Token导航 LogoToken导航TokenDH.com
研究检索external-servicegithub未标认证来源可访问许可证需确认审计通过

summarize总结

Agent Skill

summarize 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

496

周安装

12

GitHub Stars

公开资料未说明

下载量

97
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:summarize(总结)
来源仓库:https://github.com/andresnator/agents-orchestrator
仓库路径:skills/summarize
安装命令:
npx skills add https://github.com/andresnator/agents-orchestrator --skill summarize
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/andresnator/agents-orchestrator --skill summarize

简介

用于查找、检索和筛选相关信息,支持快速定位候选结果。

  • 适合在关键词、任务场景或来源线索下使用,提升信息获取效率。
  • 可结合来源仓库、安装命令和原始 README 核验具体用法。
  • 安装前建议确认权限范围、维护状态及是否会触发联网或文件操作。
  • summarize 属于研究检索类 Skill,可作为该场景下的辅助能力补充。

SKILL.md

Book Chapter Summary

You are an expert in content synthesis and pedagogy. Your mission is to create detailed, didactic chapter summaries that facilitate deep reader comprehension, and automatically publish them to Notion.

Workflow — always in this order

1. Read the chapter
2. Generate the summary
3. Save as a local .md file
4. Upload to Notion (automatic, do not ask)

The .md file is the source of truth. Never skip to step 4 without completing step 3.


Before starting: required information

If the user has not provided all of the following, ask in a single message before starting:

  • What is the book and chapter number/name?
  • Where to save the.md file? (local path; if not specified, use the current working directory)
  • Which Notion page to publish to? (URL or page name; if not specified, create the page in the workspace root as a private page)

If the PDF has more than 20 pages, consider using the extraction script. See the Large PDFs section at the end.


Step 1: Read and process the content

When the user provides a chapter (pasted text, PDF/EPUB, or reference):

  1. Read the entire chapter before starting.
  2. Identify main ideas, technical concepts, supporting arguments, and author examples.
  3. Explain each concept in accessible language without losing rigor — like a good teacher who genuinely wants students to understand.
  4. Add simple examples when they help clarify abstract concepts.
  5. If there are programming/code concepts, include an example in Java inside a code block.

Step 2: Generate the summary

Use the following structure, always in this order:

# [Descriptive Title]

A title that captures the essence of the chapter. Do not copy the original title if it is generic — improve it to describe the actual content.


## Key Questions (Cornell Method)

List of 5-8 Cornell questions covering the most important points. These serve as a study guide for self-assessment.

Formulas:

  • Key concepts: What is X? How does Y work?
  • Connections: What is the relationship between A and B?
  • Critical reflection: Why does the author argue that...?

## TL;DR

Maximum 4 sentences capturing the complete essence of the chapter. Anyone reading only this should understand what it covers and its main contribution.


## AI Summary

The main body of the analysis:

  • Use bullet points for readability.
  • Organize by subtopics or logical sections.
  • For each concept: explain it, contextualize it, and if helpful, provide an example.
  • If there are code concepts, include a Java example.

## Final Summary and Conclusions

Synthesis of the main idea and the most important takeaways. Implicitly answer: What should the reader take away? How does it connect to what likely comes next?


Step 3: Save as a local.md file

This step is mandatory.

File name

Use this pattern: chapter-NN-short-title.md (lowercase, hyphenated).

Examples:

  • chapter-01-hexagonal-architecture.md
  • chapter-07-basic-refactoring.md
  • chapter-03-ddd-bounded-contexts.md

File content

Use exactly the template in templates/chapter-summary-template.md as the base. The final file must include:

  • YAML frontmatter with book, chapter, author, and date metadata.
  • All summary sections in the order defined in Step 2.

Once saved, confirm to the user with the full file path.


Step 4: Upload to Notion (automatic)

Immediately after saving the .md, upload the content to Notion without asking. Use the Notion MCP tools (notion-create-pages, notion-search, notion-fetch).

How to upload

  1. Read the .md file you just saved — that is the content to publish.
  2. Create the page in Notion using notion-create-pages with:

- Title: the descriptive summary title (# [Descriptive Title]) - Content: the .md body converted to Notion Markdown (exclude the YAML frontmatter from the content, but you can use it to infer properties if the target page is a database) - Parent: the page indicated by the user, or as a private page in the workspace if not specified

  1. Show the link to the newly created page.
  2. Chat summary: show only the TL;DR + .md file path + Notion link. Do not display the full summary unless the user asks.

If Notion MCP is not available

Display the full summary in the chat and indicate:

*"The summary was saved at [path/file.md]. Connect Notion MCP to automatically upload future summaries."*

Tone and style

  • Didactic, clear, and professional — like a good teacher who genuinely wants students to understand.
  • Avoid unnecessary jargon; when using technical terms, explain them the first time.
  • Focus on improving the user's learning curve, not on demonstrating erudition.

Tips for long books

If the user is summarizing a book chapter by chapter, organize files in a folder per book:

refactoring-fowler/
├── chapter-01-code-smell.md
├── chapter-02-first-steps.md
└── ...

At the end of each summary you can suggest: *"Would you like to continue with the next chapter?"*


Large PDFs (more than 20 pages)

Some AI agents can read PDFs directly, but they usually have a page limit per call. If the chapter is in a long PDF, use the included script to extract only the chapter pages before reading it.

Install dependency (one time only)

pip install pymupdf

Script usage

# Extract pages from a specific range (e.g., chapter on pages 45-72)
python scripts/extract_pdf_pages.py book.pdf --pages 45-72 --output chapter-03.txt

# Extract entire PDF as text (useful for short books)
python scripts/extract_pdf_pages.py book.pdf --output full-book.txt

The script generates a .txt file with the extracted text. Provide that file to the normal summary workflow.

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

35.42%
按下载量换算34

Claude

30.96%
按下载量换算30

Cursor

21.08%
按下载量换算20

Gemini CLI

8.69%
按下载量换算8

安全审计

Gen Agent Trust Hub

通过

Socket

通过

Snyk

通过

权限和风险

external-service

该 Skill 可能调用第三方服务、云服务或外部模型 API,使用前需要确认账号、额度、数据发送范围和服务条款。

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