card-xiaohongshu · Xiaohongshu Knowledge Cards
A skill that renders copy into 1080×1440 vertical Xiaohongshu knowledge-card sets: one knowledge drop becomes a 3–9 card swipeable deck (cover → body → closing), each card carrying exactly one key point.
Before rendering, a visual style must be chosen from the card-design design system (9 named styles). Cards are rendered from pure HTML with automated screenshots, then passed through the card_audit.py quality gate — only all-PASS decks ship.
Example invocation: "Turn these book notes into 6 Xiaohongshu knowledge cards."
Full brief
Positioning
card-xiaohongshu renders finished card copy into vertical Xiaohongshu knowledge-card sets: 1080×1440 per card, 3–9 cards per deck, screenshottable as a whole or individually. It is the "render layer" — it does not plan notes or write copy.
Core capabilities
- Style first: before the first line of HTML, read the card-design design system and lock one of 9 named styles (Swiss Minimal, Editorial, Neo-Chinese Ink, Cream, Dopamine Y2K, Luxe Black-Gold, Sticker Journal, Geek Terminal, Botanical Fresh) — fonts and palette stay locked for the whole deck.
- Skeleton selection: structure per content type (cover/ledger/pipeline/compare/matrix/data/quote/closing); one key point per card, filled with information-bearing layouts — never one sentence floating in dead space.
- Automated rendering: HTML is batch-rendered to images by script, no manual screenshots; first run needs playwright + chromium installed.
- Quality gate audit: card_audit.py runs after rendering; cards failing the dead-space/density gate get content or skeleton fixes and re-render until all PASS.
- Design floor: no tech gradients, no gradient text, no glassmorphism, no emoji-as-icons, no bold-black display type; body text never pure black; small watermark on each card.
Workflow
- Pick the style (user chooses or takes a recommendation; lock fonts/palette)
- Pick skeletons (cover / body / closing cards)
- Write the HTML (cards stacked vertically)
- Script-render to images (each .card element exported individually)
- card_audit quality gate (fix content or skeleton and re-render on FAIL)
- Ship once everything passes
Inputs & outputs
| Input | Notes |
|---|---|
| Card copy | Required — finished content (note planning itself is xhs-note-creator) |
| Visual style | Recommended — one of 9 named styles; recommended then confirmed if unspecified |
| Card count | Optional — driven by content volume (3–9 works best) |
Output: card images (card_1.png, card_2.png…) under outputs/<theme>/.
Boundaries with adjacent skills
- card-quote: 16:9 horizontal quote/data cards — one hero statement or key number for Weibo/Zhihu/X/WeChat articles.
- poster-hero: 1080×1920 standalone marketing poster/share image (big title + selling points + QR code) for launches and campaigns — not the note cover; note covers use this skill's first card.
- xhs-note-creator: full-note planning and copywriting (the planning layer); this skill is the render layer.
Fit
- Splitting long-form knowledge or book notes into swipeable Xiaohongshu decks
- Knowledge accounts batch-producing image-text notes
- Card-ifying courses and methodology content
- Multi-card series needing one locked visual identity
Before you start
- Have the copy ready (this skill renders; it doesn't write).
- First render needs
pip install playwrightplus the chromium browser. - Style selection is step one: style decides whether the cards look good — don't force every topic into the same one.
card-xiaohongshu is part of the Aiglade Skill library. Invoke it from the Aiglade chat box in plain language.