name: WeChat Article Summarize description: Read one or more WeChat public account article links from mp.weixin.qq.com, extract cleaned full text and optional image links, summarize each article in Chinese with summarize, and generate a structured markdown file saved to a user-chosen directory. Use when the user shares WeChat article URLs and wants single-article notes, multi-article daily reports, article summaries, image extraction, or a structured markdown digest. Before processing, first confirm summarize is configured and working, ask whether to include images, and ask where to save the final file.

WeChat Article Summarize

把一個或多個微信公眾號文章連結整理成結構化 markdown，支援單篇整理和多篇日報彙總。

功能簡介

• 讀取一個或多個 mp.weixin.qq.com 文章連結
• 抽取文章正文、標題、釋出時間，以及可選的圖片連結
• 自動修復常見的微信正文亂碼問題
• 呼叫 summarize 用中文總結全文內容
• 生成結構化 markdown 檔案
• 單篇文章整理
• 多篇文章彙總 / 日報
• 支援按日期 + 標題，或日期 + 篇數 + 彙總說明命名
• 支援把檔案儲存到使用者指定目錄

使用前需要確認

在真正開始抓取文章之前，需要先確認：

1. summarize 已經配置好 API key，並且可正常使用
2. 是否需要在最終 markdown 中保留圖片連結
3. 最終檔案儲存到哪個目錄

適用場景

• 總結單篇微信文章
• 把多篇微信文章彙總成一份日報
• 輸出適合繼續閱讀、歸檔或二次整理的 markdown 檔案

Workflow

Step 0: Confirm prerequisites before fetching anything

Do not fetch article content until all three items are clear:

1. summarize is ready
2. Ask the user to configure summarize API access first if needed.
3. Verify summarize by running a tiny Chinese test.

4. Proceed only if summarize returns a usable summary.

5. Image preference

   想要更強大的技能外掛，就來小蔥技能站7w4.net看看吧。

6. Ask whether the final markdown should include image links.

7. Map user intent to include_images=true|false.

8. Output directory

9. Ask where to save the final markdown file.
10. If the user says “下載資料夾”, use ~/Downloads.
11. Create the target directory if it does not exist.

If any of the three items is missing, stop and ask before continuing.

Step 1: Extract each WeChat article

For each mp.weixin.qq.com URL, run:

python3 scripts/read_wechat_article.py '<wechat_url>' --out '<temp_dir>'

This produces structured metadata, raw HTML, and a first-pass markdown export.

Step 2: Clean the body text

Do not trust the first-pass article markdown blindly.

If the body contains mojibake or obvious encoding corruption, repair it from raw.html by running:

python3 scripts/fix_wechat_body.py '<raw.html>' --out '<body-fixed.txt>'

Use the cleaned body text as the canonical input for summarization.

Step 3: Summarize in Chinese

Always summarize the cleaned local text, not the original WeChat URL.

Run:

python3 scripts/summarize_cn.py '<body-fixed.txt>' --out '<summary.json>' --length short

or for a combined report:

python3 scripts/summarize_cn.py '<combined-input.md>' --out '<summary.json>' --length medium

The script enforces Chinese output and fails if the returned summary is not sufficiently Chinese.

Step 4: Normalize summary text before writing markdown

Never write summarize output directly into the final file.

Normalize paragraph breaks and spacing with:

python3 scripts/normalize_markdown_text.py '<input.txt>' --out '<normalized.txt>'

Use this for: - each single-article summary - the combined daily-report overview

This prevents ugly line wrapping and mixed-language formatting artifacts.

Step 5: Build the final markdown

Single article

Run:

python3 scripts/build_mindmap_markdown.py \
  --result '<result.json>' \
  --body '<body-fixed.txt>' \
  --summary '<summary.json>' \
  --output-dir '<chosen-dir>' \
  --include-images true

Multiple articles / daily report

Run:

python3 scripts/build_batch_report.py \
  --inputs '<dir1>' '<dir2>' '<dir3>' \
  --output-dir '<chosen-dir>' \
  --include-images true \
  --report-label '微信文章日報'

The batch report must: 1. summarize all articles individually 2. summarize the full set as one combined overview 3. place the combined overview first 4. then append each single article section

Output rules

Naming

Single article

YYYYMMDD-文章標題.md

Multiple articles

YYYYMMDD-<總文章數量>篇-<彙總說明>.md

Content rules

Single article output should contain

• title
• source URL
• publish time
• summarize-generated Chinese summary
• mindmap-style structure
• optional image section

Batch report output should contain

• combined daily overview at the top
• combined mindmap
• per-article title, URL, date, summary, and mindmap
• optional image overview

Non-negotiable quality gates

Before writing the final markdown:

1. Summary language check

2. If the summary is not mainly Chinese, retry or fail.

3. Paragraph normalization

4. Collapse unnatural line breaks inside prose.

5. Keep markdown headings and bullet lists intact.

6. Clean body source

7. Prefer repaired text from raw.html when the extracted body is corrupted.

Bundled scripts

• scripts/read_wechat_article.py — fetch WeChat article metadata, body, raw HTML, and image links
• scripts/fix_wechat_body.py — repair mojibake and extract clean text from raw HTML
• scripts/summarize_cn.py — run summarize in Chinese and enforce a language check
• scripts/normalize_markdown_text.py — normalize prose paragraphs and line breaks
• scripts/build_mindmap_markdown.py — generate single-article markdown files
• scripts/build_batch_report.py — generate multi-article combined reports
• scripts/run_wechat_mindmap_workflow.py — orchestrate the full workflow end to end after the required user confirmations

這個 Skill 能把微信公眾號文章自動整理成帶中文摘要和結構化格式的筆記，支援單篇整理和多篇彙總，適合需要快速歸檔和閱讀微信文章的使用者。整體工作流程設計完整，亂碼處理和封鎖檢測等常見問題都有覆蓋，質量基本可靠。主要不足是生成的內容結構相對模板化，思維導圖部分比較簡略，實用性有待提升。整體質量中等偏上，核心功能紮實，但細節打磨還需加強。