Token导航 LogoToken导航TokenDH.com
研究检索执行命令github未标认证来源可访问许可证需确认审计通过

study-notes-generator学习笔记生成器

Agent Skill

study-notes-generator 用于查找、检索和筛选相关信息,适合在 Codex、Claude、Cursor、Gemini CLI 中需要根据关键词、任务场景或来源线索快速定位候选结果时使用。可结合来源仓库、安装命令和原始 README 继续核验具体用法。安装前建议确认权限范围、维护状态,以及是否会触发联网、命令执行或文件读写。

总安装

225

周安装

9

GitHub Stars

2

下载量

73
CodexClaudeCursorGemini CLI

安装说明

本站只整理中文说明和来源信息,不托管安装包,也不代用户安装。

GitHub

来源数

2

许可证

unknown

最后核验

2026-05-01

来源状态

来源可访问

安装方式

通过对话安装

复制提示词发给支持本地命令或 Skills 的 AI 助手,先确认命令和权限,再让它执行。

请帮我安装这个 Agent Skill:study-notes-generator(学习笔记生成器)
来源仓库:https://github.com/touricks/fanshi_personal_skills
仓库路径:skills/study-notes-generator
安装命令:
npx skills add https://github.com/touricks/fanshi_personal_skills --skill study-notes-generator
安装前请先检查当前环境是否支持对应 CLI,并向我确认将要执行的命令、安装目录、联网范围和文件读写权限;确认后再执行。

命令行安装

复制命令到本机终端执行。该命令会通过 npx skills 从第三方来源获取 Skill;本站只展示命令,不托管安装包,也不自动执行。

skills.shnpx skills
npx skills add https://github.com/touricks/fanshi_personal_skills --skill study-notes-generator

简介

study-notes-generator 用于查找、检索和筛选相关信息。

  • 适合在 Codex、Claude、Cursor、Gemini CLI 中根据关键词、任务场景或来源线索快速定位候选结果时使用。
  • 可结合来源仓库、安装命令和原始 README 继续核验具体用法。
  • 安装前建议确认权限范围、维护状态,以及是否会触发联网、命令执行或文件读写。
  • 安装命令:npx skills add https://github.com/touricks/fanshi_personal_skills --skill study-notes-generator

SKILL.md

Study Notes Generator

Transform documents from slides/ into feature-rich HTML study notes saved to notes/.

Workflow

  1. Read the source document from slides/ (see PDF Handling below for large files)
  2. Extract and organize content into learning-focused sections
  3. Generate HTML using the template structure from assets/template.html
  4. Save to notes/ with filename {original-name}-notes.html

PDF Handling (CRITICAL)

The Claude API has a hard limit of 100 images/documents per conversation. PDF pages are converted to images when read, so any PDF over ~90 pages will fail.

For PDFs over 50 pages:

  1. Always extract text first using pdftotext: /opt/homebrew/bin/pdftotext "slides/filename.pdf" "slides/filename.txt"
  2. Generate notes from the .txt file only
  3. Do NOT read the PDF file directly — only read the extracted .txt

For PDFs under 50 pages:

  • Read the PDF directly with the Read tool (using pages parameter for 10+ pages)

Parallel Processing (Multiple Documents)

When processing multiple documents at once:

  1. Validate with one agent first: Spawn 1 background Task agent, wait for it to succeed, then spawn the rest. This avoids wasting tokens on a batch of agents that all fail the same way.
  2. Use Task tool with run_in_background: true for independent per-file agents. Do NOT use TeamCreate — team agents require message-driven coordination and will idle without producing output when tasks are independent.
  3. Batch size: Spawn at most 4 agents in parallel to avoid rate limits. If more files remain, wait for the first batch to complete before launching the next.
  4. Strong negative constraints in agent prompts: When instructing agents, use explicit prohibitions:

- GOOD: "Do NOT read any PDF files. ONLY read the.txt file." - BAD: "Prefer reading the text file over the PDF." Agents ignore weak preferences but obey strong negations.

Content Extraction Guidelines

When analyzing the document, identify and extract:

Key Concepts & Definitions

  • Core terminology and their meanings
  • Fundamental principles being taught
  • Mark with <div class="concept"><strong>Term</strong>Definition...</div>

Problem-Solution Pairs

  • What problem/challenge does each concept address?
  • How does the method/technique solve it?
  • Mark with:
<div class="problem-solution">
  <div class="problem">Problem: What challenge exists?</div>
  <div class="solution">Solution: How this concept/method addresses it</div>
</div>

Examples & Applications

  • Concrete examples from the document
  • Real-world applications mentioned
  • Mark with <div class="example">...</div>

Important Notes

  • Warnings, common mistakes, or critical points
  • Mark with <div class="important">...</div>

CRITICAL: Mandatory Examples Rule

For ANY content that is NOT a pure definition, you MUST provide at least one concrete example.

What requires examples:

Content TypeExample Required?Reason
Pure definition (e.g., "X is defined as Y")OptionalDefinition is self-contained
Properties/CharacteristicsYESNeed to show how property manifests
Theorems/ClaimsYESNeed to illustrate what theorem means
Comparisons (X vs Y)YESNeed concrete case showing difference
Security propertiesYESNeed attack scenario or counterexample
Algorithms/ProceduresYESNeed step-by-step walkthrough
"Why" questionsYESNeed counterexample or motivation case

Example format:

<div class="concept">
    <strong>Collision Resistance (抗碰撞性)</strong>
    Nobody can find x and y such that x ≠ y and H(x) = H(y).
</div>

<div class="example">
    <strong>Why this matters:</strong> Consider H(x) = last 4 bits of x.
    <br>Finding collision is trivial: H("00001010") = H("11111010") = "1010"
    <br>This H is NOT collision resistant - bad for security!
</div>

Counterexample pattern for "Why not X?" questions:

When explaining why a definition/property is formulated a certain way, ALWAYS provide a counterexample showing what goes wrong with the alternative:

<div class="problem-solution">
    <div class="problem">Why require H(x₁)=Y rather than x₁=x in one-wayness?</div>
    <div class="solution">
        <strong>Counterexample:</strong> Let H(x) = LSB(x) (last n bits).
        <ul>
            <li>Given Y, can you find the original x? NO (2^n possibilities)</li>
            <li>Given Y, can you find ANY x' with H(x')=Y? YES, trivially!</li>
        </ul>
        If we used x₁=x, this insecure function would be called "one-way"!
    </div>
</div>

HTML Structure

Generate HTML following this structure:

<!DOCTYPE html>
<html lang="en">
<head>
    <meta charset="UTF-8">
    <meta name="viewport" content="width=device-width, initial-scale=1.0">
    <title>{Document Title} - Study Notes</title>
    <!-- Copy full <style> block from assets/template.html -->
</head>
<body>
    <h1>{Document Title}</h1>

    <nav class="toc">
        <h2>Table of Contents</h2>
        <ul>
            <li><a href="#section-1">Section 1 Title</a></li>
            <!-- ... more sections -->
        </ul>
    </nav>

    <section id="section-1">
        <h2>Section 1 Title</h2>
        <!-- Content with concepts, problem-solutions, examples -->
    </section>

    <!-- More sections -->

    <!-- End with summary and quiz -->
    <div class="summary">
        <h3>Key Takeaways</h3>
        <ul>
            <li>Main point 1</li>
            <li>Main point 2</li>
        </ul>
    </div>

    <section>
        <h2>Self-Check Quiz</h2>
        <!-- Quiz questions -->
    </section>
</body>
</html>

Collapsible Sections

Use <details> for content that benefits from progressive disclosure:

<details>
    <summary>Detailed Explanation of X</summary>
    <div>
        Extended content here...
    </div>
</details>

Self-Quiz Questions

End each major topic or the document with quiz questions:

<div class="quiz">
    <div class="quiz-question">What is the main purpose of X?</div>
    <details>
        <summary>Show Answer</summary>
        <div>The answer explanation here...</div>
    </details>
</div>

Generate 3-5 quiz questions per major section covering:

  • Definitions (What is...?)
  • Application (When would you use...?)
  • Comparison (How does X differ from Y?)
  • Problem-solving (Given scenario Z, what approach...?)

Quiz answers should also include examples when explaining non-trivial concepts.

Output Checklist

Before saving, verify the notes include:

  • Clear title and table of contents
  • All major topics from source document
  • Key concepts highlighted with definitions
  • Concrete examples for ALL non-definition content (MANDATORY)
  • Problem-solution pairs explaining the "why"
  • Counterexamples for "why not X?" type questions
  • Collapsible sections for detailed content
  • Summary of key takeaways
  • Self-quiz questions with hidden answers (including examples in answers)

Resources

assets/

  • template.html - Full HTML template with CSS styles. Copy the <style> block into generated notes.

适合场景

01

用户想查找某类 Agent Skill 时

02

需要根据任务场景推荐可安装能力包时

03

需要对比不同来源的安装命令和来源信息时

能力概览

能力 1

按任务关键词查找相关 Skills

能力 2

展示可复制的安装命令

能力 3

保留来源站点、仓库和原始说明,方便继续核验

能力 4

展示第三方安全扫描或审计结果

安装后应在对应宿主中按原始 README 的触发条件使用;具体调用方式请以来源页面和 README 为准。

平台分布

Codex

36.05%
按下载量换算26

Claude

29.45%
按下载量换算21

Cursor

18.79%
按下载量换算14

Gemini CLI

9.14%
按下载量换算7

安全审计

Gen Agent Trust Hub

通过

Socket

通过

Snyk

通过

权限和风险

执行命令

安装流程涉及命令执行,可能通过 npx skills add https://github.com/touricks/fanshi_personal_skills --skill study-notes-generator 联网下载 Skill 或依赖。用户安装前应确认命令来源、仓库内容和执行环境。

安装前确认

本站仅展示第三方公开信息,不托管安装包,不提供自动安装或运行环境。安装前应自行审查源码、依赖和命令行为。当前只有一个来源,正式发布前建议补源仓库或其他目录站核验。

来源信息

继续浏览同类 Skills