SuperAPI AI Directory
Partner with us中文AI newsFavorites
Skillswechat-toolkit

wechat-toolkit

Writing & research

View repository

Updated 10-15 10:23

About this skill

📦 WeChat Official Account Toolkit (wechat-toolkit)
  • Author: Daniel Li
  • Copyright © Daniel Li. All rights reserved.

Integrates four major functional modules:Search → Download → Rewrite → Publish, covering the entire workflow of Official Account content creation.

Module Overview

Module Function Example Trigger Words
🔍 Search Search Official Account articles by keyword "Search for Official Account articles about XX"
📰 Download Download article content/images/videos "Download this Official Account article"
✍️ Rewrite AI de-tracing + original rewriting "Help me rewrite/adapt this article"
📱 Publish Publish Markdown to drafts "Publish to Official Account"
🔍 Module 1: Article Search

Get a list of Official Account articles through Sogou WeChat search, with support for scraping the body text.

First-time dependency installation

npm install -g cheerio

Usage

# 基础搜索
node {baseDir}/scripts/search/search_wechat.js "关键词"

# 指定数量
node {baseDir}/scripts/search/search_wechat.js "关键词" -n 15

# 保存到文件
node {baseDir}/scripts/search/search_wechat.js "关键词" -n 20 -o result.json

# 解析真实链接
node {baseDir}/scripts/search/search_wechat.js "关键词" -n 5 -r

# 抓取文章正文（自动启用 -r）
node {baseDir}/scripts/search/search_wechat.js "关键词" -n 5 -c

Parameter Description

  • query: Search keyword (required)
  • -n, --num: Number of results to return (default 10, maximum 50)
  • -o, --output: Output JSON file path
  • -r, --resolve-url: Parse the real link of the WeChat article
  • -c, --fetch-content: Scrape the article body content (automatically enables -r)

Output Fields

  • Article title, address, summary, publish time, source Official Account
  • content: Body content (when using -c)
  • word_count: Word count (when using -c)
📰 Module 2: Article Download

Enter an Official Account article link to automatically download the content (Markdown+HTML), images, and videos.

First-time dependency installation

cd {baseDir}/scripts/downloader && npm install

Before downloading: confirm the save location

# 查看当前配置
node {baseDir}/scripts/downloader/download.js --show-config

# 设置默认下载路径（仅需一次）
node {baseDir}/scripts/downloader/download.js --set-output ~/Downloads/wechat-articles
  • "isDefault": true → Not yet configured, need to ask the user
  • "isDefault": false → Already configured, inform the user of the current path

Download article

# 使用默认路径
node {baseDir}/scripts/downloader/download.js "<文章URL>"

# 临时指定路径
node {baseDir}/scripts/downloader/download.js "<文章URL>" --output <临时目录>

# 跳过图片/视频
node {baseDir}/scripts/downloader/download.js "<文章URL>" --no-image
node {baseDir}/scripts/downloader/download.js "<文章URL>" --no-video

Output Structure

<下载目录>/<文章标题>/
├── content/article.html      # 完整 HTML
├── metadata.json              # 标题、作者、时间等
├── images/                    # 所有配图
└── videos/                    # 所有视频/音频

Prerequisites

  • Node.js ≥ 18、Google Chrome、npm install(first time)
✍️ Module 3: Article Rewriting and Adaptation

Rewrite articles into a natural, original style, remove traces of AI writing, and improve originality.

Trigger Words

  • "Help me rewrite this article"
  • "Rewrite it into original content"
  • "Lower the plagiarism check rate"
  • "Remove the AI flavor"

Rewriting Workflow

Standard Rewriting Process

  1. Obtain the original text — Obtain the original text through search (-c scrapes the body) or download
  2. Analyze the structure — Identify the article type, core arguments, and paragraph hierarchy
  3. Deep rewriting — Perform the rewriting according to the following strategies
  4. Add frontmatter — Add title + cover
  5. Publish — Push to the Official Account drafts

Rewriting Strategies

A. Structural Reorganization (core of reducing duplication)

  • Paragraph rearrangement: Adjust the order of paragraphs and disrupt the original structure
  • Paragraph splitting and merging: Split long paragraphs into short ones, or merge fragmented paragraphs
  • Narrative perspective shift: Timeline ↔ Problem-oriented ↔ Comparative analysis ↔ Story introduction
  • Argument reorganization: Keep the core arguments, change the way they are developed

B. Language rewriting (removing AI traces)

  • Delete meaning-inflated sentences: "iconic", "milestone", "far-reaching impact" → replace with specific facts
  • Remove false authority: "experts believe", "the industry generally believes" → state the source or delete
  • Remove pseudo-deep verbs: "enhance capabilities", "empower", "advance the process" → change to specific actions
  • Remove advertising tone: "outstanding", "ultimate experience", "all-around" → objective description
  • Remove high-frequency AI words: empower, closed loop, ecosystem, lever, underlying logic, paradigm, consolidate, momentum
  • Remove filler phrases: in fact, it is worth noting that, overall, it is not hard to find
  • Remove empty endings: "the future is promising", "worth looking forward to" → actual conclusion or action item

C. Title rewriting

Generate 3-5 alternative titles for each article, covering:

  • Question type: Use questions to spark curiosity ("Why is XX still using this method?")
  • Number type: Use numbers to enhance credibility ("3 overlooked XX techniques")
  • Suspense type: Create an information gap ("The truth about XX, 90% of people don't know")
  • Pain point type: Hit the reader's pain point ("Stop XX, try this method")

D. Opening rewriting

Convert the original opening into one of the following styles:

  • Story introduction: Start with a short story or scene
  • Data introduction: Start with striking data
  • Pain point introduction: Directly address the reader's troubles
  • Rhetorical question introduction: Pose a counterintuitive question

E. SEO optimization (optional)

  • Naturally incorporate core keywords into the title and first paragraph
  • Distribute long-tail keywords in subheadings
  • Control keyword density (2%-5%), keep it natural and readable

AI trace identification checklist

After rewriting is complete, check item by item:

# Check item Handling method
1 Meaning-inflated sentences Replace with specific facts
2 False authority references State the source or delete
3 Pseudo-deep verbs Change to specific actions
4 Advertising tone Objective description
5 Template paragraphs (challenge → opportunity → outlook) Delete the template, keep the conclusion
6 Dense appearance of high-frequency AI words Replace with everyday language
7 Abuse of negative parallelism (not only... but also...) Express directly
8 Forced three-part structure Keep the key points, delete filler items
9 Mechanical synonym rotation Use fixed wording for the same concept
10 Abuse of dashes Change to periods or commas
11 Excessive use of bold emphasis Remove unnecessary emphasis
12 List template (X：…） Merge into natural paragraphs
13 Concept-stacking titles Change to conversational titles
14 Emoji overload Delete by default unless a style is specified
15 Chat-speak remnants Delete
16 Knowledge cutoff disclaimer Delete
17 Overly sycophantic tone Respond objectively
18 Filler phrases Delete
19 Overly vague (might, perhaps) Change to conditional judgments
20 Empty conclusion Change to an actual conclusion
21 False range expressions (from...to...) List specific facts

Output format

  1. Full rewritten text(Markdown format, including frontmatter)
  2. Alternative titles(3-5)
  3. Revision notes(Optional, briefly list the main changes)

Judgment criteria

Signs of a successful rewrite:

  • ✅ Reads like it was written by a real person, and can be read aloud smoothly without sounding awkward
  • ✅ No empty sentences, template paragraphs, or "AI-like" smell
  • ✅ High information density, every sentence has concrete content
  • ✅ The structure is clearly different from the original text
  • ✅ Preserves the integrity of the original text's core information
📱 Module 4: Article publishing

One-click publish Markdown to the WeChat Official Account drafts box, based on wenyan-cli.

First-time installation

npm install -g @wenyan-md/cli

Configure API credentials

Make sure the environment variables are set (or configured in TOOLS.md):

export WECHAT_APP_ID=your_wechat_app_id
export WECHAT_APP_SECRET=your_wechat_app_secret

Important: The IP must be in the WeChat Official Account backend whitelist!

Markdown format requirements

At the top of the filemustinclude complete frontmatter:

---
title: 文章标题（必填！）
cover: https://example.com/cover.jpg  # 封面图（必填！）
---

# 正文...

⚠️ title and cover both are indispensable, otherwise an error will be reported.

⚠️ Image paths must use absolute paths, to avoid wenyan path resolution issues. This includes cover and all image references in the body:

cover: /Users/minruiqing/photos/cover.jpg        # ✅ 绝对路径
cover: ./assets/cover.jpg                         # ❌ 相对路径可能出错

![配图](/Users/minruiqing/photos/image.jpg)       # ✅ 绝对路径
![配图](./images/photo.jpg)                       # ❌ 相对路径可能出错

Illustration generation

Before publishing,proactively ask the user whether they need illustrations generated：

📸 The article is ready! Do you need me to help you generate illustrations?

  • Cover image (cover, recommended 1080×864)
  • Body illustrations (generated based on paragraph themes)
  • No need, publish directly

If the user needs illustrations:

  1. Based on the article title and content, generate appropriate image description prompts
  2. Call the user-providedimage generation skill(such as doubao-image, openai-image-gen, etc.) to generate images
  3. Save the generated images to the article directory, usingabsolute pathsfor reference
  4. Set the cover image to the frontmatter's cover field
  5. Insert body illustrations at appropriate positions ![描述](绝对路径)

Prompt suggestions:

  • Cover image: strongly related to the title, concise and impactful, suitable for small-size previews
  • Body illustration: consistent with the corresponding paragraph content, aids understanding

Publishing method

# 方式 1: 使用 publish.js
node {baseDir}/scripts/publisher/publish.js /path/to/article.md

# 方式 2: 直接用 wenyan-cli
wenyan publish -f article.md -t lapis -h solarized-light

# 方式 3: stdin（推荐，解决路径问题）
# macOS/Linux:
cat "/path/to/article.md" | WECHAT_APP_ID=xxx WECHAT_APP_SECRET=xxx wenyan publish -t lapis -h solarized-light

# 方式 4: 含视频文章（必须用这个）
node {baseDir}/scripts/publisher/publish_with_video.js /path/to/article.md

Theme options

Built-in themes：default、lapis(recommended),phycat

Code highlighting：atom-one-dark、dracula、github、monokai、solarized-light(recommended),xcode

wenyan publish -f article.md -t lapis -h solarized-light
wenyan theme -l  # 查看所有主题

Video embedding (critical)

WeChat videos must use the iframe + data-mpvid format,publish_with_video.js this logic is already built in.

Reference in Markdown:

![视频描述](media/video.mp4)   # 自动上传并嵌入

Troubleshooting

Issue Solution
IP not in whitelist curl ifconfig.me → add to the official account backend
wenyan not installed npm install -g @wenyan-md/cli
Environment variable not set export WECHAT_APP_ID=xxx
Missing frontmatter Add title + cover
40001 token invalid Use publish_with_video.js(token management already built in)
Complete workflow example

Search → rewrite → publish

1. 搜索文章：node {baseDir}/scripts/search/search_wechat.js "AI教程" -n 5 -c
2. 选择目标文章，执行洗稿改写
3. 保存为 Markdown（含 frontmatter）
4. 发布：node {baseDir}/scripts/publisher/publish.js article.md

Download → rewrite → publish

1. 下载文章：node {baseDir}/scripts/downloader/download.js "https://mp.weixin.qq.com/s/xxx"
2. 读取下载的 HTML/Markdown，执行洗稿改写
3. 保存为 Markdown（含 frontmatter）
4. 发布：node {baseDir}/scripts/publisher/publish.js article.md

Notes

  • All tools are for personal learning use only, please comply with copyright laws
  • The search function has built-in anti-ban mechanisms (random UA, request delay), please do not use it at high frequency
  • Configuration file: downloader {baseDir}/scripts/downloader/config.json