返回 Skill 列表
extension
分类: 内容与媒体无需 API Key

midjourney-prompts

生成Midjourney V7版本的网站部分图片提示。分析页面组件,从CSS中提取纵横比,并输出带有SEO文件名的提示。在完成页面后使用(技能02-06)。触发关键词为“生成图片提示”、“midjourney提示”、“为页面创建图片”。

person作者: jakexiaohubgithub

Midjourney Prompts

Generate MJ V7 prompts for all images in a page.

Workflow

  1. Read page file and components to identify sections with images
  2. Read docs/styleguide.md for brand style keywords (if exists)
  3. For each image slot:
    • Extract aspect ratio from CSS classes
    • Generate prompt with brand style
    • Suggest SEO-friendly filename

Aspect Ratio Reference

| Section Type | CSS Classes | Ratio | |--------------|-------------|-------| | Hero background | h-svh, h-[620px] | 21:9 or 16:9 | | Project thumbnails | h-[50vh] | 16:9 | | Testimonial avatars | h-[288px], h-[328px] | 1:1 or 3:4 | | Feature images | max-h-[420px] | 4:3 | | CTA backgrounds | h-[620px] | 21:9 | | Process images | h-90 | 4:3 |

Prompt Formula

See references/prompt-formula.md for MJ V7 structure. See references/examples.md for 11 reference prompts.

Human Guidelines

  • Default: No humans - use objects, environments, abstract concepts
  • If humans needed: Silhouettes, backs, backlit figures (no faces)
  • Representing owner: Never show face directly

Output Format

For each image in the page:

### [Number]. [Section Name]
- **Section:** [component name]
- **Purpose:** [what the image shows]
- **Aspect Ratio:** [ratio] (--ar X:Y)
- **Suggested Filename:** [seo-friendly-name].webp
- **Prompt:**
[ready-to-copy MJ V7 prompt with flags]

Save the output in docs/midjourney-prompts/[page-name]-prompts.md.