AI Agent Hub
Back to skills
AI Image Prompt Master icon

AI Image Prompt Master

Design & Media Updated 2026.08.30

Paste the following prompt into your AI chat to install this skill:

Please follow https://skillhub.cn/install/skillhub.md and install @user_19bbb2a6/imageprompt-socawt.

About this skill

Problem

When prompts only say “premium,” “cinematic,” or “product shot,” image models often guess the subject, use case, aspect ratio, and text constraints. A reference image also needs a stable path into a reusable prompt.

How it works

The skill organizes visual requirements with the SOCAWT dimensions: Space, Object, Core concept, Artistry, Wash/color, and Type/hierarchy. It supports two workflows:

  • Text-first creation: asks for purpose, subject, style, composition, text, color, constraints, and output quality, then maps them to SOCAWT.
  • Image analysis: classifies the image as photographic, design, or abstract art, selects analysis depth by complexity, and produces a natural-language prompt.

The prompt uses full sentences instead of keyword stuffing, including canvas ratio, quoted text, and elements that must not change. For posters, infographics, or product images, it emphasizes exact wording, hierarchy, and structural limits to reduce the mess of asking for multiple assets at once.

Boundaries

Image analysis requires a model that can read images. If the image is blurry, corrupted, or under-specified, fall back to text description. It targets natural-language image models such as Jimeng, GPT-Image-2, and Nano Banana, not parameter syntax like --ar or (weight:1.2).

Use Cases

  • Write a product-image prompt for an e-commerce page, specifying a frosted serum bottle, side lighting, background, and no text.
  • Analyze a cyberpunk city-night reference image, extract composition, lighting, color, and camera feel, then generate a reusable prompt.
  • Break a poster screenshot into text hierarchy, subject, color, background, and layout, then write a GPT-Image-2 design-brief prompt.
  • Set a 3:4 vertical Xiaohongshu cover with left-of-center subject, warm orange palette, and exact title text, then output a natural-language prompt.

Best For

  • Visual designers who need e-commerce hero image prompts while locking product shape, color, and labels
  • Independent creators who want to turn reference images into reproducible prompts for Jimeng or GPT-Image-2
  • Marketers who design posters or infographics and need control over exact text, hierarchy, and canvas ratio
  • Engineering-focused prompt learners who want to decompose subject, composition, color, and style using SOCAWT