Guide · Agent skill
How to use the Xiaohongshu Image Card Series agent skill
Updated · written with AI from the project's own documentation and checked against our catalogue data
baoyu-xhs-images is an agent skill from the JimLiu/baoyu-skills collection that turns a piece of content into a series of infographic-style image cards. It is aimed mainly at Chinese social platforms such as Xiaohongshu (RED) and WeChat posts. This guide explains what the skill does, how to install it, how to ask for a series, and what you need before it will produce anything.
Xiaohongshu Image Card SeriesJimLiu/baoyu-skills · MIT Download the skill Quick start
- 1Download and unzip. Download the skill ZIP from our page and unzip it into ~/.claude/skills/ or a project's .claude/skills/ folder, or upload it under Settings → Capabilities → Skills in the Claude apps.
- 2Confirm an image backend. Make sure you have Codex imagegen, Cursor GenerateImage, or the baoyu-image-gen skill available. Without one, the skill will ask rather than generate.
- 3Make your request. Ask in plain language, using a trigger phrase such as 小红书图片 or image cards, and include your content and card count if you have them.
- 4Answer the confirmation questions. Choose or confirm the style, layout, palette and number of cards. The agent will not generate until you confirm.
- 5Review the prompt files. Check the saved prompts under prompts/ so you know what each card is meant to show.
- 6Check each card. Review every rendered card for wrong or altered text and numbers before you publish.
What the skill makes the agent do
Once installed, the agent can take a piece of content and split it into a series of 1-10 image cards in a cartoon-style look. You can choose from 12 visual styles, 8 layouts and 3 colour palettes (macaron, warm and neon).
Before generating, the agent asks you to confirm the plan. It then writes a saved prompt file for each card under prompts/ and asks an image generation tool to render the pictures.
- Splits content into 1-10 cards
- 12 visual styles, 8 layouts, 3 palettes
- Saves a prompt file per card before rendering
- Waits for your confirmation before generating
What you need before you start
The skill does not draw images itself. It needs an image generation backend. The source names Codex imagegen, Cursor GenerateImage, or the baoyu-image-gen skill as options. If the agent cannot find a raster image tool, it will stop and ask you how to proceed rather than producing SVG or HTML substitutes.
The codex CLI with an active login is optional and applies to the codex-imagegen route. An optional EXTEND.md preferences file can store your defaults, such as a preferred backend, language, watermark or batch size.
- An image generation backend (Codex imagegen, Cursor GenerateImage, or baoyu-image-gen)
- Optional: codex CLI with an active codex login, for the codex-imagegen route
- Optional: EXTEND.md preferences file for saved defaults
- Network access (the skill uses the network)
Install the skill from the ZIP or the Claude apps
Download the ZIP for this skill from our page and unzip it into ~/.claude/skills/ for a personal install, or into a project's .claude/skills/ folder to limit it to one project.
Alternatively, if you use the Claude apps, open Settings, go to Capabilities, then Skills, and upload the skill there.
- Personal: unzip into ~/.claude/skills/
- Per project: unzip into .claude/skills/ inside the project
- Claude apps: Settings → Capabilities → Skills, then upload
Trigger the skill with a natural request
The skill's description lists trigger phrases. You can name the skill directly or use one of the Chinese or English phrases from its description, or simply say you want a social media infographic series.
If you do not give a style, layout or palette, the agent will ask you to choose during its confirmation step. You can also pass options such as --style, --layout, --palette, --preset, --ref and --batch-size in the request.
- "Use baoyu-xhs-images to make 5 image cards from this article"
- "帮我做一组小红书图片卡片"
- "Make a WeChat image card series from these notes, macaron palette"
- "Turn this checklist into an image card series for social media"
Review the plan and the generated cards
Default behaviour is to confirm before generation. The skill only skips confirmation if you explicitly say so, for example with --yes or 直接生成. When generating in batches, the agent saves every prompt file first, then renders images, with image 1 used as the reference for the rest of the series.
Generated text on cards can be wrong. The skill forbids fixing text by painting over an image with tools such as ImageMagick, Pillow or SVG. If text is wrong, the agent regenerates from a corrected prompt, switches to a layout with less on-card text, or asks you which candidate to keep.
- Check every card for misspelled or altered text and numbers
- Expect to regenerate rather than edit cards by hand in the skill
- Keep the prompts/ folder if you want a reproducible record
Safety notes: scripts and network
According to the skill's listing, it does not run scripts of its own, but it does use the network. Image rendering goes through your chosen backend, so the content you submit may be sent to that service. Check the privacy terms of whichever backend you connect.
Review the files in the skill folder if you need to know exactly what instructions the agent is following. The folder includes SKILL.md and reference files covering presets, palettes, elements, typography and watermark setup.
- Runs scripts: no (per the listing)
- Uses network: yes
- Content goes to your image backend; review its terms
What people use it for
- Turning a long Xiaohongshu note into a 5-card product recommendation series
- Converting a WeChat article into a set of shareable image cards
- Producing a consistent cartoon-style series for a marketing campaign on social media
Frequently asked questions
Is baoyu-xhs-images free to use?
The skill's source is published under the MIT licence. Image generation depends on the backend you connect, such as Codex imagegen, Cursor GenerateImage or baoyu-image-gen, and each has its own access terms.
What do I need to install baoyu-xhs-images?
Download the skill ZIP and unzip it into ~/.claude/skills/ or a project's .claude/skills/ folder, or upload it in the Claude apps under Settings → Capabilities → Skills. You also need an image generation backend.
Does the skill generate images by itself?
No. It plans the cards and writes prompts, then asks an image generation backend to render them. If no raster image backend is available, it stops and asks you how to proceed.
Can I use it without any image generation tool?
Not for its intended output. The skill is designed not to substitute SVG, HTML or canvas art for raster images, so without a backend it will ask you instead.
Why is there text wrong on my cards?
Image models can misspell or alter text and numbers. The skill forbids covering text with overlays, so the fix is to regenerate from a corrected prompt, use a layout with less text, or pick a different candidate.
How does it compare with other image tools for social posts?
This skill is built mainly for Chinese social platforms such as Xiaohongshu and WeChat, and it organises content into a card series with styles, layouts and palettes. For brand-critical or print artwork, or where exact text and data must be verified, a manual design tool is a better fit.
New agent skills, repos and AI tools — every week
The best new skills and open-source projects, the fastest-growing repos, and one free prompt. Five minutes to read. Unsubscribe anytime.
Our verdict
Use baoyu-xhs-images if you make multi-image Xiaohongshu or WeChat posts and already have an image generation backend such as Codex imagegen, Cursor or baoyu-image-gen. Pick something else for print or brand-critical artwork, or for any card where exact text and numbers must be correct without review.
Download the skill