Search skills/

baoyu-skills Claude skills

Twenty-two skills from Baoyu for content work: illustrated articles, comics, infographics, translation, URL-to-markdown and posting drafts to WeChat, Weibo and X.

JimLiu/baoyu-skills 22 skills Design & frontendMarketingWriting & editing
Our verdict Recommended with caveats

Deep, practical and well tested for content workflows, but many skills need provider keys, a logged-in Chrome, or use reverse-engineered access.

4/5 Depth expertise shipped ! Security Broad access 49d Maintenance Updated 2 mo ago 27K Adoption stars OK License MIT, commercial OK

What baoyu-skills does

baoyu-skills bundles Baoyu's personal content workflow into 22 skills. Image-oriented ones include baoyu-article-illustrator (finds places in an article that need visuals and generates images by type, style and palette), baoyu-cover-image, baoyu-comic, baoyu-infographic, baoyu-xhs-images, baoyu-slide-deck, baoyu-diagram and baoyu-image-gen, which calls one of many providers through an API key (OpenAI, Google, OpenRouter, DashScope, Replicate, MiniMax and others). Text tools cover baoyu-translate, baoyu-format-markdown, baoyu-markdown-to-html, baoyu-url-to-markdown and baoyu-youtube-transcript. Publishing skills (baoyu-post-to-wechat, post-to-weibo, post-to-x) fill content into a browser session and leave the final publish click to you. Two skills carry a danger prefix, baoyu-danger-gemini-web and baoyu-danger-x-to-markdown, because they use reverse-engineered web access and need your consent. Skills average about 15 KB, with roughly 180 reference files, 110 script files and tests; most run through bun or npx. Chinese-language platform support is a particular strength.

Good for

  • Chinese and bilingual content creators publishing to WeChat, Weibo and Xiaohongshu
  • Turning articles into illustrated, formatted posts
  • Batch image generation across many providers

Not for

  • Teams that do not want browser-session or reverse-engineered access
  • Anyone without image-provider API keys

What's inside depth 4/5

22 skills averaging 15 KB with about 180 reference files, about 110 script files and some tests, across a 900-file repo with CI.

Skills
22
Reference docs
253
Scripts
177
Tests / evals
27
Templates
0
Avg SKILL.md
15.1 KB

Counted from the repository file tree. How we score depth.

Dependencies and costs

  • bun or npx
  • Google Chrome with a logged-in session for browser-based skills
  • API keys for the image providers you use
  • WeChat API credentials for the API posting route

Security review

Broad access

This skill asks for more access than most, or sends data to another service. That is usually a design choice rather than bad intent, but you should know before you install it.

  • Uses your browser cookies or logged-in profileLooks for a running Chrome with a remote-debugging port, attaches over CDP, reads the Google/Gemini cookies (including __Secure-1PSID), caches them in a cookie.json in the skill data dir, and sends them only to Google endpoints; otherwise opens a dedicated Chrome profile for you to log in.skills/baoyu-danger-gemini-web/scripts/gemini-webapi/utils/load-browser-cookies.ts:105

A set of content-production skills whose scripts reach out in several ways: baoyu-danger-gemini-web reads your Google/Gemini session cookies from a running Chrome over the debugging port, baoyu-danger-x-to-markdown prefers X token env vars and only otherwise reads X cookies from a dedicated Chrome profile, the post-to-* skills drive a dedicated logged-in Chrome profile to publish for you (they ask before the final submit), and image-gen sends prompts and reference images to whichever provider you pick, with an optional Codex provider that runs with the sandbox off. Scripts run through npx -y bun and install unpinned-range npm packages, and one skill needs the external wx-cli tool to read local WeChat data. Only a portion of the code was read line by line.

Worth knowing

  • browser_session_optionalFallback only: when X_AUTH_TOKEN/X_CT0 env vars are not set, opens a dedicated Chrome profile (not your main one) for you to log in to X, reads the auth_token and ct0 cookies from it via CDP and caches them in a cookie file.skills/baoyu-danger-x-to-markdown/scripts/cookies.ts:104
  • browser_session_optionalpost-to-x, post-to-weibo and post-to-wechat drive a dedicated, skill-specific Chrome profile you log in to once over CDP (real-profile Chrome/Computer Use modes are used only if you ask); they ask for explicit confirmation before the final publish.skills/baoyu-post-to-x/scripts/x-browser.ts:1
  • Relies on a CLI or package we could not reviewThe optional (never auto-selected) codex-cli image provider spawns a nested codex exec --sandbox danger-full-access --skip-git-repo-check agent, i.e. with the Codex sandbox off; you must pin that provider yourself.skills/baoyu-image-gen/scripts/codex-imagegen/spawn.ts:24
  • Runs remote code at an unpinned versionEvery script skill runs via npx -y bun when bun is not installed, and baoyu-electron-extract and baoyu-format-markdown call npx -y @electron/asar, prettier and autocorrect-node at runtime without version pins.skills/baoyu-danger-gemini-web/SKILL.md:35
  • Installs packages or toolsScript folders declare npm dependencies with caret ranges (baoyu-chrome-cdp, baoyu-md, marked, jsdom, defuddle, jimp, socks and others) that are installed when the scripts run.skills/baoyu-post-to-x/scripts/package.json:5
  • Relies on a CLI or package we could not reviewwechat-summary depends on the third-party wx-cli binary, which reads WeChat's local databases and keys; other skills use codex, ssh, agent-browser, Chrome, curl, swift/osascript or xdotool.skills/baoyu-wechat-summary/SKILL.md:44
  • Makes network requestsbaoyu-image-gen sends your prompt and any reference images to the image provider you select (Google, OpenAI, Azure, OpenRouter, DashScope, Z.ai, MiniMax, Jimeng, Seedream, Replicate, Agnes); url-to-markdown and youtube-transcript fetch the pages you give them.skills/baoyu-image-gen/scripts/providers/google.ts:120
  • Needs an API key for its own serviceProvider API keys (GOOGLE_API_KEY, OPENAI_API_KEY and so on) are read from env or .env and sent only to the matching provider's API.skills/baoyu-image-gen/scripts/providers/google.ts:114
  • Makes network requestsIf TELEGRAM_BOT_TOKEN and TELEGRAM_CHAT_ID are set, the WeChat login QR image is sent to that Telegram chat; WeChat API credentials in .env/EXTEND.md go only to WeChat, optionally through an SSH SOCKS tunnel to a server you configure.skills/baoyu-post-to-wechat/scripts/wechat-article.ts:34

What we checked

The repository (921 files) on Oct 29, 2026. Not a full review: 331 of 921 files were not downloaded (large repository; we reviewed the skill folders); we read 11 of 186 code files; it relies on code outside the repository: npm packages baoyu-chrome-cdp and baoyu-md (and other caret-range dependencies) installed for the scripts; wx-cli (github.com/jackwener/wx-cli), codex CLI and agent-browser used by individual skills; 331 of 921 repository files were not available (.claude-plugin manifests, CLAUDE.md, workflows, githooks); 175 of the 186 code files were not opened in full, including most of x-to-markdown, post-to-*, url-to-markdown adapters, and image-gen providers.

NVIDIA SkillSpector static scan: 318 pattern hits (MCP Rug Pull 120, Privilege Escalation 69, Supply Chain 39, Excessive Agency 35, Data Exfiltration 16, YARA Match 13). The CRITICAL YARA credential-exfiltration hit is wechat-article.ts sending the WeChat login QR to a Telegram chat only when the user sets TELEGRAM_BOT_TOKEN and TELEGRAM_CHAT_ID. The info_stealer YARA hits are the CDP cookie readers in the gemini-web and x-to-markdown skills (documented, consent-gated, cookies only used against their own sites). Most remaining HIGH/MEDIUM hits are README env-var tables, prompt text, 'npx' references and provider model names.

A review, not a certification. Skills change after we check them — read the files before installing.

License

MIT — commercial use allowed. Using the danger skills may conflict with the terms of the services they access.

Can it run unattended?

Not really. Most skills involve per-article choices, image review and a manual publish step.

Alternatives to baoyu-skills

md2wechat

geekjourneyx/md2wechat-skill

A CLI plus agent skill that converts Markdown into WeChat Official Account HTML, makes covers and images, and pushes drafts to the WeChat draft box.

With caveats Check first Depth 3/5

Oh Story

zenstory-ai/oh-story-claudecode

13 skills for writing Chinese web novels: ranking scans, teardown of hit books, long and short drafting with file-based continuity tracking, de-AI editing and covers.

With caveats Broad access Depth 4/5

Archify

tt-a1i/archify

A skill that turns a description or a codebase into a validated, interactive standalone HTML diagram or one-page explainer.

Recommended Check first Depth 5/5

LinkedIn Marketing Skills

sergebulaev/linkedin-skills

12 skills that draft LinkedIn posts, comments, replies and profile copy in your voice, read engagement data via Apify, and publish through Publora after you approve.

With caveats Check first Depth 3/5

FAQ

What is baoyu-skills?

It is a set of 22 agent skills from the creator known as Baoyu for content workflows: illustrating and translating articles, generating covers and infographics, converting URLs to markdown, and preparing posts for WeChat, Weibo and X.

How do I install baoyu-skills?

It has a plugin manifest for Claude Code, and the skills can be added with the skills CLI. Most skills run through bun or npx, so one of them must be installed.

Why are some baoyu skills named danger?

baoyu-danger-gemini-web and baoyu-danger-x-to-markdown rely on reverse-engineered web APIs and need your consent. They are flagged so you know they may break or violate a service's terms.

Do baoyu-skills need API keys?

Image generation needs a key for at least one provider such as OpenAI, Google or OpenRouter. Browser-based posting uses a logged-in Chrome session instead, and no API setup.

Sources: GitHub repository (stars, license, file tree, last push Sep 10, 2026). Review based on the README and SKILL.md files as published.