Guizang Social Card Skill: Xiaohongshu & WeChat Covers

Guizang Social Card Skill is a 6.3k-star agent skill that turns articles, screenshots, product notes, and user videos into Xiaohongshu / Rednote carousels, Live Photo motion cards, and WeChat 21:9 + 1:1 cover pairs. The official README calls it the sister project of the author’s Guizang PPT Skill: shared visual language, separate maintenance. PPT handles horizontal swipe talks; this one handles static feed images.

What It Is

The skill ships the same two design systems as its PPT sibling. Editorial draws on restrained magazine layouts — think Monocle, Kinfolk, Cereal — for storytelling, lifestyle, travel, and personal observation. Swiss International goes grid-first with a single anchor color, sharp hairlines, and extreme type contrast, built for product reviews, data, frameworks, and tool roundups. Both share one workflow, so a five-card set stays visually coherent from cover to last page.

Output is single-file HTML rendered to PNG with Playwright — node render.mjs, no bundler, no build chain, no server. Install it with npx skills add https://github.com/op7418/guizang-social-card-skill --skill guizang-social-card-skill, then ask for “a Swiss-style Xiaohongshu carousel from this article, 5 cards, IKB blue.”

What Makes It Different

Generic image tools give you pretty pictures and hope. This skill gives you a design grid.

The layout system alone is worth the install: 28 skeletons — 16 Editorial (M01–M16, including Image-Led Cover, Pipeline, Before/After) plus 12 Swiss (S01–S12, including KPI Tower, H-Bar Chart, Matrix + Hero) — and 10 theme presets. Editorial themes run from Ink Classic and Indigo Porcelain to a Midnight Ink dark mode; Swiss themes lock onto one of four anchor colors: IKB Klein Blue, Lemon, Lemon Green, Safety Orange.

Text-on-image gets treated as an engineering problem, not a styling afterthought. Full-bleed images pass through quiet-zone detection and subject mapping first, so copy dodges faces, products, and key actions; the skill adds localized tint only where needed, never a blanket full-canvas mask. A validator script then auto-checks four failure classes — overflow, type-cap violations, four-band density gaps, footer collisions — before you render.

Live Photo cards follow an information budget the README spells out. WeChat gets 3 seconds: one action point or one state change. Xiaohongshu gets 5 seconds: one compact process. Triple collages exist for three parallel results, not sequential stories. Real talk: I’ve seen people shove 30-second clips into these formats and wonder why the numbers tank. The most useful line in the repo is the rule to make the first frame work as a still card, then let 3–5 seconds of motion add evidence.

Who It’s For

Content teams repurposing one piece across platforms get the fastest win — the same article becomes a WeChat banner, a 1:1 share card, and a Xiaohongshu 3:4 set without re-prompting from scratch.

Indie creators on Chinese platforms are the core audience. The workflow handles platform quirks directly, including .pvt packaging so Live Photos actually publish from an iPhone — desktop and web paths generally can’t.

Agent builders get a clean automation surface. Because output is single-file HTML, you can generate, validate, and screenshot cards in a pipeline without maintaining your own browser stack.

Skip it if you need true motion graphics or long-form video editing. Live Photo here means “user video inside a card layout,” not a video editor. The project is AGPL-3.0 licensed.

How It Compares

Same author, same two visual systems, opposite deliverable: Guizang PPT Skill builds horizontal swipe decks, this one builds static feed cards. Both beat generic image tools because neither leaves layout to chance.

Bottom Line

Guizang Social Card Skill closes the gap between finished copy and platform-ready cards — 28 layout skeletons, 10 themes, a validator, and an honest motion budget, all delivered as one HTML file. If your agents already produce articles, this turns them into Xiaohongshu carousels and WeChat cover pairs in a single prompt. Pair it with the HTML Anything skill for broader HTML generation, or grab the sister PPT skill when the deliverable is a talk, not a feed post.