加载中
user_d69b0d4a
Reverse-engineer reference images into topology-correct, confidence-labeled visual analyses and high-fidelity image-generation prompts. Use when analyzing or recreating an image, extracting or improving a prompt, comparing references, diagnosing a failed reproduction, or revising an analysis after a user corrects spatial relationships, occlusion, reflection, pose, styling, camera, lighting, color, or other visual facts. Works with a host agent's native image understanding or an included OpenAI vision adapter for agents without vision. Produces Scene Topology, Reproduction Priority Map, Spatial Constraints, High-fidelity Prompt, Negative Prompt, Reproduction Risks, and dependency-aware local recompilation.
PURPOSE
Reverse-engineer reference images into topology-correct, confidence-labeled visual analyses and high-fidelity image-generation prompts. Use when analyzing or recreating an image, extracting or improving a prompt, comparing references, diagnosing a failed reproduction, or revising an analysis after a user corrects spatial relationships, occlusion, reflection, pose, styling, camera, lighting, color, or other visual facts. Works with a host agent's native image understanding or an included OpenAI vision adapter for agents without vision. Produces Scene Topology, Reproduction Priority Map, Spatial Constraints, High-fidelity Prompt, Negative Prompt, Reproduction Risks, and dependency-aware local recompilation.
USER EXPERIENCE
暂无真实用户评价
如果你实际使用过这个 Skill，可以分享具体场景、效果和踩坑。平台不会用评价替代 VERIFIED、安全或兼容性证据。
REQUIREMENTS
当前版本没有声明额外依赖。
权限、网络访问和静态扫描结果放在“安全与权限”页签，避免把工程细节堆在概览里。
EVIDENCE