SuperAPI AI Directory
Partner with us中文AI newsFavorites
Skillslearning-master

learning-master

Writing & research

View repository

Updated 10-15 22:34

About this skill

Learning Master

Structured learning skill based on thesix-stage learning method. Foran already determined course or book, guides the user from establishing a plan to the final output.

建立计划 → 资源转换 → 预习 → 通读 → 内化 → 外化
    ↓          ↓        ↓       ↓       ↓       ↓
目录+规划   转为MD    框架梳理  深度笔记  Anki卡片  作品产出

Command Quick Reference

Command Description Stage
/study-init-course Create learning project (directory + plan + resource conversion) Establish Plan
/study-step1-preview Generate preview notes Preview
/study-step2-notes Generate detailed read-through notes Read-through
/study-step3-anki-review Generate Anki cards and export TSV Internalization
/study-step4-output-works Generate externalization suggestions and action plan Externalization
/study-run-full-flow Complete workflow (Preview → Read-through → Internalization → Externalization) All Stages
/study-step5-final-summary Course completion summary All Stages

General Conventions

Project Directory Structure

projects/<课程名>/
├── plan.md                # 学习计划
├── summary.md             # 课程总结
├── 0-resources/           # 原始资源与转换后的 Markdown
├── 1-preview/             # 预习笔记
├── 2-notes/               # 通读笔记
├── 3-review/              # Anki 卡片（Markdown 格式，-3.md）
│   └── anki-export/       # Anki 导入文件（TSV 格式，-3-anki.txt）
├── 4-works/               # 外化产出
└── 9-archive/             # 归档

File Naming Rules

Stage Suffix Example
Preview -1.md 第1章-1.md
Read-through -2.md 第1章-2.md
Review -3.md 第1章-3.md
Externalization -4.md 第1章-4.md
  • Mandatory requirements: Preserve the complete information of the original file name, only add the stage suffix
  • Course-level files (plan.md、summary.md) are placed directly in the project root directory
  • Original resources and converted Markdown are uniformly placed in 0-resources/

Progress Status

Status Meaning
⏳ Not yet started
🔄 Currently learning
📝 Notes generated, learning to be completed
✅ Learning completed

Mandatory Rules：

  • After generating notes, mark as 📝，Cannotdirectly mark as ✅
  • Only after the user explicitly confirms "finished learning" can it be marked as ✅ and fill in the completion date
  • plan.md is the sole record point for course-level progress
  • Chapter status updates must remain consistent with the corresponding output directory

Update Timing：

  • After executing Preview, Read-through, Internalization, Externalization: update to 📝 or 🔄
  • After the user confirms completion of chapter learning: update to ✅
  • After completing the course summary: supplement the completion log

Writing Standards

  • UseSimplified Chinese, use Chinese punctuation: ，。！？：；""''
  • Use English punctuation for English/code content
  • Use ……for ellipsis (6 dots)
  • Add a space between Chinese and English:学习 Python 语言
  • Add a space between numbers and Chinese:共 3 个步骤
  • Wrap code and file names in backticks
  • For the first occurrence of an English term, append Chinese:Markdown（MD）, use a consistent form thereafter
  • Keep proper nouns in their original form: GitHub, Python

File Conflict Handling

When the target file already exists, first ask the user whether tooverwriteorappend. Default append—add at the end of the original file ## 补充 block.

Command: /study-init-course

Create the course directory, study plan, and complete resource preparation.

Steps

  1. Confirm project information—Ask the user:

    • Topic/name (course name, book title, or article title)
    • Type (online course, book, offline training, WeChat official account article, etc.)
    • Source (platform, publisher, or URL)
    • Content volume (total duration, page count, or word count)
    • Schedule (time invested per day/week, how long you plan to take to finish)
    • Learning motivation (why learn? What problem do you want to solve?)
    • Output goals (what results do you plan to produce after learning?)

  2. Confirm understanding—Briefly restate and confirm the project name

  3. Create the directory structure and organize the original files

    mkdir -p "projects/<主题名称>/"{0-resources,1-preview,2-notes,3-review,4-works,9-archive}
    • Mandatory requirement: if the user provides original materials (such as PDF, Word, video, etc.), the materials must be organized under the projects/<主题名称>/0-resources/ directory; copy by default, and only move when the user explicitly requests organizing the source files.

  4. Generate study plan—Refer to the template templates/0-plan.md, write to projects/<主题名称>/plan.md

    • Three-dimensional review: analyze relevance, value, and goal orientation
    • SMART goals: transform vague goals into measurable specific goals
    • Structural adaptation: books by "chapter/section", courses by "week/module", articles by "core viewpoint"

  5. Resource conversion(if original materials exist)—invoke the markdown-conversion skill to convert materials to Markdown:

    • Document types (Word/HTML/EPUB) → Pandoc
    • Native PDF → local parsing
    • Scanned/complex PDF → MinerU
    • Subtitle directory → subtitle conversion script
    • See details in skills/markdown-conversion/SKILL.md
    • Conversion results are stored in projects/<课程>/0-resources/
    • Check quality: no garbled text, smooth reading order

  6. Next steps—After resource preparation is complete, proceed to /study-step1-preview

Command: /study-step1-preview

Quickly build a framework and enter deep learning with questions. Preview does not seek depth, only "knowing what will be covered", 5-10 minutes/chapter.

Steps

  1. Confirm course—Scan projects/ and list existing courses for the user to choose from, confirm the chapter range

  2. Get content—Prioritize reading the content in projects/<课程>/0-resources/ content in

  3. Generate preview notes—Generate according to template templates/1-preview.md , including:

    • Author/instructor introduction
    • Content overview (2-3 sentences)
    • Content structure (tree diagram)
    • Core concept table
    • Key points (3-5 items)
    • Preview questions (3 items)

  4. Save——projects/<课程>/1-preview/[原名]-1.md

  5. Follow-up suggestions—Can upload to NotebookLM to generate an audio podcast → proceed to /study-step2-notes

Command: /study-step2-notes

Deeply understand the content and generate detailed chapter notes. Read for understanding; after finishing a chapter, you should be able to explain it to someone else.

Steps

  1. Confirm the course——Scan projects/ List existing courses for the user to choose from, and confirm the chapter scope

  2. Generate notes——Follow template templates/2-notes.md Generate:

    • Deep summary, not simple extraction
    • Reorganize in your own words
    • Keep key terms and definitions
    • Each knowledge point includes: definition, key points, examples
    • Mark the key points and record places where understanding is difficult

  3. Save——projects/<课程>/2-notes/[原名]-2.md

  4. Handling difficult-to-understand points——For concepts you do not understand, provide: a simple explanation + a concrete example + practical application scenarios

  5. Follow-up——It is recommended to proceed to /study-step3-anki-review, and try to recall with the notes closed

Command: /study-step3-anki-review

Convert chapter knowledge into Anki cards and automatically export an importable TSV file.

Card design principles

  • One card, one point: each card tests only one knowledge point
  • Question first, answer after: the front is the question, the back is the answer
  • Each chapter outputs 10-20 cards

Card types

Type Front Example
Definition What is X? What is the Feynman Technique?
Comparison What is the difference between X and Y? TCP vs UDP
Application How to use X to solve Y? How to use MECE to analyze a problem?
Connection What is the relationship between X and Y? What is the relationship between OKR and KPI?
Further inquiry Why X? Why use dependency injection?
Analogy What is X like? What is the TCP three-way handshake like?

Card format

**Q**: [问题]
**A**: [答案]
---

Group by knowledge point type (e.g., ## 概念类), use --- to separate each card.

Steps

  1. Confirm the course——Scan projects/ List existing courses for the user to choose from, and confirm the chapter scope

  2. Read——Prioritize reading projects/<课程>/2-notes/

  3. Generate cards——Follow template templates/3-review.md

  4. Mark weak points——At the bottom of the document, ## 薄弱点 record concepts you repeatedly forget

  5. Save Markdown——projects/<课程>/3-review/[原名]-3.md

  6. Export TSV——Automatically call the script to generate the Anki import file:

    python3 skills/learning-master/scripts/generate_anki_cards.py "projects/<课程>/3-review/[原名]-3.md"
    • When using this skill in other projects, first locate this skill directory, then call the scripts/generate_anki_cards.py
    • Output file:[原名]-3-anki.txt, saved under the 3-review/anki-export/ directory
    • TSV format:Front内容<TAB>Back内容
    • Markdown will be converted to HTML; physical line breaks within fields become <br>, and tabs become spaces, to avoid breaking the TSV
    • When importing into Anki: allow HTML, delimiter is Tab

  7. Follow-up——Enter /study-step4-output-works

Command: /study-step4-output-works

Turn knowledge into practice. Externalization standard: "able to teach others" or "able to solve practical problems".

Steps

  1. Confirm the course——Scan projects/ List existing courses for the user to choose from, and confirm the chapter range

  2. Read——projects/<课程>/2-notes/ and 3-review/

  3. Recommend output types——Recommend 1-2 specific outputs based on the course content:

    • Technical → code demo / technical blog outline
    • Theoretical → core viewpoints summary / reading notes article framework
    • Skill-based → hands-on checklist / practice plan
    • General → explanation script outline for teaching others

  4. Generate output framework——According to template templates/4-works.md, directly outputan actionable concise framework(not a list of suggestions), kept within 1-2 pages

  5. Save——projects/<课程>/4-works/[原名]-4.md

  6. Follow-up——After completing the course → /study-step5-final-summary

Command: /study-run-full-flow

One complete learning cycle = Preview → Understand → Internalize → Externalize

Steps

  1. Confirm——Course name, chapter range (can specify a single chapter or multiple chapters)

  2. Execute in sequence——Call in order:

    • /study-step1-preview
    • /study-step2-notes
    • /study-step3-anki-review
    • /study-step4-output-works
    • After each stage is completed, confirm whether to continue
    • The user can specify at any time to skip a certain stage

  3. Update progress——Update plan.md Chapter plan table status (see the progress status in the general conventions)

Command: /study-step5-final-summary

After the course study is completed, conduct an overall review and action planning.

Steps

  1. Confirm course completion status——Scan projects/ Confirm the course, check plan.md, and confirm that all core chapters have been completed

  2. Generate summary report——According to template templates/5-summary.md write to projects/<课程>/summary.md：

    • Learning data: total time spent, number of notes, number of cards
    • Core takeaways: the 3-5 most important cognitive models from the whole book
    • Methodology/tools: recommended tools and operating steps
    • Resource quick reference: book list, websites, app list
    • Action plan: short-term (this week/this month) and long-term (3-6 months)

  3. Update plan.md——Mark the summary stage as completed, and add a course completion log

Best practice: Do not list each chapter title; instead, do cross-chapter knowledge synthesis (Synthesis). The purpose of the summary is "what to do next", not "what I read".

Script tools

# Anki 卡片生成（Markdown Q/A → TSV），输出至 3-review/anki-export/
python3 skills/learning-master/scripts/generate_anki_cards.py <review_file_or_dir>

# 指定自定义输出目录
python3 skills/learning-master/scripts/generate_anki_cards.py <review_dir> --output-dir <output_dir>

Reference materials