---
name: amazon-listing-image-skill
description: Creates production-ready Amazon listing images and Amazon ad creatives with any image-generation model. Use when an operator needs Amazon main images, secondary listing images, Sponsored Display or DSP creative, product photography prompts, creative briefs, or a model-agnostic Amazon asset QA workflow.
---

# Amazon Listing Image Skill

Create truthful, product-accurate Amazon creative. The operator may use any capable image model, design tool, or editor. Do not assume a particular API, dashboard, model, or filesystem.

## Intake gate

Collect these before generating:

1. Product name, category, marketplace, ASIN/SKU, and target customer.
2. Scope: `listing set`, `ad set`, or both.
3. Canonical product photos from every relevant angle, packaging photos, dimensions, color/variant, and included items.
4. Brand kit: logo rules, colors, typefaces, approved voice, and approved claims.
5. Differentiators with evidence. Treat every performance, comparison, certification, review, price, or guarantee statement as unapproved until evidence is supplied.
6. For ads: campaign type, placement, destination, CTA, offer, and required image dimensions.

Stop and request missing assets when product identity, included items, or claims cannot be verified. Never use a competitor listing as a visual source or imitate its branding.

## Choose the output path

### Listing image set

Build a seven-image plan unless the brand has a different approved template:

| Slot | Goal | Copy |
|---|---|---|
| 01 | Main image | None |
| 02 | Product-in-use lifestyle | Minimal or none |
| 03 | Primary benefit | Outcome-led, evidence-backed |
| 04 | Features and materials | Short, specific callouts |
| 05 | Size, fit, or compatibility | Accurate dimensions/diagram |
| 06 | Detail or use case | Show the real use |
| 07 | Comparison, bundle, or proof | Only substantiated facts |

For each slot, produce: goal, exact visual composition, approved copy, source assets, model prompt, negative prompt or constraints, and acceptance checks.

### Amazon ad set

Design an ad system, not isolated images:

1. Define one audience, one pain point, one product truth, and one CTA per concept.
2. Produce 3 distinct concepts: product-led, outcome-led, and use-case-led.
3. Make every message understandable before the click. Never imply an Amazon endorsement, ranking, badge, review count, discount, or scarcity without current written approval.
4. Create a placement matrix from the campaign brief. Confirm the current Amazon-spec requirements in the ad console or official documentation before export; placement specs change.

For each concept, deliver a headline, primary visual, supporting proof, CTA, on-image copy, model prompt, and crop-safe layout notes.

## Main image rules

The main image is a compliance-sensitive product record, not an ad:

- Use pure white background (`#FFFFFF`).
- Show only the item(s) actually included; show the correct quantity, variant, color, and packaging where appropriate.
- Product should occupy roughly 85% of the frame without clipping.
- No text, graphics, badges, borders, insets, lifestyle scenes, props, or unverified accessories.
- Export a square high-resolution image. Verify the current category and marketplace requirements before delivery. Amazon’s general image guidance requires at least 500 px on the longest side and recommends 1,000 px or more for zoom; use the brand’s current seller-console requirement if it differs.

## Prompt method for any model

Use supplied product photos as visual references whenever the model supports them. Generate the environment and lighting around the verified product, rather than asking a model to invent the product.

Use this prompt structure:

```text
TASK: [listing slot or ad concept]
PRODUCT TRUTH: [exact product, variant, included items, materials, dimensions]
REFERENCE PRIORITY: Match the supplied product reference images exactly: silhouette,
color, proportions, hardware, labels, finish, packaging, and count.
COMPOSITION: [camera angle, framing, scene, subject action, whitespace/crop safe area]
LIGHTING AND STYLE: [brand-approved visual direction]
COPY IN IMAGE: [exact approved text, or "none"]
OUTPUT: [aspect ratio, resolution, file type]
CONSTRAINTS: No invented features, accessories, labels, logos, certifications, text,
watermarks, extra units, altered proportions, or competitor marks.
```

If the model renders copy unreliably, generate the clean visual first and add approved typography in a design tool. Do not ask the model to guess a logo or product label.

## Production loop

1. Create an asset manifest before generation.
2. Produce 2–4 visual variations per slot or concept.
3. Review at 100% zoom against the references.
4. Regenerate or edit only the defect. Do not accept a near-match for product identity.
5. Add final text after visual approval where needed.
6. Export organized final files and the manifest.

Use descriptive filenames:

```text
[brand]_[asin-or-sku]_[variant]_[listing-01-main|ad-concept]_[size]_v01.png
```

## QA checklist

Before handoff, confirm every final asset:

- [ ] Matches the supplied product, color, count, components, labels, and packaging.
- [ ] Contains only evidence-backed claims and exact approved copy.
- [ ] Has no unreadable, hallucinated, or misspelled text.
- [ ] Has no competitor brand, marketplace logo, restricted badge, watermark, or misleading visual.
- [ ] Main image follows the main-image rules above.
- [ ] Secondary images have legible mobile-scale copy and do not overload the frame.
- [ ] Ad assets have correct dimensions, safe areas, CTA, destination, and placement-specific requirements.
- [ ] Filenames, source references, prompts, approvals, and final exports are recorded.

## Handoff format

Return this concise manifest with links or paths to the assets:

```markdown
# [Product] Amazon Creative Handoff

## Inputs verified
- Product / variant:
- Source photos:
- Approved claims:
- Brand kit:

## Deliverables
| Asset | Purpose | Size | Final file | Status |
|---|---|---:|---|---|

## Listing slots or ad concepts
- [Name]: [one-sentence strategy] | Copy: “[exact text]”

## QA result
- Product fidelity:
- Compliance checks:
- Open risks or required approvals:
```

## Non-negotiables

- Do not create deceptive before-and-after images, medical or efficacy claims without substantiation, fake reviews, fake rankings, fake urgency, or fabricated certifications.
- Do not misrepresent the product through generated packaging, accessories, quantities, scale, or features.
- Do not claim compliance approval. The operator performs a preflight check; the seller remains responsible for the current Amazon policies and category rules.