Generate Amazon Image Set

What It Solves

Amazon product images are not simply “more generated pictures” to ship. MAIN, listing secondary images, and A+ modules serve different purchase decisions, and their sizes, placement, device behavior, and evidence sources affect compliance. If brand assets or reference images are passed directly to an image model, the common risk is that unverified dimensions, materials, packaging, certifications, or performance claims end up in the final images, causing rework and platform risk. This skill turns Amazon image-set generation into a reviewable operating specification.

How It Works

It first builds a product fact lock: only USER_CONFIRMED, DOCUMENT_CONFIRMED, and unambiguous IMAGE_OBSERVED evidence can support accurate claims, while INFERRED and UNKNOWN items must not be promoted to product facts. It then defines delivery scope and creates a slot manifest for MAIN, up to seven listing secondary images, and paired A+ PC/Mobile modules, requiring each slot to answer one main purchase question.

• Every slot lists required facts, reference-image roles, confirmed copy, missing evidence, and status.
• Statuses are limited to READY, NEEDS_REVIEW, and BLOCKED, so unrunnable checks are not disguised as passes.
• Generation briefs are self-contained, including product identity, slot task, size constraints, and prohibited claims.
• Plan QA and final-image QA run separately; QUALIFIED is output only when hard gates pass.

Boundaries

The skill does not assume Python, Shell, OCR, image metadata access, or a specific image-generation tool. If dimensions, packaging contents, or product states are not confirmed, it does not generate the corresponding images. When A+ permissions, marketplace category rules, or module sizes cannot be verified, it should output a draft and mark final compliance as BLOCKED. It fits as a specification layer in an agent workflow, not as an auto-out tool that bypasses compliance or invents product facts.