The 12 Skills for Xiaohongshu Operations: Installing Them Doesn’t Mean the Account Will Survive

Someone broke Xiaohongshu operations into 12 Skills, covering research, covers, layout, publishing, distribution, topic diagnosis, and copy generation. The image states it plainly: a one-stop AI toolset from topic selection and creation to publishing.

Here, Skill refers to operating instructions and tool wrappers that can be called by Agents such as Codex and Claude. The list is ordered by operations stage; Star figures are recorded as in the original post and may have changed.

1. Trending-topic integration and cross-platform research: Agent-Reach
   About 107.2k stars. It can search and read content on platforms such as Xiaohongshu, X, YouTube, and Bilibili. Suitable for people who need to track trending topics, find material for topic selection, and observe cross-platform trends. It sits at the information collection layer, used to discover recent hot topics and user discussions.
   Link: https://github.com/Panniantong/Agent-Reach

2. Cover images and image-text cards: baoyu-xhs-images
   The repository it belongs to has about 27.1k stars. Splits an article into 1 to 10 Xiaohongshu covers and content cards. Suitable for knowledge bloggers, tutorial accounts, and AI content creators. Supports multiple visual styles, layouts, color schemes, and watermarks, and can directly generate a series of image-text posts.
   Repo: https://github.com/JimLiu/baoyu-skills

3. Layout, publishing, and account execution: xiaohongshu-mcp
   About 16.7k stars. Log in, search notes, get details, and publish image-text posts and videos. Suitable for people who need AI to actually operate a Xiaohongshu account. It handles the final execution step of the workflow and can connect to Agents such as Codex and Claude to complete uploads and publishing.
   Link: https://github.com/xpzouying/xiaohongshu-mcp

4. Multi-platform distribution: social-auto-upload
   About 16.0k stars. Automatically uploads videos to Xiaohongshu, Douyin, WeChat Channels, Bilibili, TikTok, and YouTube. Suitable for short-video matrices, multi-account, and multi-platform operations teams. After the final video is ready, distribute it in batches to reduce repeated uploading.
   Link: https://github.com/dreammis/social-auto-upload

5. Image-text cards and advanced layout: guizang-social-card-skill
   About 7.9k stars. Generates Xiaohongshu carousel images, knowledge cards, and WeChat Official Account cover pairs. Suitable for content creators who value layout quality and visual consistency. Built-in visual systems such as editorial style and Swiss style, offering 28 layouts and 10 theme sets.
   Link: https://github.com/op7418/guizang-social-card-skill

6. Publishing, engagement, and account operations: XiaohongshuSkills
   About 3.5k stars. Search, publish, comment, and automate engagement. Suitable for Codex, Claude Code, and OpenClaw users. Wraps common Xiaohongshu operations into Agent Skills; suitable as an execution tool for automated operations.
   Link: https://github.com/white0dew/XiaohongshuSkills

7. Topic strategy and data diagnosis: xiaohongshu-ops-skill
   About 2.5k stars. Recommendation feed analysis, account diagnosis, topic generation, viral post breakdowns, and data review. Suitable for operators who already have an account but whose topic selection is scattered or data is unstable. Combines account positioning, historical performance, and platform content to generate topics and identify reasons for data differences.
   Repo: https://github.com/Xiangyu-CAS/xiaohongshu-ops-skill

8. Copy generation and full-chain creation: Auto-Redbook-Skills
   About 2.4k stars. Generates titles, body copy, covers, and content cards, and completes publishing. Suitable for beginners who want to quickly get through the Xiaohongshu image-text creation workflow. It basically covers everything from topic input to final publishing.
   Skill: https://github.com/comeonzhj/Auto-Redbook-Skills

9. Search and collection, publishing and engagement: xiaohongshu-skills
   About 2.0k stars. Search notes, retrieve content, publish works, and engage via comments. Suitable for users who need a basic set of Xiaohongshu automation capabilities. It is more at the platform operations layer and can be combined with topic selection, copywriting, and visual Skills.
   Link: https://github.com/autoclaw-cc/xiaohongshu-skills

10. Cover strategy and visual direction: xhs-visual-director-skill
    About 1.5k stars. Determines visual direction, number of image-text pages, cover structure, and page-by-page image plan. Suitable for people who can write content but cannot plan covers and image-text visuals. First analyzes the content task, then outputs covers, page-by-page visuals, drawing prompts, and a pre-publish self-check checklist.
    Skill: https://github.com/ziguishian/xhs-visual-director-skill

11. Cover design and character consistency: gbro-cover-design
    About 937 stars. Generates 3:4 Xiaohongshu cover compositions and drawing prompts. Suitable for accounts that need covers with real people on camera and consistent character appearance. Confirms the theme, style, and character requirements through step-by-step questions, then outputs a cover plan.
    Link: https://github.com/pyang5166/gbro-cover-design

12. Competitor research and note collection: xhs-cli
    About 675 stars. Search, read, like, save, comment, and retrieve Xiaohongshu content. Suitable for people doing competitor analysis, viral post collection, and engagement research. It can serve as a data collection tool before topic strategy, passing popular notes to analysis Skills for further processing.
    Link: https://github.com/jackwener/xhs-cli

A netizen added a collection: https://github.com/UnderleveledBuilder/awesome-social-media-skills . The author replied that they mainly work on Xiaohongshu and just collected these themselves.

There was also a comment in the comments section: no matter how complete the toolchain is, the bottleneck is risk control. A netizen said that Xiaohongshu accounts batch-published by mcp won't survive two weeks. Topics that can pass moderation are even scarcer. The author also replied that in practice they still use manual supervision because they don't feel at ease.

Another netizen gave workflow-building advice: several of these can publish, and when actually building a workflow, choosing one main tool for each step is enough. Give each draft an ID and carry it through collection, image generation, and publishing; checking duplicates and reviewing sources will be much clearer.

Someone asked the author whether they make money from this. The author replied that they mainly work on Xiaohongshu, currently take brand deals, and have nothing else. Another netizen said they had installed a row of Skills, but their account still had zero followers. The author posted a screenshot of their account to prove it: 11.7k followers, 88k likes and saves.

This list is more like a parts library. First pick one main pipeline and get it working: Agent-Reach for collection, baoyu-xhs-images or guizang-social-card-skill for images, xiaohongshu-mcp for publishing, and xiaohongshu-ops-skill for review. Add social-auto-upload for multi-platform distribution. Keep manual checks in the publishing step; don't just hand the account to automation and walk away. Whether a topic can pass moderation, whether anyone will click the cover, and why the data won't take off—these judgments still have to be made by yourself.