gpt-image-2
面向 GPT Image 2 的图像生成 / 编辑技能。可在 3 种环境下使用：(A) Garden 本地模式，通过 OpenAI 兼容接口直接出图并落盘；(B) Host-Native 模式，把本 Skill 当作提示词工程指引，把渲染好的 prompt 交给宿主 Agent 自带的图像工具出图；(C) Advisor 模式，宿主无任何图像工具时退化为高质量 prompt 顾问。涵盖 18 大类、80+ 个结构化模板，覆盖海报 / UI / 产品 / 信息图 / 学术图 / 技术架构图 / 漫画 / 头像 / 流程板 / 电影分镜 / IP 周边 / 编辑工作流等场景。
- installs 8w
- 1,754
- 30-day movement
- starts with the next reading
- Related entries
- 4
- Connections
- 0
An agent skill for GPT Image 2 image generation and editing, packaged as a SKILL.md with detection and generation scripts plus a reference library of 18 template categories and 80+ structured prompt templates. It adapts to three runtime modes: direct generation via an OpenAI-compatible API, delegating to the host agent's image tools, or acting as a prompt-engineering advisor when no image tool exists.
Reach for it when you want consistent, high-quality GPT Image 2 prompts and images regardless of whether your environment can generate images directly.
Use it to
- Generate images from text prompts via the OpenAI-compatible API
- Edit existing images, including masked partial edits
- Render structured templates for posters, UI mockups, product visuals, and diagrams
- Get expert prompt guidance when no image tool is available
- Save prompts and generated images with task-based naming conventions
For AI coding agent users who need image generation or prompt engineering
- Host repository
- ConardLi/garden-skills
- Installs, lifetime
- 4,700
- Installs, 8 weeks
- 1,754
- Host stars
- 12k
- Host language
- CSS