yuque-personal-reading-digest
8
总安装量
5
周安装量
#33800
全站排名
安装命令
npx skills add https://github.com/yuque/yuque-plugin --skill yuque-personal-reading-digest
Agent 安装分布
codex
5
claude-code
5
Skill 文档
Reading Digest â Document Summarization & Reading Notes
Help the user quickly distill a Yuque document into a structured reading digest with key takeaways, core arguments, and actionable insights.
When to Use
- User wants a summary of a long Yuque document
- User says “å¸®ææ»ç»è¿ç¯ææ¡£”, “summarize this doc”, “çæé 读æè¦”
- User wants to extract key points from an article for future reference
- User says “è¿ç¯æç« 讲äºä»ä¹”, “帮ææç¼è¦ç¹”
Required MCP Tools
All tools are from the yuque-mcp server:
yuque_searchâ Find the target document by keywordyuque_get_docâ Read the full document contentyuque_list_reposâ List personal repos to find the save targetyuque_create_docâ Save the reading digest as a new document
Workflow
Step 1: Locate the Document
The user may provide:
- A document title or keyword
- A direct Yuque document URL
- A repo + doc reference
If the user provides a keyword or title:
Tool: yuque_search
Parameters:
query: "<keyword>"
type: "doc"
If the user provides a URL or specific reference, extract the repo_id and doc_id directly.
Step 2: Read the Document
Tool: yuque_get_doc
Parameters:
repo_id: "<namespace>"
doc_id: "<slug>"
If the document is very long, note the total length and proceed with the full content.
Step 3: Generate the Reading Digest
Analyze the document and produce a structured digest:
# ð é
读æè¦ï¼[ææ¡£æ é¢]
> **åæ**ï¼[ææ¡£æ é¢](ææ¡£é¾æ¥)
> **ä½è
**ï¼[ä½è
]
> **é
è¯»æ¥æ**ï¼YYYY-MM-DD
> **é¢è®¡é
读æ¶é´**ï¼çº¦ X åé
---
## ð¯ ä¸å¥è¯æ»ç»
[ç¨ä¸å¥è¯æ¦æ¬ææ¡£çæ ¸å¿è§ç¹æç®ç]
---
## ð å
³é®è¦ç¹
1. **[è¦ç¹ 1]**ï¼[ç®è¦è¯´æ]
2. **[è¦ç¹ 2]**ï¼[ç®è¦è¯´æ]
3. **[è¦ç¹ 3]**ï¼[ç®è¦è¯´æ]
4. **[è¦ç¹ 4]**ï¼[ç®è¦è¯´æ]
5. **[è¦ç¹ 5]**ï¼[ç®è¦è¯´æ]
---
## ð§ æ ¸å¿è®ºç¹ä¸é»è¾
[æ¢³çææ¡£çæ ¸å¿è®ºè¯é»è¾ï¼2-3 段]
---
## ð¡ å¯å䏿è
- [è¿ç¯ææ¡£å¯¹æçå¯å 1]
- [è¿ç¯ææ¡£å¯¹æçå¯å 2]
- [å¯ä»¥åºç¨å°çåºæ¯]
---
## ð åæéå¥
> [æå½ææ¡£ä¸å¼å¾è®°ä½çç²¾å½©æ®µè½ 1]
> [æå½ææ¡£ä¸å¼å¾è®°ä½çç²¾å½©æ®µè½ 2]
---
## ð ç¸å
³å»¶ä¼¸
- [å¯ä»¥è¿ä¸æ¥é
è¯»çæ¹å 1]
- [å¯ä»¥è¿ä¸æ¥é
è¯»çæ¹å 2]
---
> æ¬æè¦ç± AI 婿çæï¼å»ºè®®ç»ååæé
读ã
Step 4: Review with User
Present the digest to the user and ask:
- “æè¦æ¯å¦åç¡®ï¼æéè¦è°æ´çå°æ¹åï¼”
- “è¦ä¿åå°ä½ çè¯éç¥è¯åºåï¼”
Step 5: (Optional) Save to Yuque
If the user wants to save:
Tool: yuque_list_repos
Parameters:
type: "user"
Find or ask for the target repo (often “é 读ç¬è®°” or “读书æè¦”).
Tool: yuque_create_doc
Parameters:
repo_id: "<namespace>"
title: "ð é
读æè¦ï¼[åææ é¢]"
body: "<formatted digest>"
format: "markdown"
Step 6: Confirm
â
é
读æè¦å·²çæå¹¶ä¿åï¼
ð **[ð é
读æè¦ï¼åææ é¢](ææ¡£é¾æ¥)**
ð 已彿¡£å°ï¼ãç¥è¯åºåç§°ã
### æè¦æ¦è§
- æç¼äº X 个å
³é®è¦ç¹
- æå½äº X æ¡åæéå¥
- çæäº X 个延伸é
读æ¹å
Guidelines
- Preserve the author’s original intent â don’t distort or over-simplify
- Key points should be specific and actionable, not vague generalizations
- Include direct quotes from the original for important claims
- The “å¯åä¸æè” section should be personalized â relate to the user’s context if known
- Default language is Chinese; match the original document’s language
- For very long documents (>5000 words), consider breaking the digest into sections matching the original structure
Error Handling
| Situation | Action |
|---|---|
yuque_search returns no results |
Ask user for the exact document URL or repo/slug |
yuque_get_doc fails (404) |
Document may have been deleted; inform user |
yuque_get_doc fails (403) |
User may lack permission; suggest checking access |
| Document is very short (<200 words) | Generate a brief summary instead of full digest template |
| Document is non-text (slides, spreadsheet) | Inform user this skill works best with text documents |