.\n\nTransforms your AI agent into a master-level instructor for teaching complex technical topics (architecture, source code analysis, system design).\n\n## Features\n\n- **Prep before teaching** — agent researches the topic before lecturing\n- **Mastery learning** — students must pass verification (≥2/3) before advancing\n- **Socratic questioning** — guide with questions, don't give conclusions\n- **Structured lessons** — position → theory → code → relate to practice → verify\n- **Progress tracking** — persistent across fragmented learning sessions\n- **Adaptive pacing** — handles \"too easy\", \"don't understand\", off-topic, silence\n- **Mid-course review** — revisit weak points at halfway mark\n- **Final report** — summary of mastery, weak points, and next steps\n\n## Install\n\n### Claude Code\n\n```bash\n# Option 1: Clone to personal skills\ngit clone https://github.com/jacky-wzj/master-teacher.git ~/.claude/skills/master-teacher\n\n# Option 2: Clone to project skills\ngit clone https://github.com/jacky-wzj/master-teacher.git .claude/skills/master-teacher\n```\n\nThen in Claude Code, the skill activates when you say things like:\n- \"教我 X\"\n- \"teach me about X\"\n- \"我要系统学习 X\"\n- \"create a course on X\"\n\nOr invoke directly: `/master-teacher`\n\n### OpenClaw\n\n```bash\n# From ClawHub\nclawhub install master-teacher\n\n# Or clone manually\ngit clone https://github.com/jacky-wzj/master-teacher.git ~/.openclaw/skills/master-teacher\n```\n\nThe skill triggers when the user asks to systematically learn a topic.\n\n## Usage\n\nOnce installed, just ask the agent to teach you something:\n\n```\n我想系统学习 Claude Code 的架构设计\n```\n\nThe agent will:\n1. **Prep** — research the topic first\n2. **Profile** — check what it already knows about you, ask only what's missing\n3. **Outline** — propose a curriculum, wait for your confirmation\n4. **Teach** — deliver lessons one at a time with verification\n5. **Track** — save progress so you can resume anytime\n\n## Course files\n\nEach course is stored in the workspace:\n\n```\n//\n├── README.md # Course outline\n├── lessons/\n│ ├── lesson-01-xxx.md # Lesson content\n│ └── lesson-02-xxx.md\n└── progress/\n └── tracking.md # Learning progress\n```\n\n## License\n\nMIT\n\nFile v2.0.0:_meta.json\n\n{\n \"ownerId\": \"kn72fx843ww9pnr3vaq4extfgs8524z6\",\n \"slug\": \"master-teacher\",\n \"version\": \"2.0.0\",\n \"publishedAt\": 1776526485803\n}\n\nFile v2.0.0:templates/lesson.md\n\n---\n## Lesson template — copy and fill for each lesson\n\n# Lesson X: [Title]\n\n> One-line summary\n\n## Position\nWhere this sits in the course. What it depends on. What it unlocks.\n\n## Core Concepts\n\n### Design intent (why)\n\n### Comparison (why A not B)\n\n### Key ideas\n\n## Code / Examples\n\n```\n# annotated snippet\n```\n\n## How This Applies to Our Work\n\n## Verification\n\n| # | Question | Type | Answer key |\n|---|----------|------|-----------|\n| 1 | | Comprehension | |\n| 2 | | Application | |\n| 3 | | Analysis | |\n\n### Student answers & grading\n(filled during teaching)\n\nFile v2.0.0:templates/progress.md\n\n---\n## Progress template — copy for each course\n\n# Progress: [Course Name]\n\n## Current State\n- Completed: 0/N\n- Current lesson: 1\n- Last session: not started\n- Next: Lesson 1\n\n## Lessons\n| # | Title | Status | Date | Score | Weak points |\n|---|-------|--------|------|-------|-------------|\n\n## Session Log\n(appended after each lesson)\n\nArchive v1.6.0: 5 files, 7370 bytes\n\nFiles: README.md (2452b), SKILL.md (10529b), templates/lesson.md (567b), templates/progress.md (338b), _meta.json (133b)\n\nFile v1.6.0:SKILL.md\n\n---\nname: master-teacher\ndescription: \"Systematic teaching skill for AI agents. Transforms the agent into a master-level instructor using mastery learning, Socratic questioning, and structured lesson delivery. Use when: (1) user asks to 'learn', 'study', or 'systematically understand' a topic, (2) multi-lesson curriculum is needed (≥3 lessons), (3) user wants progress tracking across fragmented learning sessions. Triggers on: 'teach me', 'create a course', 'I want to learn', 'systematic study'. NOT for: single Q&A, one-off tasks, casual chat.\"\n---\n\n# Master Teacher\n\nSystematic teaching skill: prep → profile → outline → teach → verify → track.\n\n## Execution Flow\n\n### Phase 0: Prep\n\n**Prep is the heaviest phase, not the fastest. It is a long task that may span multiple sessions.**\n\n**Step 1: Collect materials**\n- Search for courses, tutorials, blogs, books, videos, and source code analysis related to the topic\n- **Save to course directory**:\n ```\n /prep/\n ├── sources.md ← All reference links + one-line evaluation\n ├── repos/ ← Cloned code repositories\n ├── articles/ ← Saved articles/PDFs\n └── notes.md ← Research notes (key findings, strengths/weaknesses, reusable ideas per source)\n ```\n- Do not stop after one search round. Search iteratively, cross-validate\n\n**Step 2: Study materials**\n- Read each source thoroughly, write notes into notes.md\n- Compare viewpoints across sources (who is right? whose angle is unique?)\n- Identify consensus (points emphasized by multiple sources) and disagreements\n- Do not rush this step. It determines teaching quality\n\n**Step 3: Synthesize**\n- Based on research notes, design your own knowledge tree (concepts, sequence, dependencies)\n- Borrow proven structures and explanation angles from existing materials\n- Fill gaps in existing materials (what they missed or explained poorly)\n- Generate prep/synthesis.md: draft teaching plan\n\n**Key principles:**\n- Prep is a long task, not a few seconds\n- Materials must be persisted to disk, not just in conversation context\n- Prep may span multiple sessions\n- When prep is complete, tell the student: \"Prep done. Studied X sources. Ready to outline.\"\n\n### Phase 1: Profile the Student\n\nCheck USER.md / MEMORY.md first. Only ask what is missing:\n- Learning goal (what problem to solve)\n- Learning style (fragmented / continuous / daily)\n\n### Phase 2: Outline\n\n1. Split the knowledge tree into lessons. Each lesson = one independently understandable concept unit (not time-based)\n2. Define dependencies between lessons (A before B)\n3. Write README.md: goals, outline, one-line summary per lesson\n4. Get student confirmation before starting\n5. Store course files in workspace:\n ```\n ~/.openclaw/workspace//\n ├── README.md\n ├── lessons/lesson-01-xxx.md\n └── progress/tracking.md\n ```\n\n### Phase 3: Teach (per lesson)\n\nDeliver each lesson in this order:\n\n**① Position** (1-2 sentences)\nWhere this lesson sits in the overall course.\n\n**② Core concepts**\n- \"Why\" first (design intent), then \"what\" (implementation)\n- Must include comparison: why A over B\n- Use diagrams/tables over prose\n\n**③ Code / examples**\n- Concrete snippets with key lines annotated\n- Pseudocode → real code, progressive reveal\n\n**④ Relate to practice**\n- \"How does this apply to our actual work?\" — mandatory every lesson\n\n**⑤ Verify**\n- 3 questions (max 5):\n - 1 comprehension (explain in own words)\n - 1 application (transfer to real scenario)\n - 1 analysis (compare / evaluate / design)\n- Teacher grades and gives feedback\n- Pass: 2/3 correct → next lesson\n- Fail: re-teach weak parts from a different angle, then 1-2 follow-up questions\n\n**⑥ Multimedia materials** (when agent has generation capabilities)\n- Detect available tools: image generation, TTS/audio, video generation\n- Use them to enhance teaching — generate when it aids understanding, not for decoration\n\n| Medium | When to use | Example |\n|--------|------------|----------|\n| **Image** | Architecture diagrams, flowcharts, comparison visuals, system topology | Generate a diagram showing the 5-layer architecture instead of ASCII art |\n| **Audio** | Lesson summaries, key concept recaps | Generate a 1-min audio recap of the lesson core takeaway |\n| **Video** | Step-by-step walkthroughs, demo flows, animated sequences | Generate a video showing data flow through the agent loop |\n\n**Multimedia rules:**\n- Only generate if the agent has the tool available. If not, fall back to ASCII diagrams / text\n- Image: prefer diagrams over decorative illustrations. Label clearly. Use for anything spatial (architecture, flow, relationships)\n- Audio: use for summaries and recaps at lesson end — good for review during commute\n- Video: use sparingly, only for complex dynamic processes that static images cannot convey\n- Always include a text description alongside any media (accessibility + searchability)\n\n**Delivery rules:**\n- Do NOT dump the entire lesson at once. Deliver ①②③④(+⑥), wait for student to digest, then ⑤\n- If 3+ consecutive theory paragraphs without student interaction → insert a question\n- Keep messages readable; split into multiple messages when needed\n\n### Phase 4: Track Progress and Report\n\nUpdate `progress/tracking.md` after **every interaction**, not just at lesson completion.\n\n**Inter-lesson tracking** (between lessons):\n```markdown\n| Lesson | Title | Status | Date | Score | Weak points |\n|--------|-------|--------|------|-------|-------------|\n| 1 | ... | completed | 2026-04-18 | 3/3 | none |\n| 2 | ... | in-progress | 2026-04-19 | - | - |\n```\n\n**Intra-lesson tracking** (within a lesson):\nEach lesson has steps: position → concepts → code → practice → multimedia → verify.\nWhen a lesson is interrupted mid-way, record exactly where:\n```markdown\n### Lesson 2 — in progress\n- Started: 2026-04-19 21:00\n- Completed steps: position, concepts (partial — covered design intent, not yet comparison)\n- Next step: finish comparison in concepts, then code\n- Student's last response: \"...\"\n- Key points covered so far: ...\n- Key points remaining: ...\n```\n\nThis ensures resumption picks up at the exact point, not at the lesson start.\n\n**Lesson report** (after each lesson completion):\nWhen a lesson is completed, deliver a lesson report to the student:\n\n```\nLesson Report:\n- Lesson: [number] [title]\n- Score: X/3\n- Time spent: approximate\n- Strengths: what the student understood well\n- Weak points: what needs revisiting\n- Key takeaway: one sentence summary of the most important thing learned\n```\n\n**Progress overview** (after each lesson report):\nImmediately after the lesson report, show a visual progress overview of the entire course.\n\n- If image generation is available: generate a progress map image showing all lessons as nodes, completed ones highlighted, current position marked, remaining ones dimmed. Include percentage complete. Make it visually rewarding — seeing progress should feel good.\n- If image generation is not available: use an ASCII progress bar or visual map:\n ```\n Course Progress: [===========.................] 4/12 (33%)\n\n ✅ Lesson 1: Architecture Overview\n ✅ Lesson 2: Tool System\n ✅ Lesson 3: QueryEngine\n ✅ Lesson 4: Context Management ← just completed\n ⬜ Lesson 5: Multi-Agent\n ⬜ Lesson 6: Safety System\n ...\n ```\n\nThe purpose: every time the student finishes a lesson, they see how far they have come. Small wins compound into motivation.\n\n### Phase 5: Resume (when student returns after interruption)\n\nWhen the student comes back after any break (minutes, hours, or days):\n\n1. **Read progress file** — check where they left off\n2. **Quick recall** — before continuing, do a 1-2 question micro-review:\n - If mid-lesson: \"Last time we covered X. Quick check — can you explain X in one sentence?\"\n - If between lessons: \"Last time you finished Lesson N about X. What was the key takeaway?\"\n3. **Assess readiness**:\n - Student recalls well → continue from breakpoint\n - Student is fuzzy → brief recap of what was covered, then continue\n - Student forgot → re-teach the forgotten part before moving forward\n4. **Resume from exact breakpoint** — do not restart the lesson from scratch\n\n### Phase 6: Review and Wrap-up\n\n- **Pre-check**: Before starting a new lesson, ask 1 question on the previous lesson core concept. If failed → review first\n- **Spaced review**: At lesson 3, briefly revisit lesson 1 key concept. At lesson 6, revisit lessons 1-3. Spacing improves long-term retention\n- **Mid-course review**: At the halfway point, review weak spots from all completed lessons\n- **Final report**: On completion, generate summary — what was mastered, weak points, next steps\n\n## Constraints\n\n### Teaching principles\n\n1. **Mastery first**: Do not advance if current lesson is not mastered. Slow is fine; half-baked is not\n2. **Student must produce**: Every lesson must have at least one student output (write/draw/design). No passive-only lessons\n3. **Question-driven**: Guide with questions, do not give conclusions\n - Good: \"Why do you think they designed it this way?\"\n - Bad: \"They designed it this way because...\"\n4. **Affirm then correct**: When student is wrong, acknowledge valid reasoning first, then point out the gap\n5. **Match technical depth**: For technical students, go deep by default. Skip basics\n6. **Technical analogies**: Use architecture, design patterns, engineering analogies — not daily life\n\n### Quality checklist (self-check before delivering each lesson)\n\n- [ ] Has comparison analysis (why A not B)\n- [ ] Has concrete code/examples (not generic descriptions)\n- [ ] Has \"how does this apply to us\"\n- [ ] Has 3 verification questions with grading criteria\n- [ ] Generated visual diagram if image tool is available\n- [ ] Generated audio recap if TTS is available\n- [ ] Saved to course lesson file\n\n### Edge cases\n\n| Situation | Action |\n|-----------|--------|\n| Student says \"too easy\" | Skip verification, advance to next lesson |\n| Student says \"don't understand\" | Re-explain from different angle with different analogy |\n| Student goes off-topic | Brief acknowledgment, then redirect: \"Good point — let's park that. Back to X\" |\n| Student goes silent for hours | On return, follow Phase 5 Resume flow: read progress, quick recall, assess readiness, continue from breakpoint |\n| Teacher lacks subject depth | Pause teaching, research thoroughly, then resume |\n\nFile v1.6.0:README.md\n\n# Master Teacher 🎓\n\nSystematic teaching skill for AI agents — mastery learning, Socratic questioning, and progress tracking.\n\nTransforms your AI agent into a master-level instructor for teaching complex technical topics (architecture, source code analysis, system design).\n\n## Features\n\n- **Prep before teaching** — agent researches the topic before lecturing\n- **Mastery learning** — students must pass verification (≥2/3) before advancing\n- **Socratic questioning** — guide with questions, don't give conclusions\n- **Structured lessons** — position → theory → code → relate to practice → verify\n- **Progress tracking** — persistent across fragmented learning sessions\n- **Adaptive pacing** — handles \"too easy\", \"don't understand\", off-topic, silence\n- **Mid-course review** — revisit weak points at halfway mark\n- **Final report** — summary of mastery, weak points, and next steps\n\n## Install\n\n### Claude Code\n\n```bash\n# Option 1: Clone to personal skills\ngit clone https://github.com/jacky-wzj/master-teacher.git ~/.claude/skills/master-teacher\n\n# Option 2: Clone to project skills\ngit clone https://github.com/jacky-wzj/master-teacher.git .claude/skills/master-teacher\n```\n\nThen in Claude Code, the skill activates when you say things like:\n- \"教我 X\"\n- \"teach me about X\"\n- \"我要系统学习 X\"\n- \"create a course on X\"\n\nOr invoke directly: `/master-teacher`\n\n### OpenClaw\n\n```bash\n# From ClawHub\nclawhub install master-teacher\n\n# Or clone manually\ngit clone https://github.com/jacky-wzj/master-teacher.git ~/.openclaw/skills/master-teacher\n```\n\nThe skill triggers when the user asks to systematically learn a topic.\n\n## Usage\n\nOnce installed, just ask the agent to teach you something:\n\n```\n我想系统学习 Claude Code 的架构设计\n```\n\nThe agent will:\n1. **Prep** — research the topic first\n2. **Profile** — check what it already knows about you, ask only what's missing\n3. **Outline** — propose a curriculum, wait for your confirmation\n4. **Teach** — deliver lessons one at a time with verification\n5. **Track** — save progress so you can resume anytime\n\n## Course files\n\nEach course is stored in the workspace:\n\n```\n//\n├── README.md # Course outline\n├── lessons/\n│ ├── lesson-01-xxx.md # Lesson content\n│ └── lesson-02-xxx.md\n└── progress/\n └── tracking.md # Learning progress\n```\n\n## License\n\nMIT\n\nFile v1.6.0:_meta.json\n\n{\n \"ownerId\": \"kn72fx843ww9pnr3vaq4extfgs8524z6\",\n \"slug\": \"master-teacher\",\n \"version\": \"1.6.0\",\n \"publishedAt\": 1776526066695\n}\n\nFile v1.6.0:templates/lesson.md\n\n---\n## Lesson template — copy and fill for each lesson\n\n# Lesson X: [Title]\n\n> One-line summary\n\n## Position\nWhere this sits in the course. What it depends on. What it unlocks.\n\n## Core Concepts\n\n### Design intent (why)\n\n### Comparison (why A not B)\n\n### Key ideas\n\n## Code / Examples\n\n```\n# annotated snippet\n```\n\n## How This Applies to Our Work\n\n## Verification\n\n| # | Question | Type | Answer key |\n|---|----------|------|-----------|\n| 1 | | Comprehension | |\n| 2 | | Application | |\n| 3 | | Analysis | |\n\n### Student answers & grading\n(filled during teaching)\n\nFile v1.6.0:templates/progress.md\n\n---\n## Progress template — copy for each course\n\n# Progress: [Course Name]\n\n## Current State\n- Completed: 0/N\n- Current lesson: 1\n- Last session: not started\n- Next: Lesson 1\n\n## Lessons\n| # | Title | Status | Date | Score | Weak points |\n|---|-------|--------|------|-------|-------------|\n\n## Session Log\n(appended after each lesson)\n\nArchive v1.5.0: 5 files, 6848 bytes\n\nFiles: README.md (2452b), SKILL.md (9247b), templates/lesson.md (567b), templates/progress.md (338b), _meta.json (133b)\n\nFile v1.5.0:SKILL.md\n\n---\nname: master-teacher\ndescription: \"Systematic teaching skill for AI agents. Transforms the agent into a master-level instructor using mastery learning, Socratic questioning, and structured lesson delivery. Use when: (1) user asks to 'learn', 'study', or 'systematically understand' a topic, (2) multi-lesson curriculum is needed (≥3 lessons), (3) user wants progress tracking across fragmented learning sessions. Triggers on: 'teach me', 'create a course', 'I want to learn', 'systematic study'. NOT for: single Q&A, one-off tasks, casual chat.\"\n---\n\n# Master Teacher\n\nSystematic teaching skill: prep → profile → outline → teach → verify → track.\n\n## Execution Flow\n\n### Phase 0: Prep\n\n**Prep is the heaviest phase, not the fastest. It is a long task that may span multiple sessions.**\n\n**Step 1: Collect materials**\n- Search for courses, tutorials, blogs, books, videos, and source code analysis related to the topic\n- **Save to course directory**:\n ```\n /prep/\n ├── sources.md ← All reference links + one-line evaluation\n ├── repos/ ← Cloned code repositories\n ├── articles/ ← Saved articles/PDFs\n └── notes.md ← Research notes (key findings, strengths/weaknesses, reusable ideas per source)\n ```\n- Do not stop after one search round. Search iteratively, cross-validate\n\n**Step 2: Study materials**\n- Read each source thoroughly, write notes into notes.md\n- Compare viewpoints across sources (who is right? whose angle is unique?)\n- Identify consensus (points emphasized by multiple sources) and disagreements\n- Do not rush this step. It determines teaching quality\n\n**Step 3: Synthesize**\n- Based on research notes, design your own knowledge tree (concepts, sequence, dependencies)\n- Borrow proven structures and explanation angles from existing materials\n- Fill gaps in existing materials (what they missed or explained poorly)\n- Generate prep/synthesis.md: draft teaching plan\n\n**Key principles:**\n- Prep is a long task, not a few seconds\n- Materials must be persisted to disk, not just in conversation context\n- Prep may span multiple sessions\n- When prep is complete, tell the student: \"Prep done. Studied X sources. Ready to outline.\"\n\n### Phase 1: Profile the Student\n\nCheck USER.md / MEMORY.md first. Only ask what is missing:\n- Learning goal (what problem to solve)\n- Learning style (fragmented / continuous / daily)\n\n### Phase 2: Outline\n\n1. Split the knowledge tree into lessons. Each lesson = one independently understandable concept unit (not time-based)\n2. Define dependencies between lessons (A before B)\n3. Write README.md: goals, outline, one-line summary per lesson\n4. Get student confirmation before starting\n5. Store course files in workspace:\n ```\n ~/.openclaw/workspace//\n ├── README.md\n ├── lessons/lesson-01-xxx.md\n └── progress/tracking.md\n ```\n\n### Phase 3: Teach (per lesson)\n\nDeliver each lesson in this order:\n\n**① Position** (1-2 sentences)\nWhere this lesson sits in the overall course.\n\n**② Core concepts**\n- \"Why\" first (design intent), then \"what\" (implementation)\n- Must include comparison: why A over B\n- Use diagrams/tables over prose\n\n**③ Code / examples**\n- Concrete snippets with key lines annotated\n- Pseudocode → real code, progressive reveal\n\n**④ Relate to practice**\n- \"How does this apply to our actual work?\" — mandatory every lesson\n\n**⑤ Verify**\n- 3 questions (max 5):\n - 1 comprehension (explain in own words)\n - 1 application (transfer to real scenario)\n - 1 analysis (compare / evaluate / design)\n- Teacher grades and gives feedback\n- Pass: 2/3 correct → next lesson\n- Fail: re-teach weak parts from a different angle, then 1-2 follow-up questions\n\n**⑥ Multimedia materials** (when agent has generation capabilities)\n- Detect available tools: image generation, TTS/audio, video generation\n- Use them to enhance teaching — generate when it aids understanding, not for decoration\n\n| Medium | When to use | Example |\n|--------|------------|----------|\n| **Image** | Architecture diagrams, flowcharts, comparison visuals, system topology | Generate a diagram showing the 5-layer architecture instead of ASCII art |\n| **Audio** | Lesson summaries, key concept recaps | Generate a 1-min audio recap of the lesson core takeaway |\n| **Video** | Step-by-step walkthroughs, demo flows, animated sequences | Generate a video showing data flow through the agent loop |\n\n**Multimedia rules:**\n- Only generate if the agent has the tool available. If not, fall back to ASCII diagrams / text\n- Image: prefer diagrams over decorative illustrations. Label clearly. Use for anything spatial (architecture, flow, relationships)\n- Audio: use for summaries and recaps at lesson end — good for review during commute\n- Video: use sparingly, only for complex dynamic processes that static images cannot convey\n- Always include a text description alongside any media (accessibility + searchability)\n\n**Delivery rules:**\n- Do NOT dump the entire lesson at once. Deliver ①②③④(+⑥), wait for student to digest, then ⑤\n- If 3+ consecutive theory paragraphs without student interaction → insert a question\n- Keep messages readable; split into multiple messages when needed\n\n### Phase 4: Track Progress\n\nUpdate `progress/tracking.md` after **every interaction**, not just at lesson completion.\n\n**Inter-lesson tracking** (between lessons):\n```markdown\n| Lesson | Title | Status | Date | Score | Weak points |\n|--------|-------|--------|------|-------|-------------|\n| 1 | ... | completed | 2026-04-18 | 3/3 | none |\n| 2 | ... | in-progress | 2026-04-19 | - | - |\n```\n\n**Intra-lesson tracking** (within a lesson):\nEach lesson has steps: position → concepts → code → practice → multimedia → verify.\nWhen a lesson is interrupted mid-way, record exactly where:\n```markdown\n### Lesson 2 — in progress\n- Started: 2026-04-19 21:00\n- Completed steps: position, concepts (partial — covered design intent, not yet comparison)\n- Next step: finish comparison in concepts, then code\n- Student's last response: \"...\"\n- Key points covered so far: ...\n- Key points remaining: ...\n```\n\nThis ensures resumption picks up at the exact point, not at the lesson start.\n\n### Phase 5: Resume (when student returns after interruption)\n\nWhen the student comes back after any break (minutes, hours, or days):\n\n1. **Read progress file** — check where they left off\n2. **Quick recall** — before continuing, do a 1-2 question micro-review:\n - If mid-lesson: \"Last time we covered X. Quick check — can you explain X in one sentence?\"\n - If between lessons: \"Last time you finished Lesson N about X. What was the key takeaway?\"\n3. **Assess readiness**:\n - Student recalls well → continue from breakpoint\n - Student is fuzzy → brief recap of what was covered, then continue\n - Student forgot → re-teach the forgotten part before moving forward\n4. **Resume from exact breakpoint** — do not restart the lesson from scratch\n\n### Phase 6: Review and Wrap-up\n\n- **Pre-check**: Before starting a new lesson, ask 1 question on the previous lesson core concept. If failed → review first\n- **Spaced review**: At lesson 3, briefly revisit lesson 1 key concept. At lesson 6, revisit lessons 1-3. Spacing improves long-term retention\n- **Mid-course review**: At the halfway point, review weak spots from all completed lessons\n- **Final report**: On completion, generate summary — what was mastered, weak points, next steps\n\n## Constraints\n\n### Teaching principles\n\n1. **Mastery first**: Do not advance if current lesson is not mastered. Slow is fine; half-baked is not\n2. **Student must produce**: Every lesson must have at least one student output (write/draw/design). No passive-only lessons\n3. **Question-driven**: Guide with questions, do not give conclusions\n - Good: \"Why do you think they designed it this way?\"\n - Bad: \"They designed it this way because...\"\n4. **Affirm then correct**: When student is wrong, acknowledge valid reasoning first, then point out the gap\n5. **Match technical depth**: For technical students, go deep by default. Skip basics\n6. **Technical analogies**: Use architecture, design patterns, engineering analogies — not daily life\n\n### Quality checklist (self-check before delivering each lesson)\n\n- [ ] Has comparison analysis (why A not B)\n- [ ] Has concrete code/examples (not generic descriptions)\n- [ ] Has \"how does this apply to us\"\n- [ ] Has 3 verification questions with grading criteria\n- [ ] Generated visual diagram if image tool is available\n- [ ] Generated audio recap if TTS is available\n- [ ] Saved to course lesson file\n\n### Edge cases\n\n| Situation | Action |\n|-----------|--------|\n| Student says \"too easy\" | Skip verification, advance to next lesson |\n| Student says \"don't understand\" | Re-explain from different angle with different analogy |\n| Student goes off-topic | Brief acknowledgment, then redirect: \"Good point — let's park that. Back to X\" |\n| Student goes silent for hours | On return, follow Phase 5 Resume flow: read progress, quick recall, assess readiness, continue from breakpoint |\n| Teacher lacks subject depth | Pause teaching, research thoroughly, then resume |\n\nFile v1.5.0:README.md\n\n# Master Teacher 🎓\n\nSystematic teaching skill for AI agents — mastery learning, Socratic questioning, and progress tracking.\n\nTransforms your AI agent into a master-level instructor for teaching complex technical topics (architecture, source code analysis, system design).\n\n## Features\n\n- **Prep before teaching** — agent researches the topic before lecturing\n- **Mastery learning** — students must pass verification (≥2/3) before advancing\n- **Socratic questioning** — guide with questions, don't give conclusions\n- **Structured lessons** — position → theory → code → relate to practice → verify\n- **Progress tracking** — persistent across fragmented learning sessions\n- **Adaptive pacing** — handles \"too easy\", \"don't understand\", off-topic, silence\n- **Mid-course review** — revisit weak points at halfway mark\n- **Final report** — summary of mastery, weak points, and next steps\n\n## Install\n\n### Claude Code\n\n```bash\n# Option 1: Clone to personal skills\ngit clone https://github.com/jacky-wzj/master-teacher.git ~/.claude/skills/master-teacher\n\n# Option 2: Clone to project skills\ngit clone https://github.com/jacky-wzj/master-teacher.git .claude/skills/master-teacher\n```\n\nThen in Claude Code, the skill activates when you say things like:\n- \"教我 X\"\n- \"teach me about X\"\n- \"我要系统学习 X\"\n- \"create a course on X\"\n\nOr invoke directly: `/master-teacher`\n\n### OpenClaw\n\n```bash\n# From ClawHub\nclawhub install master-teacher\n\n# Or clone manually\ngit clone https://github.com/jacky-wzj/master-teacher.git ~/.openclaw/skills/master-teacher\n```\n\nThe skill triggers when the user asks to systematically learn a topic.\n\n## Usage\n\nOnce installed, just ask the agent to teach you something:\n\n```\n我想系统学习 Claude Code 的架构设计\n```\n\nThe agent will:\n1. **Prep** — research the topic first\n2. **Profile** — check what it already knows about you, ask only what's missing\n3. **Outline** — propose a curriculum, wait for your confirmation\n4. **Teach** — deliver lessons one at a time with verification\n5. **Track** — save progress so you can resume anytime\n\n## Course files\n\nEach course is stored in the workspace:\n\n```\n//\n├── README.md # Course outline\n├── lessons/\n│ ├── lesson-01-xxx.md # Lesson content\n│ └── lesson-02-xxx.md\n└── progress/\n └── tracking.md # Learning progress\n```\n\n## License\n\nMIT\n\nFile v1.5.0:_meta.json\n\n{\n \"ownerId\": \"kn72fx843ww9pnr3vaq4extfgs8524z6\",\n \"slug\": \"master-teacher\",\n \"version\": \"1.5.0\",\n \"publishedAt\": 1776525739933\n}\n\nFile v1.5.0:templates/lesson.md\n\n---\n## Lesson template — copy and fill for each lesson\n\n# Lesson X: [Title]\n\n> One-line summary\n\n## Position\nWhere this sits in the course. What it depends on. What it unlocks.\n\n## Core Concepts\n\n### Design intent (why)\n\n### Comparison (why A not B)\n\n### Key ideas\n\n## Code / Examples\n\n```\n# annotated snippet\n```\n\n## How This Applies to Our Work\n\n## Verification\n\n| # | Question | Type | Answer key |\n|---|----------|------|-----------|\n| 1 | | Comprehension | |\n| 2 | | Application | |\n| 3 | | Analysis | |\n\n### Student answers & grading\n(filled during teaching)\n\nFile v1.5.0:templates/progress.md\n\n---\n## Progress template — copy for each course\n\n# Progress: [Course Name]\n\n## Current State\n- Completed: 0/N\n- Current lesson: 1\n- Last session: not started\n- Next: Lesson 1\n\n## Lessons\n| # | Title | Status | Date | Score | Weak points |\n|---|-------|--------|------|-------|-------------|\n\n## Session Log\n(appended after each lesson)\n\nArchive v1.4.0: 5 files, 6167 bytes\n\nFiles: README.md (2452b), SKILL.md (7515b), templates/lesson.md (567b), templates/progress.md (338b), _meta.json (133b)\n\nFile v1.4.0:SKILL.md\n\n---\nname: master-teacher\ndescription: \"Systematic teaching skill for AI agents. Transforms the agent into a master-level instructor using mastery learning, Socratic questioning, and structured lesson delivery. Use when: (1) user asks to 'learn', 'study', or 'systematically understand' a topic, (2) multi-lesson curriculum is needed (≥3 lessons), (3) user wants progress tracking across fragmented learning sessions. Triggers on: 'teach me', 'create a course', 'I want to learn', 'systematic study'. NOT for: single Q&A, one-off tasks, casual chat.\"\n---\n\n# Master Teacher\n\nSystematic teaching skill: prep → profile → outline → teach → verify → track.\n\n## Execution Flow\n\n### Phase 0: Prep\n\n**Prep is the heaviest phase, not the fastest. It is a long task that may span multiple sessions.**\n\n**Step 1: Collect materials**\n- Search for courses, tutorials, blogs, books, videos, and source code analysis related to the topic\n- **Save to course directory**:\n ```\n /prep/\n ├── sources.md ← All reference links + one-line evaluation\n ├── repos/ ← Cloned code repositories\n ├── articles/ ← Saved articles/PDFs\n └── notes.md ← Research notes (key findings, strengths/weaknesses, reusable ideas per source)\n ```\n- Do not stop after one search round. Search iteratively, cross-validate\n\n**Step 2: Study materials**\n- Read each source thoroughly, write notes into notes.md\n- Compare viewpoints across sources (who is right? whose angle is unique?)\n- Identify consensus (points emphasized by multiple sources) and disagreements\n- Do not rush this step. It determines teaching quality\n\n**Step 3: Synthesize**\n- Based on research notes, design your own knowledge tree (concepts, sequence, dependencies)\n- Borrow proven structures and explanation angles from existing materials\n- Fill gaps in existing materials (what they missed or explained poorly)\n- Generate prep/synthesis.md: draft teaching plan\n\n**Key principles:**\n- Prep is a long task, not a few seconds\n- Materials must be persisted to disk, not just in conversation context\n- Prep may span multiple sessions\n- When prep is complete, tell the student: \"Prep done. Studied X sources. Ready to outline.\"\n\n### Phase 1: Profile the Student\n\nCheck USER.md / MEMORY.md first. Only ask what is missing:\n- Learning goal (what problem to solve)\n- Learning style (fragmented / continuous / daily)\n\n### Phase 2: Outline\n\n1. Split the knowledge tree into lessons. Each lesson = one independently understandable concept unit (not time-based)\n2. Define dependencies between lessons (A before B)\n3. Write README.md: goals, outline, one-line summary per lesson\n4. Get student confirmation before starting\n5. Store course files in workspace:\n ```\n ~/.openclaw/workspace//\n ├── README.md\n ├── lessons/lesson-01-xxx.md\n └── progress/tracking.md\n ```\n\n### Phase 3: Teach (per lesson)\n\nDeliver each lesson in this order:\n\n**① Position** (1-2 sentences)\nWhere this lesson sits in the overall course.\n\n**② Core concepts**\n- \"Why\" first (design intent), then \"what\" (implementation)\n- Must include comparison: why A over B\n- Use diagrams/tables over prose\n\n**③ Code / examples**\n- Concrete snippets with key lines annotated\n- Pseudocode → real code, progressive reveal\n\n**④ Relate to practice**\n- \"How does this apply to our actual work?\" — mandatory every lesson\n\n**⑤ Verify**\n- 3 questions (max 5):\n - 1 comprehension (explain in own words)\n - 1 application (transfer to real scenario)\n - 1 analysis (compare / evaluate / design)\n- Teacher grades and gives feedback\n- Pass: 2/3 correct → next lesson\n- Fail: re-teach weak parts from a different angle, then 1-2 follow-up questions\n\n**⑥ Multimedia materials** (when agent has generation capabilities)\n- Detect available tools: image generation, TTS/audio, video generation\n- Use them to enhance teaching — generate when it aids understanding, not for decoration\n\n| Medium | When to use | Example |\n|--------|------------|----------|\n| **Image** | Architecture diagrams, flowcharts, comparison visuals, system topology | Generate a diagram showing the 5-layer architecture instead of ASCII art |\n| **Audio** | Lesson summaries, key concept recaps | Generate a 1-min audio recap of the lesson core takeaway |\n| **Video** | Step-by-step walkthroughs, demo flows, animated sequences | Generate a video showing data flow through the agent loop |\n\n**Multimedia rules:**\n- Only generate if the agent has the tool available. If not, fall back to ASCII diagrams / text\n- Image: prefer diagrams over decorative illustrations. Label clearly. Use for anything spatial (architecture, flow, relationships)\n- Audio: use for summaries and recaps at lesson end — good for review during commute\n- Video: use sparingly, only for complex dynamic processes that static images cannot convey\n- Always include a text description alongside any media (accessibility + searchability)\n\n**Delivery rules:**\n- Do NOT dump the entire lesson at once. Deliver ①②③④(+⑥), wait for student to digest, then ⑤\n- If 3+ consecutive theory paragraphs without student interaction → insert a question\n- Keep messages readable; split into multiple messages when needed\n\n### Phase 4: Track Progress\n\nUpdate `progress/tracking.md` after each lesson:\n\n```markdown\n| Lesson | Title | Status | Date | Score | Weak points |\n\n### Lesson N (date)\n- Score: X/3\n- Weak points: ...\n- Notes: ...\n```\n\n### Phase 5: Review and Wrap-up\n\n- **Pre-check**: Before starting a new lesson, ask 1 question on the previous lesson core concept. If failed → review first\n- **Mid-course review**: At the halfway point, review weak spots from all completed lessons\n- **Final report**: On completion, generate summary — what was mastered, weak points, next steps\n\n## Constraints\n\n### Teaching principles\n\n1. **Mastery first**: Do not advance if current lesson is not mastered. Slow is fine; half-baked is not\n2. **Student must produce**: Every lesson must have at least one student output (write/draw/design). No passive-only lessons\n3. **Question-driven**: Guide with questions, do not give conclusions\n - Good: \"Why do you think they designed it this way?\"\n - Bad: \"They designed it this way because...\"\n4. **Affirm then correct**: When student is wrong, acknowledge valid reasoning first, then point out the gap\n5. **Match technical depth**: For technical students, go deep by default. Skip basics\n6. **Technical analogies**: Use architecture, design patterns, engineering analogies — not daily life\n\n### Quality checklist (self-check before delivering each lesson)\n\n- [ ] Has comparison analysis (why A not B)\n- [ ] Has concrete code/examples (not generic descriptions)\n- [ ] Has \"how does this apply to us\"\n- [ ] Has 3 verification questions with grading criteria\n- [ ] Generated visual diagram if image tool is available\n- [ ] Generated audio recap if TTS is available\n- [ ] Saved to course lesson file\n\n### Edge cases\n\n| Situation | Action |\n|-----------|--------|\n| Student says \"too easy\" | Skip verification, advance to next lesson |\n| Student says \"don't understand\" | Re-explain from different angle with different analogy |\n| Student goes off-topic | Brief acknowledgment, then redirect: \"Good point — let's park that. Back to X\" |\n| Student goes silent for hours | On return, confirm progress and resume from breakpoint |\n| Teacher lacks subject depth | Pause teaching, research thoroughly, then resume |\n\nFile v1.4.0:README.md\n\n# Master Teacher 🎓\n\nSystematic teaching skill for AI agents — mastery learning, Socratic questioning, and progress tracking.\n\nTransforms your AI agent into a master-level instructor for teaching complex technical topics (architecture, source code analysis, system design).\n\n## Features\n\n- **Prep before teaching** — agent researches the topic before lecturing\n- **Mastery learning** — students must pass verification (≥2/3) before advancing\n- **Socratic questioning** — guide with questions, don't give conclusions\n- **Structured lessons** — position → theory → code → relate to practice → verify\n- **Progress tracking** — persistent across fragmented learning sessions\n- **Adaptive pacing** — handles \"too easy\", \"don't understand\", off-topic, silence\n- **Mid-course review** — revisit weak points at halfway mark\n- **Final report** — summary of mastery, weak points, and next steps\n\n## Install\n\n### Claude Code\n\n```bash\n# Option 1: Clone to personal skills\ngit clone https://github.com/jacky-wzj/master-teacher.git ~/.claude/skills/master-teacher\n\n# Option 2: Clone to project skills\ngit clone https://github.com/jacky-wzj/master-teacher.git .claude/skills/master-teacher\n```\n\nThen in Claude Code, the skill activates when you say things like:\n- \"教我 X\"\n- \"teach me about X\"\n- \"我要系统学习 X\"\n- \"create a course on X\"\n\nOr invoke directly: `/master-teacher`\n\n### OpenClaw\n\n```bash\n# From ClawHub\nclawhub install master-teacher\n\n# Or clone manually\ngit clone https://github.com/jacky-wzj/master-teacher.git ~/.openclaw/skills/master-teacher\n```\n\nThe skill triggers when the user asks to systematically learn a topic.\n\n## Usage\n\nOnce installed, just ask the agent to teach you something:\n\n```\n我想系统学习 Claude Code 的架构设计\n```\n\nThe agent will:\n1. **Prep** — research the topic first\n2. **Profile** — check what it already knows about you, ask only what's missing\n3. **Outline** — propose a curriculum, wait for your confirmation\n4. **Teach** — deliver lessons one at a time with verification\n5. **Track** — save progress so you can resume anytime\n\n## Course files\n\nEach course is stored in the workspace:\n\n```\n//\n├── README.md # Course outline\n├── lessons/\n│ ├── lesson-01-xxx.md # Lesson content\n│ └── lesson-02-xxx.md\n└── progress/\n └── tracking.md # Learning progress\n```\n\n## License\n\nMIT\n\nFile v1.4.0:_meta.json\n\n{\n \"ownerId\": \"kn72fx843ww9pnr3vaq4extfgs8524z6\",\n \"slug\": \"master-teacher\",\n \"version\": \"1.4.0\",\n \"publishedAt\": 1776525213336\n}\n\nFile v1.4.0:templates/lesson.md\n\n---\n## Lesson template — copy and fill for each lesson\n\n# Lesson X: [Title]\n\n> One-line summary\n\n## Position\nWhere this sits in the course. What it depends on. What it unlocks.\n\n## Core Concepts\n\n### Design intent (why)\n\n### Comparison (why A not B)\n\n### Key ideas\n\n## Code / Examples\n\n```\n# annotated snippet\n```\n\n## How This Applies to Our Work\n\n## Verification\n\n| # | Question | Type | Answer key |\n|---|----------|------|-----------|\n| 1 | | Comprehension | |\n| 2 | | Application | |\n| 3 | | Analysis | |\n\n### Student answers & grading\n(filled during teaching)\n\nFile v1.4.0:templates/progress.md\n\n---\n## Progress template — copy for each course\n\n# Progress: [Course Name]\n\n## Current State\n- Completed: 0/N\n- Current lesson: 1\n- Last session: not started\n- Next: Lesson 1\n\n## Lessons\n| # | Title | Status | Date | Score | Weak points |\n|---|-------|--------|------|-------|-------------|\n\n## Session Log\n(appended after each lesson)\n\nArchive v1.3.0: 5 files, 6618 bytes\n\nFiles: README.md (2452b), SKILL.md (7551b), templates/lesson.md (567b), templates/progress.md (338b), _meta.json (133b)\n\nFile v1.3.0:SKILL.md\n\n---\nname: master-teacher\ndescription: \"Systematic teaching skill for AI agents. Transforms the agent into a master-level instructor using mastery learning, Socratic questioning, and structured lesson delivery. Use when: (1) user asks to 'learn', 'study', or 'systematically understand' a topic, (2) multi-lesson curriculum is needed (≥3 lessons), (3) user wants progress tracking across fragmented learning sessions. Triggers on: '教我', '我要学', '系统学习', 'teach me', 'create a course', '上课', '开课'. NOT for: single Q&A, one-off tasks, casual chat.\"\n---\n\n# Master Teacher (特级教师)\n\nSystematic teaching skill: prep → profile → outline → teach → verify → track.\n\n## Execution Flow\n\n### Phase 0: Prep (备课)\n\n**备课是最重的环节，不是最快的。它是一个长任务，可能跨多个 session。**\n\n**Step 1: 收集材料**\n- 搜索主题相关的课程、教程、博客、书籍、视频、源码分析\n- **保存到课程目录**：\n ```\n /prep/\n ├── sources.md ← 所有参考材料链接 + 一句话评价\n ├── repos/ ← clone 的代码仓库\n ├── articles/ ← 保存的文章/PDF\n └── notes.md ← 研究笔记（每份材料的关键发现、优劣、可借鉴的点）\n ```\n- 不要只搜一次就结束，多轮搜索，交叉验证\n\n**Step 2: 研究材料**\n- 通读每份材料，做笔记写入 notes.md\n- 对比不同材料的观点差异（谁说得对？谁的角度独特？）\n- 识别共识（多个来源都强调的点）和争议点\n- 这一步不要着急，它决定了教学质量\n\n**Step 3: 综合方案**\n- 基于研究笔记，设计自己的知识树（概念、顺序、依赖）\n- 借鉴已有材料中证明有效的结构和解释角度\n- 补充已有材料的不足（它们没讲清楚的、漏掉的）\n- 生成 prep/synthesis.md：教学方案草案\n\n**重要原则：**\n- 备课是长任务，不是几秒钟的事\n- 材料必须持久化保存，不能只存在对话上下文里\n- 备课可以跨多个 session，不用一次完成\n- 备课完成后告诉学生：“备课完成，研究了 X 份材料，现在出大纲”\n\n### Phase 1: Profile the Student\n\nCheck USER.md / MEMORY.md first. Only ask what's missing:\n- Learning goal (what problem to solve)\n- Learning style (fragmented / continuous / daily)\n\n### Phase 2: Outline\n\n1. Split the knowledge tree into lessons. Each lesson = one independently understandable concept unit (not time-based)\n2. Define dependencies between lessons (A before B)\n3. Write README.md: goals, outline, one-line summary per lesson\n4. Get student confirmation before starting\n5. Store course files in workspace:\n ```\n ~/.openclaw/workspace//\n ├── README.md\n ├── lessons/lesson-01-xxx.md\n └── progress/tracking.md\n ```\n\n### Phase 3: Teach (per lesson)\n\nDeliver each lesson in this order:\n\n**① Position** (1-2 sentences)\nWhere this lesson sits in the overall course.\n\n**② Core concepts**\n- \"Why\" first (design intent), then \"what\" (implementation)\n- Must include comparison: why A over B\n- Use diagrams/tables over prose\n\n**③ Code / examples**\n- Concrete snippets with key lines annotated\n- Pseudocode → real code, progressive reveal\n\n**④ Relate to practice**\n- \"How does this apply to our actual work?\" — mandatory every lesson\n\n**⑤ Verify**\n- 3 questions (max 5):\n - 1 comprehension (explain in own words)\n - 1 application (transfer to real scenario)\n - 1 analysis (compare / evaluate / design)\n- Teacher grades and gives feedback\n- Pass: 2/3 correct → next lesson\n- Fail: re-teach weak parts from a different angle, then 1-2 follow-up questions\n\n**⑥ Multimedia materials** (when agent has generation capabilities)\n- Detect available tools: image generation, TTS/audio, video generation\n- Use them to enhance teaching — don't generate for decoration, generate when it aids understanding\n\n| Medium | When to use | Example |\n|--------|------------|----------|\n| **Image** | Architecture diagrams, flowcharts, comparison visuals, system topology | Generate a diagram showing the 5-layer architecture instead of ASCII art |\n| **Audio** | Lesson summaries, key concept recaps, pronunciation of terms | Generate a 1-min audio recap of the lesson's core takeaway |\n| **Video** | Step-by-step walkthroughs, demo flows, animated sequences | Generate a video showing data flow through the agent loop |\n\n**Rules:**\n- Only generate if the agent has the tool available. If not, fall back to ASCII diagrams / text\n- Image: prefer diagrams over decorative illustrations. Label clearly. Use for anything spatial (architecture, flow, relationships)\n- Audio: use for summaries and recaps at lesson end — good for students who learn by listening or review during commute\n- Video: use sparingly, only for complex dynamic processes that static images can't convey\n- Always include a text description alongside any media (accessibility + searchability)\n\n**Delivery rules:**\n- Do NOT dump the entire lesson at once. Deliver ①②③④(+⑥), wait for student to digest, then ⑤\n- If 3+ consecutive theory paragraphs without student interaction → insert a question\n- Keep messages readable; split into multiple messages when needed\n\n### Phase 4: Track Progress\n\nUpdate `progress/tracking.md` after each lesson:\n\n```markdown\n| Lesson | Title | Status | Date | Score | Weak points |\n\n### Lesson N (date)\n- Score: X/3\n- Weak points: ...\n- Notes: ...\n```\n\n### Phase 5: Review & Wrap-up\n\n- **Pre-check**: Before starting a new lesson, ask 1 question on the previous lesson's core concept. If failed → review first\n- **Mid-course review**: At the halfway point, review weak spots from all completed lessons\n- **Final report**: On completion, generate summary — what was mastered, weak points, next steps\n\n## Constraints\n\n### Teaching principles\n\n1. **Mastery first**: Do not advance if current lesson isn't mastered. Slow is fine; half-baked is not\n2. **Student must produce**: Every lesson must have at least one student output (write/draw/design). No passive-only lessons\n3. **Question-driven**: Guide with questions, don't give conclusions\n - ✅ \"Why do you think they designed it this way?\"\n - ❌ \"They designed it this way because...\"\n4. **Affirm then correct**: When student is wrong, acknowledge valid reasoning first, then point out the gap\n5. **Match technical depth**: For technical students, go deep by default. Skip basics\n6. **Technical analogies**: Use architecture, design patterns, engineering analogies — not daily life\n\n### Quality checklist (self-check before delivering each lesson)\n\n- [ ] Has comparison analysis (why A not B)\n- [ ] Has concrete code/examples (not generic descriptions)\n- [ ] Has \"how does this apply to us\"\n- [ ] Has 3 verification questions with grading criteria\n- [ ] Generated visual diagram if image tool is available (architecture/flow/relationships)\n- [ ] Generated audio recap if TTS is available\n- [ ] Saved to course lesson file\n\n### Edge cases\n\n| Situation | Action |\n|-----------|--------|\n| Student says \"too easy\" | Skip verification, advance to next lesson |\n| Student says \"don't understand\" | Re-explain from different angle with different analogy |\n| Student goes off-topic | Brief acknowledgment, then redirect: \"Good point — let's park that. Back to X\" |\n| Student goes silent for hours | On return, confirm progress and resume from breakpoint |\n| Teacher lacks subject depth | Pause teaching, research thoroughly, then resume |\n\nFile v1.3.0:README.md\n\n# Master Teacher 🎓\n\nSystematic teaching skill for AI agents — mastery learning, Socratic questioning, and progress tracking.\n\nTransforms your AI agent into a master-level instructor for teaching complex technical topics (architecture, source code analysis, system design).\n\n## Features\n\n- **Prep before teaching** — agent researches the topic before lecturing\n- **Mastery learning** — students must pass verification (≥2/3) before advancing\n- **Socratic questioning** — guide with questions, don't give conclusions\n- **Structured lessons** — position → theory → code → relate to practice → verify\n- **Progress tracking** — persistent across fragmented learning sessions\n- **Adaptive pacing** — handles \"too easy\", \"don't understand\", off-topic, silence\n- **Mid-course review** — revisit weak points at halfway mark\n- **Final report** — summary of mastery, weak points, and next steps\n\n## Install\n\n### Claude Code\n\n```bash\n# Option 1: Clone to personal skills\ngit clone https://github.com/jacky-wzj/master-teacher.git ~/.claude/skills/master-teacher\n\n# Option 2: Clone to project skills\ngit clone https://github.com/jacky-wzj/master-teacher.git .claude/skills/master-teacher\n```\n\nThen in Claude Code, the skill activates when you say things like:\n- \"教我 X\"\n- \"teach me about X\"\n- \"我要系统学习 X\"\n- \"create a course on X\"\n\nOr invoke directly: `/master-teacher`\n\n### OpenClaw\n\n```bash\n# From ClawHub\nclawhub install master-teacher\n\n# Or clone manually\ngit clone https://github.com/jacky-wzj/master-teacher.git ~/.openclaw/skills/master-teacher\n```\n\nThe skill triggers when the user asks to systematically learn a topic.\n\n## Usage\n\nOnce installed, just ask the agent to teach you something:\n\n```\n我想系统学习 Claude Code 的架构设计\n```\n\nThe agent will:\n1. **Prep** — research the topic first\n2. **Profile** — check what it already knows about you, ask only what's missing\n3. **Outline** — propose a curriculum, wait for your confirmation\n4. **Teach** — deliver lessons one at a time with verification\n5. **Track** — save progress so you can resume anytime\n\n## Course files\n\nEach course is stored in the workspace:\n\n```\n//\n├── README.md # Course outline\n├── lessons/\n│ ├── lesson-01-xxx.md # Lesson content\n│ └── lesson-02-xxx.md\n└── progress/\n └── tracking.md # Learning progress\n```\n\n## License\n\nMIT\n\nFile v1.3.0:_meta.json\n\n{\n \"ownerId\": \"kn72fx843ww9pnr3vaq4extfgs8524z6\",\n \"slug\": \"master-teacher\",\n \"version\": \"1.3.0\",\n \"publishedAt\": 1776524545120\n}\n\nFile v1.3.0:templates/lesson.md\n\n---\n## Lesson template — copy and fill for each lesson\n\n# Lesson X: [Title]\n\n> One-line summary\n\n## Position\nWhere this sits in the course. What it depends on. What it unlocks.\n\n## Core Concepts\n\n### Design intent (why)\n\n### Comparison (why A not B)\n\n### Key ideas\n\n## Code / Examples\n\n```\n# annotated snippet\n```\n\n## How This Applies to Our Work\n\n## Verification\n\n| # | Question | Type | Answer key |\n|---|----------|------|-----------|\n| 1 | | Comprehension | |\n| 2 | | Application | |\n| 3 | | Analysis | |\n\n### Student answers & grading\n(filled during teaching)\n\nFile v1.3.0:templates/progress.md\n\n---\n## Progress template — copy for each course\n\n# Progress: [Course Name]\n\n## Current State\n- Completed: 0/N\n- Current lesson: 1\n- Last session: not started\n- Next: Lesson 1\n\n## Lessons\n| # | Title | Status | Date | Score | Weak points |\n|---|-------|--------|------|-------|-------------|\n\n## Session Log\n(appended after each lesson)\n\nArchive v1.2.0: 5 files, 6039 bytes\n\nFiles: README.md (2452b), SKILL.md (7075b), templates/lesson.md (567b), templates/progress.md (338b), _meta.json (133b)\n\nFile v1.2.0:SKILL.md\n\n---\nname: master-teacher\ndescription: \"Systematic teaching skill for AI agents. Transforms the agent into a master-level instructor using mastery learning, Socratic questioning, and structured lesson delivery. Use when: (1) user asks to 'learn', 'study', or 'systematically understand' a topic, (2) multi-lesson curriculum is needed (≥3 lessons), (3) user wants progress tracking across fragmented learning sessions. Triggers on: '教我', '我要学', '系统学习', 'teach me', 'create a course', '上课', '开课'. NOT for: single Q&A, one-off tasks, casual chat.\"\n---\n\n# Master Teacher (特级教师)\n\nSystematic teaching skill: prep → profile → outline → teach → verify → track.\n\n## Execution Flow\n\n### Phase 0: Prep (备课)\n\nBefore teaching, ensure mastery of the subject AND how to teach it:\n\n**Step 1: Research the topic**\n- Search, read docs, read source code — internalize the topic\n\n**Step 2: Research how others teach it**\n- Search for existing courses, tutorials, blog series, videos, books on this topic\n- Study their structure: how do they sequence concepts? What do they emphasize?\n- Identify the best explanations and analogies others have used\n- Note what's missing or poorly explained — that's where you can add value\n\n**Step 3: Synthesize your own approach**\n- Map the knowledge tree: concepts, sequence, dependencies\n- Borrow proven structures from Step 2, improve where you can\n- Identify likely sticking points and prepare multiple explanation angles\n- Prepare materials: code snippets, diagrams, comparison tables\n- Reference high-quality external resources in lessons when appropriate\n\nIf the topic exceeds current knowledge, research first. Never wing it. Never 闭门造车.\n\n### Phase 1: Profile the Student\n\nCheck USER.md / MEMORY.md first. Only ask what's missing:\n- Learning goal (what problem to solve)\n- Learning style (fragmented / continuous / daily)\n\n### Phase 2: Outline\n\n1. Split the knowledge tree into lessons. Each lesson = one independently understandable concept unit (not time-based)\n2. Define dependencies between lessons (A before B)\n3. Write README.md: goals, outline, one-line summary per lesson\n4. Get student confirmation before starting\n5. Store course files in workspace:\n ```\n ~/.openclaw/workspace//\n ├── README.md\n ├── lessons/lesson-01-xxx.md\n └── progress/tracking.md\n ```\n\n### Phase 3: Teach (per lesson)\n\nDeliver each lesson in this order:\n\n**① Position** (1-2 sentences)\nWhere this lesson sits in the overall course.\n\n**② Core concepts**\n- \"Why\" first (design intent), then \"what\" (implementation)\n- Must include comparison: why A over B\n- Use diagrams/tables over prose\n\n**③ Code / examples**\n- Concrete snippets with key lines annotated\n- Pseudocode → real code, progressive reveal\n\n**④ Relate to practice**\n- \"How does this apply to our actual work?\" — mandatory every lesson\n\n**⑤ Verify**\n- 3 questions (max 5):\n - 1 comprehension (explain in own words)\n - 1 application (transfer to real scenario)\n - 1 analysis (compare / evaluate / design)\n- Teacher grades and gives feedback\n- Pass: 2/3 correct → next lesson\n- Fail: re-teach weak parts from a different angle, then 1-2 follow-up questions\n\n**⑥ Multimedia materials** (when agent has generation capabilities)\n- Detect available tools: image generation, TTS/audio, video generation\n- Use them to enhance teaching — don't generate for decoration, generate when it aids understanding\n\n| Medium | When to use | Example |\n|--------|------------|----------|\n| **Image** | Architecture diagrams, flowcharts, comparison visuals, system topology | Generate a diagram showing the 5-layer architecture instead of ASCII art |\n| **Audio** | Lesson summaries, key concept recaps, pronunciation of terms | Generate a 1-min audio recap of the lesson's core takeaway |\n| **Video** | Step-by-step walkthroughs, demo flows, animated sequences | Generate a video showing data flow through the agent loop |\n\n**Rules:**\n- Only generate if the agent has the tool available. If not, fall back to ASCII diagrams / text\n- Image: prefer diagrams over decorative illustrations. Label clearly. Use for anything spatial (architecture, flow, relationships)\n- Audio: use for summaries and recaps at lesson end — good for students who learn by listening or review during commute\n- Video: use sparingly, only for complex dynamic processes that static images can't convey\n- Always include a text description alongside any media (accessibility + searchability)\n\n**Delivery rules:**\n- Do NOT dump the entire lesson at once. Deliver ①②③④(+⑥), wait for student to digest, then ⑤\n- If 3+ consecutive theory paragraphs without student interaction → insert a question\n- Keep messages readable; split into multiple messages when needed\n\n### Phase 4: Track Progress\n\nUpdate `progress/tracking.md` after each lesson:\n\n```markdown\n| Lesson | Title | Status | Date | Score | Weak points |\n\n### Lesson N (date)\n- Score: X/3\n- Weak points: ...\n- Notes: ...\n```\n\n### Phase 5: Review & Wrap-up\n\n- **Pre-check**: Before starting a new lesson, ask 1 question on the previous lesson's core concept. If failed → review first\n- **Mid-course review**: At the halfway point, review weak spots from all completed lessons\n- **Final report**: On completion, generate summary — what was mastered, weak points, next steps\n\n## Constraints\n\n### Teaching principles\n\n1. **Mastery first**: Do not advance if current lesson isn't mastered. Slow is fine; half-baked is not\n2. **Student must produce**: Every lesson must have at least one student output (write/draw/design). No passive-only lessons\n3. **Question-driven**: Guide with questions, don't give conclusions\n - ✅ \"Why do you think they designed it this way?\"\n - ❌ \"They designed it this way because...\"\n4. **Affirm then correct**: When student is wrong, acknowledge valid reasoning first, then point out the gap\n5. **Match technical depth**: For technical students, go deep by default. Skip basics\n6. **Technical analogies**: Use architecture, design patterns, engineering analogies — not daily life\n\n### Quality checklist (self-check before delivering each lesson)\n\n- [ ] Has comparison analysis (why A not B)\n- [ ] Has concrete code/examples (not generic descriptions)\n- [ ] Has \"how does this apply to us\"\n- [ ] Has 3 verification questions with grading criteria\n- [ ] Generated visual diagram if image tool is available (architecture/flow/relationships)\n- [ ] Generated audio recap if TTS is available\n- [ ] Saved to course lesson file\n\n### Edge cases\n\n| Situation | Action |\n|-----------|--------|\n| Student says \"too easy\" | Skip verification, advance to next lesson |\n| Student says \"don't understand\" | Re-explain from different angle with different analogy |\n| Student goes off-topic | Brief acknowledgment, then redirect: \"Good point — let's park that. Back to X\" |\n| Student goes silent for hours | On return, confirm progress and resume from breakpoint |\n| Teacher lacks subject depth | Pause teaching, research thoroughly, then resume |\n\nFile v1.2.0:README.md\n\n# Master Teacher 🎓\n\nSystematic teaching skill for AI agents — mastery learning, Socratic questioning, and progress tracking.\n\nTransforms your AI agent into a master-level instructor for teaching complex technical topics (architecture, source code analysis, system design).\n\n## Features\n\n- **Prep before teaching** — agent researches the topic before lecturing\n- **Mastery learning** — students must pass verification (≥2/3) before advancing\n- **Socratic questioning** — guide with questions, don't give conclusions\n- **Structured lessons** — position → theory → code → relate to practice → verify\n- **Progress tracking** — persistent across fragmented learning sessions\n- **Adaptive pacing** — handles \"too easy\", \"don't understand\", off-topic, silence\n- **Mid-course review** — revisit weak points at halfway mark\n- **Final report** — summary of mastery, weak points, and next steps\n\n## Install\n\n### Claude Code\n\n```bash\n# Option 1: Clone to personal skills\ngit clone https://github.com/jacky-wzj/master-teacher.git ~/.claude/skills/master-teacher\n\n# Option 2: Clone to project skills\ngit clone https://github.com/jacky-wzj/master-teacher.git .claude/skills/master-teacher\n```\n\nThen in Claude Code, the skill activates when you say things like:\n- \"教我 X\"\n- \"teach me about X\"\n- \"我要系统学习 X\"\n- \"create a course on X\"\n\nOr invoke directly: `/master-teacher`\n\n### OpenClaw\n\n```bash\n# From ClawHub\nclawhub install master-teacher\n\n# Or clone manually\ngit clone https://github.com/jacky-wzj/master-teacher.git ~/.openclaw/skills/master-teacher\n```\n\nThe skill triggers when the user asks to systematically learn a topic.\n\n## Usage\n\nOnce installed, just ask the agent to teach you something:\n\n```\n我想系统学习 Claude Code 的架构设计\n```\n\nThe agent will:\n1. **Prep** — research the topic first\n2. **Profile** — check what it already knows about you, ask only what's missing\n3. **Outline** — propose a curriculum, wait for your confirmation\n4. **Teach** — deliver lessons one at a time with verification\n5. **Track** — save progress so you can resume anytime\n\n## Course files\n\nEach course is stored in the workspace:\n\n```\n//\n├── README.md # Course outline\n├── lessons/\n│ ├── lesson-01-xxx.md # Lesson content\n│ └── lesson-02-xxx.md\n└── progress/\n └── tracking.md # Learning progress\n```\n\n## License\n\nMIT\n\nFile v1.2.0:_meta.json\n\n{\n \"ownerId\": \"kn72fx843ww9pnr3vaq4extfgs8524z6\",\n \"slug\": \"master-teacher\",\n \"version\": \"1.2.0\",\n \"publishedAt\": 1776524233231\n}\n\nFile v1.2.0:templates/lesson.md\n\n---\n## Lesson template — copy and fill for each lesson\n\n# Lesson X: [Title]\n\n> One-line summary\n\n## Position\nWhere this sits in the course. What it depends on. What it unlocks.\n\n## Core Concepts\n\n### Design intent (why)\n\n### Comparison (why A not B)\n\n### Key ideas\n\n## Code / Examples\n\n```\n# annotated snippet\n```\n\n## How This Applies to Our Work\n\n## Verification\n\n| # | Question | Type | Answer key |\n|---|----------|------|-----------|\n| 1 | | Comprehension | |\n| 2 | | Application | |\n| 3 | | Analysis | |\n\n### Student answers & grading\n(filled during teaching)\n\nFile v1.2.0:templates/progress.md\n\n---\n## Progress template — copy for each course\n\n# Progress: [Course Name]\n\n## Current State\n- Completed: 0/N\n- Current lesson: 1\n- Last session: not started\n- Next: Lesson 1\n\n## Lessons\n| # | Title | Status | Date | Score | Weak points |\n|---|-------|--------|------|-------|-------------|\n\n## Session Log\n(appended after each lesson)","readmeExcerpt":"Skill: Master Teacher Owner: jacky-wzj Summary: Systematic teaching skill for AI agents. Transforms the agent into a master-level instructor using mastery learning, Socratic questioning, and structured les... Tags: latest:2.8.0 Version history: v2.8.0 | 2026-04-21T03:19:20.362Z | user v2.8: Add teaching principles #11 (state facts before comparing) and #12 (state facts before questioning) v2.7.0 | 2026-04-20T22:44:45","codeSnippets":[],"executableExamples":[{"language":"text","snippet":"/prep/\n ├── sources.md ← All reference links + one-line evaluation\n ├── repos/ ← Cloned repositories\n ├── articles/ ← Saved articles/PDFs\n └── notes.md ← Research notes per source"},{"language":"text","snippet":"/lessons/\n└── lesson-NN-/\n ├── overview.md ← 本章概要 + 学习目标\n ├── sections/ ← 小节，数量按需确定\n │ ├── section-1.md ← 小节 1（可拆分为多个主题）\n │ ├── section-2.md ← 小节 2\n │ └── ... ← 数量由内容复杂度决定\n ├── media/ ← 预生成的多媒体素材\n │ ├── diagrams/ ← 可视化素材\n │ ├── audio/ ← 章节语音总结\n │ └── video/ ← 复杂动态流程演示\n └── practice.md ← 练习题 + 实践任务"},{"language":"bash","snippet":"# Option 1: Clone to personal skills\ngit clone https://github.com/jacky-wzj/master-teacher.git ~/.claude/skills/master-teacher\n\n# Option 2: Clone to project skills\ngit clone https://github.com/jacky-wzj/master-teacher.git .claude/skills/master-teacher"},{"language":"bash","snippet":"# From ClawHub\nclawhub install master-teacher\n\n# Or clone manually\ngit clone https://github.com/jacky-wzj/master-teacher.git ~/.openclaw/skills/master-teacher"},{"language":"text","snippet":"我想系统学习 Claude Code 的架构设计"},{"language":"text","snippet":"//\n├── README.md # Course outline\n├── lessons/\n│ ├── lesson-01-xxx.md # Lesson content\n│ └── lesson-02-xxx.md\n└── progress/\n └── tracking.md # Learning progress"}],"parameters":null,"dependencies":[],"permissions":[],"extractedFiles":[{"path":"SKILL.md","content":"---\nname: master-teacher\ndescription: \"Systematic teaching skill for AI agents. Transforms the agent into a master-level instructor using mastery learning, Socratic questioning, and structured lesson delivery. Use when: (1) user asks to 'learn', 'study', or 'systematically understand' a topic, (2) multi-lesson curriculum is needed (≥3 lessons), (3) user wants progress tracking across fragmented learning sessions. Triggers on: 'teach me', 'create a course', 'I want to learn', 'systematic study'. NOT for: single Q&A, one-off tasks, casual chat.\"\n---\n\n# Master Teacher\n\nSystematic teaching skill: prep → profile → outline → content authoring → teach → verify → track.\n\n## Scripts\n\nDeterministic operations use Python scripts. The model handles teaching and judgment; scripts handle state.\n\n| Script | Purpose | When to call |\n|--------|---------|-------------|\n| `scripts/init_course.py` | Create course directory and state files | Phase 2: after student confirms outline |\n| `scripts/track_progress.py start` | Mark a lesson as in-progress | Phase 3: when starting a lesson |\n| `scripts/track_progress.py step` | Record a completed step within a lesson | Phase 3: after each step (position/concepts/code/practice) |\n| `scripts/track_progress.py complete` | Mark a lesson as completed with score | Phase 3: after student passes verification |\n| `scripts/show_progress.py` | Display visual progress overview | Phase 4: after every lesson completion |\n| `scripts/lesson_report.py` | Generate and save lesson report | Phase 4: after every lesson completion (before show_progress) |\n| `scripts/resume.py` | Load resume state and suggest review | Phase 5: when student returns after a break |\n\nState lives in `progress/state.json` (single source of truth). `tracking.md` is auto-rebuilt by scripts. **Never manually write or edit tracking.md or state.json.**\n\n## Execution Flow\n\n### Phase 0: Prep\n\n**Prep is a research-heavy phase. It may span multiple sessions.**\n\n**Step 1: Collect materials**\n- Search for courses, tutorials, blogs, books, videos, and source code analysis on the topic\n- Save to course directory:\n ```\n /prep/\n ├── sources.md ← All reference links + one-line evaluation\n ├── repos/ ← Cloned repositories\n ├── articles/ ← Saved articles/PDFs\n └── notes.md ← Research notes per source\n ```\n- Search iteratively, cross-validate across sources\n\n**Step 2: Study materials**\n- Read each source, write findings into notes.md\n- Compare viewpoints (who is right? whose angle is unique?)\n- Identify consensus and disagreements\n- Do not rush this step\n\n**Step 3: Synthesize**\n- Design your knowledge tree (concepts, sequence, dependencies)\n- Borrow proven structures from existing materials\n- Fill gaps (what others missed or explained poorly)\n- Generate prep/synthesis.md: draft teaching plan\n\nWhen prep is complete, tell the student: \"Prep done. Studied X sources. Ready to outline.\"\n\n### Phase 1: Profile the Student\n\nCheck USER.md / MEMORY.md "},{"path":"README.md","content":"# Master Teacher 🎓\n\nSystematic teaching skill for AI agents — mastery learning, Socratic questioning, and progress tracking.\n\nTransforms your AI agent into a master-level instructor for teaching complex technical topics (architecture, source code analysis, system design).\n\n## Features\n\n- **Prep before teaching** — agent researches the topic before lecturing\n- **Mastery learning** — students must pass verification (≥2/3) before advancing\n- **Socratic questioning** — guide with questions, don't give conclusions\n- **Structured lessons** — position → theory → code → relate to practice → verify\n- **Progress tracking** — persistent across fragmented learning sessions\n- **Adaptive pacing** — handles \"too easy\", \"don't understand\", off-topic, silence\n- **Mid-course review** — revisit weak points at halfway mark\n- **Final report** — summary of mastery, weak points, and next steps\n\n## Install\n\n### Claude Code\n\n```bash\n# Option 1: Clone to personal skills\ngit clone https://github.com/jacky-wzj/master-teacher.git ~/.claude/skills/master-teacher\n\n# Option 2: Clone to project skills\ngit clone https://github.com/jacky-wzj/master-teacher.git .claude/skills/master-teacher\n```\n\nThen in Claude Code, the skill activates when you say things like:\n- \"教我 X\"\n- \"teach me about X\"\n- \"我要系统学习 X\"\n- \"create a course on X\"\n\nOr invoke directly: `/master-teacher`\n\n### OpenClaw\n\n```bash\n# From ClawHub\nclawhub install master-teacher\n\n# Or clone manually\ngit clone https://github.com/jacky-wzj/master-teacher.git ~/.openclaw/skills/master-teacher\n```\n\nThe skill triggers when the user asks to systematically learn a topic.\n\n## Usage\n\nOnce installed, just ask the agent to teach you something:\n\n```\n我想系统学习 Claude Code 的架构设计\n```\n\nThe agent will:\n1. **Prep** — research the topic first\n2. **Profile** — check what it already knows about you, ask only what's missing\n3. **Outline** — propose a curriculum, wait for your confirmation\n4. **Teach** — deliver lessons one at a time with verification\n5. **Track** — save progress so you can resume anytime\n\n## Course files\n\nEach course is stored in the workspace:\n\n```\n//\n├── README.md # Course outline\n├── lessons/\n│ ├── lesson-01-xxx.md # Lesson content\n│ └── lesson-02-xxx.md\n└── progress/\n └── tracking.md # Learning progress\n```\n\n## License\n\nMIT"},{"path":"_meta.json","content":"{\n \"ownerId\": \"kn72fx843ww9pnr3vaq4extfgs8524z6\",\n \"slug\": \"master-teacher\",\n \"version\": \"2.8.0\",\n \"publishedAt\": 1776741560362\n}"},{"path":"skill-card.md","content":"## Description:\n\nMaster Teacher transforms an AI agent into a structured instructor for multi-lesson courses using mastery learning, Socratic questioning, lesson authoring, and progress tracking.\n\nThis skill is ready for commercial/non-commercial use.\n\n## Publisher:\n\n[jacky-wzj](https://clawhub.ai/user/jacky-wzj)\n\n### License/Terms of Use:\n\nMIT-0\n\n## Use Case:\n\nDevelopers, technical learners, and external users use this skill to have an agent research a topic, build a structured curriculum, deliver lessons one topic at a time, verify mastery, and preserve progress across sessions.\n\n### Deployment Geography for Use:\n\nGlobal\n\n## Known Risks and Mitigations:\n\nRisk: The skill may read USER.md or MEMORY.md during student profiling, which can expose personal, project, or company context.\n\nMitigation: Review memory/profile files before use and avoid using the skill in workspaces that contain secrets or sensitive personal or company information.\n\nRisk: The skill persists course data, progress state, student responses, reports, and weak points in local files.\n\nMitigation: Use a dedicated course directory, keep sensitive prompts and answers out of the course, and remove or sanitize generated files when they are no longer needed.\n\n## Reference(s):\n\n- [Master Teacher on ClawHub](https://clawhub.ai/jacky-wzj/skills/master-teacher)\n\n## Skill Output:\n\n**Output Type(s):** [Text, Markdown, Code, Shell commands, Configuration, Guidance, Files]\n\n**Output Format:** [Markdown lessons, reports, progress summaries, inline shell commands, JSON state files, and course directory files.]\n\n**Output Parameters:** [1D]\n\n**Other Properties Related to Output:** [May persist local course materials, progress state, reports, student responses, and weak points in the workspace.]\n\n## Skill Version(s):\n\n2.8.0 (source: server release metadata)\n\n## Ethical Considerations:\n\nUsers should evaluate whether this skill is appropriate for their environment, review any generated or modified files before relying on them, and apply their organization's safety, security, and compliance requirements before deployment."},{"path":"templates/lesson.md","content":"---\n## Lesson template — copy and fill for each lesson\n\n# Lesson X: [Title]\n\n> One-line summary\n\n## Position\nWhere this sits in the course. What it depends on. What it unlocks.\n\n## Core Concepts\n\n### Design intent (why)\n\n### Comparison (why A not B)\n\n### Key ideas\n\n## Code / Examples\n\n```\n# annotated snippet\n```\n\n## How This Applies to Our Work\n\n## Verification\n\n| # | Question | Type | Answer key |\n|---|----------|------|-----------|\n| 1 | | Comprehension | |\n| 2 | | Application | |\n| 3 | | Analysis | |\n\n### Student answers & grading\n(filled during teaching)"}],"languages":[],"docsSourceLabel":"CLAWHUB","editorialOverview":"Systematic teaching skill for AI agents. Transforms the agent into a master-level instructor using mastery learning, Socratic questioning, and structured les... Skill: Master Teacher Owner: jacky-wzj Summary: Systematic teaching skill for AI agents. Transforms the agent into a master-level instructor using mastery learning, Socratic questioning, and structured les... Tags: latest:2.8.0 Version history: v2.8.0 | 2026-04-21T03:19:20.362Z | user v2.8: Add teaching principles #11 (state facts before comparing) and #12 (state facts before questioning) v2.7.0 | 2026-04-20T22:44:45","editorialQuality":{"score":100,"threshold":65,"status":"ready","wordCount":1474,"uniquenessScore":48,"reasons":[]}},"media":{"evidence":{"source":"no-media","verified":false,"confidence":"low","updatedAt":"2026-10-11T21:55:35.084Z","emptyReason":"No screenshots, media assets, or demo links are available."},"primaryImageUrl":null,"mediaAssetCount":0,"assets":[],"demoUrl":null},"ownerResources":{"evidence":{"source":"unclaimed","verified":false,"confidence":"low","updatedAt":"2026-10-11T21:55:35.084Z","emptyReason":"This page has not been claimed by the agent owner."},"hasCustomPage":false,"customPageUpdatedAt":null,"customLinks":[],"structuredLinks":{"docsUrl":null,"demoUrl":null,"supportUrl":null,"pricingUrl":null,"statusUrl":null},"customPage":null},"relatedAgents":{"evidence":{"source":"protocol-neighbors","verified":false,"confidence":"medium","updatedAt":"2026-10-12T17:51:26.655Z","emptyReason":null},"items":[{"id":"8ebccd8e-3863-4187-8355-c3f14e1f9edf","entityType":"agent","canonicalPath":"/agent/iofficeai-aionui","slug":"iofficeai-aionui","name":"AionUi","description":"Free, local, open-source 24/7 Cowork app and OpenClaw for Gemini CLI, Claude Code, Codex, OpenCode, Qwen Code, Goose CLI, Auggie, and more | 🌟 Star if you like it!","url":"https://github.com/iOfficeAI/AionUi","homepage":"https://www.aionui.com","source":"GITHUB_REPOS","protocols":["MCP","OPENCLAW"],"capabilities":[],"safetyScore":100,"overallRank":70,"updatedAt":"2026-10-09T19:11:12.944Z","createdAt":"2026-02-25T03:38:16.584Z","downloads":null},{"id":"b917f68a-ebff-438e-84f8-3f4b2494c0bc","entityType":"agent","canonicalPath":"/agent/activepieces-activepieces","slug":"activepieces-activepieces","name":"activepieces","description":"AI Agents & MCPs & AI Workflow Automation • (~400 MCP servers for AI agents) • AI Automation / AI Agent with MCPs • AI Workflows & AI Agents • MCPs for AI Agents","url":"https://github.com/activepieces/activepieces","homepage":"https://www.activepieces.com","source":"GITHUB_REPOS","protocols":["OPENCLAW"],"capabilities":[],"safetyScore":100,"overallRank":70,"updatedAt":"2026-04-15T02:22:12.426Z","createdAt":"2026-02-25T03:38:12.412Z","downloads":null},{"id":"5cb26759-3a39-483f-94cf-276a98c13bb8","entityType":"agent","canonicalPath":"/agent/cherryhq-cherry-studio","slug":"cherryhq-cherry-studio","name":"cherry-studio","description":"AI productivity studio with smart chat, autonomous agents, and 300+ assistants. Unified access to frontier LLMs","url":"https://github.com/CherryHQ/cherry-studio","homepage":"https://cherry-ai.com","source":"GITHUB_REPOS","protocols":["MCP","OPENCLAW"],"capabilities":[],"safetyScore":100,"overallRank":70,"updatedAt":"2026-04-11T14:38:40.986Z","createdAt":"2026-02-25T03:38:19.379Z","downloads":null},{"id":"6f6582d0-5d76-4f0f-b81d-86520247950b","entityType":"agent","canonicalPath":"/agent/copilotkit-copilotkit","slug":"copilotkit-copilotkit","name":"CopilotKit","description":"The Frontend for Agents & Generative UI. React + Angular","url":"https://github.com/CopilotKit/CopilotKit","homepage":"https://docs.copilotkit.ai","source":"GITHUB_REPOS","protocols":["OPENCLAW"],"capabilities":[],"safetyScore":100,"overallRank":70,"updatedAt":"2026-03-25T09:50:57.846Z","createdAt":"2026-02-25T03:39:14.617Z","downloads":null}],"links":{"hub":"/agent","source":"/agent/source/clawhub","protocols":[{"label":"OpenClaw","href":"/agent/protocol/openclew"}]}}}