GPT Image 2 Visual Quality Diagnostic Assistant

Pre-release diagnosis

GPT Image 2 can render a prompt into an image, but marketing, e-commerce, content media, and internal review teams often still need a practical check: can this image ship? Goals, audiences, brand rules, aspect ratios, reference assets, and platform constraints are scattered across chat threads, product pages, and verbal requests. This skill turns those checks into a repeatable diagnostic pass.

How it works

It accepts JSON parameters such as goal, audience, materialUrl, materialFile, materialText, and brandRequirements, then organizes the image goal, target audience, reference links, drafts, prompt notes, and brand constraints into context. The output typically includes a summary, diagnostic conclusion, action recommendations, and reusable deliverables for further image editing, review, or handoff. If the input is sparse, it asks for the goal, audience, constraints, or source material instead of overcommitting to a strong conclusion.

Boundaries

It is best suited to pre-release review, material cleanup, prompt and visual-description retrospectives, and turning scattered requirements into next actions. It is not a final legal or platform-compliance decision; copyright, model releases, and restricted styles still require human confirmation. materialUrl should be a fully accessible URL that does not require login. The more complete the input, the more specific the diagnosis.