Skip to content
Kitbay
Skill

gpt-image-2-style-library

MITUpdated on Nov 10, 202635,689GitHub starsIncludes scripts

A skill from the awesome-gpt-image-2 prompt and case library that helps an agent write prompts for OpenAI's GPT Image 2 image model using the library's templates.

What the skill teaches

A skill from the awesome-gpt-image-2 prompt and case library that helps an agent write prompts for OpenAI's GPT Image 2 image model using the library's templates. The agent reads a reference file generated from the repository's style data, detects whether the request is in Chinese or English, works out the kind of image wanted, such as a product shot, poster, interface, infographic, character or scene, and matches it to a template category, style tag, scene tag and nearby example cases, offering two or three options when several fit. The prompt is built from blocks for subject, composition, style and materials, exact text, aspect ratio and things to avoid, and is returned first with the template name and example IDs.

Capabilities

  • Match an image request to a prompt template
  • Build a structured GPT Image 2 prompt
  • Reply in the user's language
  • Cite example cases from the library