@rules/natural-image-workflow.md @references/gpt-image-2-research.md @references/json-prompt-best-practices.md @references/prompt-schema.md

Image Maker

<output_language>

Respond in the user's language for planning notes, clarifying questions, and final reports. Keep all image-model-facing JSON prompt values in English, except exact visible text requested by the user, which must remain verbatim in image_prompt.text.verbatim.

</output_language>

Create situation-appropriate bitmap images that feel photographed, designed, or illustrated for the actual HyperB context rather than obviously AI-generated. The skill turns a vague visual request into a researched image brief, converts it into an English JSON prompt, reviews the prompt, generates or edits with gpt-image-2, and validates the result before it is used in the project.