Xiaohongshu image direction workflow for Chinese creators. Use when Codex needs to turn a note, draft, or topic into high-quality Xiaohongshu image directions: propose two to three distinct design routes first, choose the right visual hierarchy for the content, prefer 3:4 vertical native layouts, decide between AI image generation and HTML fallback, and review the result for both design quality and feed attractiveness before finalizing.
这个 Skill 处理的是小红书配图的“方向、提案、生成与审查”。
核心原则只有一句:
先提案,后生成。
默认采用以下约定,除非用户另有说明:
先提炼:
必须先给 2-3 个方向,每个方向写清:
默认读取:
| 路径 | 什么时候用 |
|---|---|
| AI 生成 | 需要质感、氛围、插画感、纸张感、手作感 |
| HTML 兜底 | 文字必须精确、数据复杂、表格信息多 |
每次都至少检查:
Do not:
Always: