微信公众号排版

Create a local HTML preview page that renders a provided article as WeChat-compatible rich HTML and includes a “复制 HTML” button. Preserve the article's meaning and content; only make light structural adjustments for layout readability.

公众号配图适配

定位：这是公众号「整篇视觉/排版」这一路, , 用户给文章内容，你输出可直接贴进公众号编辑器的 HTML 排版（复制 HTML 一键搬运）。是本集合里配图 Skill 的搭档：space-chart-image / space-text-logic-diagram 出单张配图，本 Skill 出整篇排版。

输出：默认 HTML（本 Skill 主路径）。需要把某个区块（如头图/金句卡）导出成图片时，对该区块截图导出 PNG，或交给 Codex/workbuddy 内置出图模型重绘。

内容不改写：只做排版层面的轻调整，保留全部事实、顺序、引用、结论。

Workflow

1. Collect the article

   • If the user has not provided the article, ask them to paste it.
   • Detect title from the first # heading or first strong title-like line. If unclear, ask for a title.

2. Ask for style

   • Ask which style to use unless the user already specified one:
     • Claude: warm off-white, calm editorial, restrained borders, generous whitespace.
     • OpenAI: crisp white, black/gray, utilitarian documentation feel.
     • Google: bright white, soft blue/green/yellow/red accents, modular cards.
   • If the user says “你来定”, choose based on content:
     • reflective/narrative/opinion → Claude
     • technical/tutorial/product notes → OpenAI
     • data/list/comparison/education → Google
   • Read references/style-guide.md only when implementing the chosen style or when comparing styles.

3. Prepare the content

   • Do not rewrite the article substantially.
   • Keep all facts, order, quotes, scores, names, and conclusions intact.
   • Allow only layout-oriented optimization:
     • normalize heading levels;
     • split very long paragraphs;
     • convert obvious tables/lists into styled HTML tables/cards;
     • wrap long body sections in scrollable preview windows if the user asks or if a section is very long.