Claude Code skills for automating everyday content work
Maintained by Baoyu (Jim Liu), baoyu-skills is MIT-licensed, has 20.9k GitHub stars, and is written in TypeScript. It is a Claude Code plugin collection intended to improve everyday content production with AI agents. It covers generating covers and infographics, turning webpages and YouTube into clean Markdown, translation, and one-click posting to WeChat, Weibo, and X.
Each skill is a separate folder containing SKILL.md and reference files. Invoke it with natural language ('make a cover image,' 'convert this URL to markdown'); Claude Code loads the corresponding skill and executes it. Other runtimes, including Codex, are also supported. It is ready to use after installation without configuring each skill individually.
The 21 skills fall into three categories: content generation (10), AI generation (2), and utilities (9). Section 03 below is the complete reference, with a one-sentence purpose and one usage example for each skill. Section 04 shows how to connect several skills into a content workflow.
Acquire content
Organize and translate
Generate visuals
Format
Publish
A collection of skills shared by Baoyu to improve AI agents' everyday work.
One-line installation and natural-language invocation
There are two installation methods: install the whole package with the skills CLI, or register it as a plugin marketplace inside Claude Code. After installation, you do not need to memorize command names. Describe what you want in natural language, and Claude Code loads the corresponding skill.
# Method one: install the whole package with the skills CLI
npx skills add jimliu/baoyu-skills
# Method two: register a plugin marketplace in Claude Code
/plugin marketplace add JimLiu/baoyu-skillsNatural-language skill invocation
Describe the task in natural language, and Claude Code selects the corresponding skill. If you are unsure what is available, see the reference in Section 03, or name the function directly ('make an infographic,' 'translate this article,' 'convert to markdown').
# Generate visuals
Create a cover for this article:./posts/intro.md
Turn these highlights into a vertical infographic
# Fetch and organize content
Convert this URL to clean markdown:https://example.com/post
Translate this article into English in refined modePurposes and usage examples for 21 skills
All 21 skills are listed below in three main categories. Each row gives the skill name, a one-sentence purpose, and a usage example you can adapt directly. Invoke skills with natural language; you do not need to copy the examples word for word. Skills named danger use unofficial APIs at your own risk.
① Content generation · Content (10)
| Skill | Purpose | Usage example |
|---|---|---|
baoyu-cover-image | Article covers with five dimensions (type × palette × rendering × text × mood) | 為這篇文章做封面:./post.md --palette warm |
baoyu-infographic | Professional infographics with 21 layouts × 21 visual styles | 把這份重點做成一張直式資訊圖 |
baoyu-diagram | Publication-ready SVG diagrams from source material (flow, sequence, structure) | 幫這段流程畫一張 SVG 流程圖 |
baoyu-slide-deck | Professional slide images from content, including a complete outline | 把這篇文章做成 10 頁投影片 |
baoyu-comic | Knowledge comics with flexible art style × tone combinations | 把這個概念畫成四格知識漫畫 |
baoyu-article-illustrator | Article illustrations with three dimensions: type × style × palette | 為這篇長文配三張插圖 |
baoyu-xhs-images | Xiaohongshu image-card series with Style × Layout and palette overrides | 做一組小紅書圖卡,5 張,清新風 |
baoyu-post-to-x | Posting to X (Twitter), with images and X Articles | 把這篇整理成 thread 發到 X |
baoyu-post-to-wechat | Posting to WeChat Official Accounts in image-and-text or article mode | 把這篇用圖文模式發到微信公眾號 |
baoyu-post-to-weibo | Posting to Weibo, with text / images / video / headline articles | 把這段配圖發到微博 |
② AI generation · Generation (2)
| Skill | Purpose | Usage example |
|---|---|---|
baoyu-image-gen | AI SDK-based image generation supporting OpenAI / Azure / Google / OpenRouter / DashScope and others | 用 image-gen 出一張封面背景,1024×1024 |
baoyu-danger-gemini-web | Text and image generation through Gemini Web (reverse-engineered API, danger) | 用 gemini-web 生成這張示意圖 |
③ Utilities · Utility (9)
| Skill | Purpose | Usage example |
|---|---|---|
baoyu-youtube-transcript | Downloading YouTube subtitles / transcripts, with language, translation, and chapters | 抓這支 YouTube 的中文逐字稿與章節 |
baoyu-url-to-markdown | Fetching any URL with Chrome CDP and converting it to clean Markdown | 把這個網址轉成乾淨 markdown |
baoyu-danger-x-to-markdown | Converting X (Twitter) content to Markdown (danger) | 把這則 X thread 轉成 markdown |
baoyu-compress-image | Compressing images and reducing file size while retaining quality | 把這張圖壓到 200KB 以內 |
baoyu-format-markdown | Formatting plain text / Markdown with frontmatter and layout | 幫這份草稿加 frontmatter 並排版 |
baoyu-markdown-to-html | Converting Markdown to styled HTML, including WeChat-compatible themes | 把這份 markdown 轉成微信相容 HTML |
baoyu-translate | Translating articles / documents in quick / normal / refined modes | 用精修模式把這篇翻成英文 |
baoyu-wechat-summary | Organizing WeChat group-chat highlights into a structured summary | 把這段群聊整理成今日重點摘要 |
baoyu-electron-extract | Extracting resources and JavaScript from installed Electron apps | 抽出這個 Electron app 的 JS 資源 |
Connecting multiple skills into a content workflow
A single skill is already useful; connecting skills can provide greater benefit. The following are common combined workflows. Each step's output becomes the input to the next, and one sentence is enough to assign the whole workflow to Claude Code.
COMBO 01 · Fully automated blog → WeChat
url-to-markdown fetches the original → translate translates → format-markdown formats → cover-image generates a cover → markdown-to-html (WeChat theme) → post-to-wechat publishes.
Assign the whole workflow in one sentence
COMBO 02 · YouTube → image-and-text thread
youtube-transcript fetches the transcript and chapters → translate → infographic or article-illustrator creates visuals → post-to-x publishes a thread.
Turn video into readable content
COMBO 03 · Xiaohongshu image cards
Give the highlights to xhs-images to generate a set of image cards (Style × Layout), then use post-to-weibo to post to Weibo, or upload to Xiaohongshu manually.
Complete social image cards in one pass
COMBO 04 · Knowledge comics
Give a concept to comic, choose the art style and tone, create a multi-panel knowledge comic, then publish it with post-to-x. Suitable for science communication and teaching.
Easy-to-understand visual explanations of concepts
COMBO 05 · Unified image generation backend
When visual skills (cover, infographic, illustrator) need images, connect image-gen to your chosen provider (OpenAI / Azure / Google / OpenRouter / DashScope), with centralized key management.
One configuration shared by multiple skills
COMBO 06 · Content fetching and organization tools
url-to-markdown, x-to-markdown, and youtube-transcript fetch external content; compress-image and format-markdown finish it, forming your content organization workflow.
Fetch first, then process
A content workflow started with one sentence
The example below gives Claude Code a URL and connects url-to-markdown, translate, cover-image, and markdown-to-html to prepare content ready for a WeChat Official Account. One sentence is enough; Claude Code selects the skills itself, in order.
$ You ›
Turn https://example.com/llm-guide into a Chinese illustrated article,
create a cover, and export HTML ready for WeChat.
ok: → article.md(2,140 characters)
ok: → article.zh.md
ok: → cover-image/llm-guide/cover.png
ok: → article.wechat.html
claude: Claude ›
Done; the four skills have completed:
hl: Original : article.md
hl: Chinese text : article.zh.md
hl: Cover : cover.png
hl: WeChat HTML : article.wechat.html
$ You › Post directly to the Official Account in image-and-text mode
ok: ✓ Draft created; confirm publication in the Official Account dashboard
Describe the task in one sentence, and Claude Code selects the skills and runs them in order.
Goal descriptions and skill selection
Every skill's description explains when it should be used. Describe the result you want, and Claude Code automatically matches and selects the right skill. The same applies to connected workflows: 'turn it into Chinese image-and-text content, make a cover, and output WeChat HTML' is enough to invoke four skills in order.
You do not need to memorize 21 names. Use the reference in Section 03 to understand the available capabilities; when giving instructions, describe the goal in natural language. For steps with side effects, such as posting, it first creates a draft and waits for your confirmation rather than submitting directly.
Usage boundaries and risks
Advanced content workflows with baoyu-skills
After installation, you can build a stable content workflow by fixing styles, creating combinations, and even adding your own skills.
Advanced usage
1. Fixed style and backend settings. Fix the palette, fonts, and image generation backend in each visual skill's configuration (such as cover-image's EXTEND.md) to maintain brand consistency.
2. Combined workflows for recurring tasks. Save common connections (such as blog → WeChat) as a one-sentence prompt template. Paste it later to rerun the whole workflow.
3. Custom skills. baoyu-skills is an open-source template. Follow the same SKILL.md structure to extend it with your own skills and manage them in the same plugin.
4. Sharing across runtimes. These skills also run in runtimes such as Codex. Team members using different agents can share the same capabilities.
5. Official documentation reference. Each skill's references/ contains complete options. The repo README and CHANGELOG are the basis for features and versions.
Priority further reading
① JimLiu/baoyu-skills: README, CHANGELOG, and all 21 skills. ② skills/<name>/SKILL.md in each skill folder: complete options and reference files for that skill. ③ Public material shared by Baoyu (Jim Liu): usage and approaches behind these skills.
21 skills share one purpose: AI handles each step of content work; you decide what you want.