xiaohongshu-card-notes（小红书卡片笔记）

Xiaohongshu Card Notes

Follow shared release-shell rules in:

• postplus-shared release-shell rules

Use this skill when the user wants to turn existing material into a Xiaohongshu card-style note.

Typical source inputs:

• an HTML landing page
• a spoken script
• a markdown memo
• a result table or dashboard
• a case study page
• a workflow explanation that should become carousel cards

This skill is for card-style image notes, not for generic captions or long prose posts.

Default Goal

Produce a Xiaohongshu card note that is:

• easy to scan on mobile
• visually structured
• page-by-page clear
• information-dense without losing scanability
• faithful to the source material

The target is not "rewrite beautifully."

The target is:

• extract the strongest message from the source
• map it into a small number of strong cards
• keep each page readable even when density is high
• make the output directly usable for design

First Decision

Decide what the user actually needs:

1. content-to-cards

Use when the user has source material and wants a full Xiaohongshu card note structure.

Return:

• page plan
• cover title
• per-page headline
• per-page body copy

2. layout-only

Use when the user already has the page copy and wants packaging only.

Do not rewrite the substance unless asked.

3. html-adaptation

Use when the source is an HTML page or landing page.

Inspect:

• section hierarchy
• proof/data modules
• title system
• visual rhythm

Then convert it into Xiaohongshu pages rather than mirroring the full web layout.

4. script-adaptation

Use when the source is a spoken script.

Do not preserve spoken redundancy. Compress the script into card-ready statements.

5. hybrid-source

Use when the user provides both an HTML page and a script.

Default behavior:

• take structure from the HTML if it is strong
• take judgment and phrasing from the script if it is stronger
• merge them into one card sequence

Layout Rules

Read references/card-layout-spec.md whenever the user asks for design-ready card output.

Non-negotiable defaults:

• aspect ratio: 3:4
• canvas size: 1242 x 1656
• do not assume every page must be sparse
• some pages should be image-led
• some pages should be dense memo pages
• left alignment is the default
• use only a small number of emphasis treatments
• when a repo already contains a strong HTML or landing page reference, inherit its color system, card language, corner radius, and visual rhythm instead of inventing a new style
• when the user provides a strong HTML or landing page reference, treat that file as the visual source of truth for card rendering
• use the landing page's dark floating pill, luminous blue gradient blocks, pale blue surfaces, large radius cards, and black headlines on light backgrounds
• older black memo-card constraints are deprecated in the current release style and should be overridden when they conflict with the landing page reference

Source Adaptation Workflow

Read references/source-adaptation-workflow.md when adapting from HTML, scripts, or mixed sources.

The working order is:

1. identify the core claim
2. decide the card sequence
3. compress each page to one job
4. assign a page role
5. write card copy that fits the page role

Page Roles

Common page roles:

• cover
• image-led cover
• problem
• hidden truth
• framework
• dense memo explanation
• proof / data
• case result
• workflow snapshot
• table screenshot page
• closing judgment

Not every note needs all of them.

Choose the smallest set that makes the story clear.

Copy Rules

For each page:

• one page = one main takeaway
• do not flatten everything into short slogans
• prefer short declarative lines
• keep keywords strong and sparse
• preserve the user's judgment and operator tone
• avoid 不是……而是…… by default
• avoid 也就是说 by default
• prefer direct statements over rhetorical contrast templates

For dense memo pages:

• allow 3-6 short paragraphs when needed
• each paragraph should push one sub-point
• paragraph spacing matters more than bullet count
• the page should still feel screenshot-worthy on its own

For image-led pages:

• let the image carry pause, mood, or proof
• keep the copy compact but sharp
• use these pages to create rhythm between dense pages

If adapting from a spoken script:

• remove repeated setup
• remove oral filler
• keep the strongest lines
• convert explanation into page headlines and short supporting text

If adapting from HTML:

• do not reproduce navbar, footer, or web-only scaffolding
• keep hero claim, proof blocks, workflow blocks, and result table logic only if they help the card sequence
• web sections often need to be split into multiple cards

Do not force every section into a sparse card. If the source contains strong reasoning, preserve it as a dense memo page.

Output Shapes

Return the smallest useful output for the ask.

Common shapes:

• 图文结构
• 封面标题备选
• 逐页文案
• 设计备注
• 图文 brief

For full card generation, default to this format:

封面

• title
• subtitle or tag

第2页

• headline
• body
• optional design note

Repeat page by page.

Design Notes

Only include design notes when they help execution.

Useful design notes:

• text hierarchy
• suggested emphasis word
• suggested chart / table / screenshot use
• whether the page should be mostly text or mostly visual
• which parts of the reference HTML visual language should be inherited
• which card should use hero-gradient treatment
• which card should use white proof-table treatment
• which card should use pale-blue supporting surfaces instead of dark memo panels

Avoid vague notes like "make it aesthetic."

Density Preference

This workspace now prefers a denser Xiaohongshu note style than generic carousel cards.

Default preference:

• mix sparse and dense pages
• let some pages read like high-value memo screenshots
• allow stronger paragraph density when the reasoning matters
• use images or screenshots to create breathing room between dense pages

Do not use 3 lines max as a universal rule. That only applies to light pages, not to memo pages.

Anti-Patterns

Do not:

• dump the whole script onto cards unchanged
• mirror a webpage section-for-section without compression
• overload each page with too many bullets
• use too many highlight colors
• make every page equally loud
• add generic Xiaohongshu fluff that weakens trust

Default Input Shape

A common input shape is:

• professional scripts
• structured HTML pages
• result tables
• proof-heavy workflow explanations

Default preference:

• preserve the strongest judgments from the source
• convert proof into visible card pages
• turn tables or dashboards into one or more dedicated proof cards
• mix high-density memo pages with visual proof pages
• make the final output feel like a professional Xiaohongshu carousel, not a repackaged web page

Release-Shell Execution Contract

• keep page plans, source extracts, and intermediate layout notes under <work-folder>/.postplus/xiaohongshu-card-notes/
• keep only final user-facing card copy or design briefs outside .postplus/
• start with a bounded first pass on one source document and a small card set
• fail fast if the required source material is missing or under-specified instead of inventing missing proof or using customer-specific fallback assumptions