feichanggege/ecommerce-visual-copywriting-skill
Security testingThis page was translated automatically. Technical details may differ from the original. View the original
E-commerce Visual Copy Design SOP Skill - Enables e-commerce design beginners to create product images with 5 years of senior design experience with one click. Compatible with SKILL.md Agents such as Claude / Codex / Cursor / Workbuddy
Overview
Refine product information into a main image, detail page storyboard script, in-image copy, design description, and raw image prompt. These diagrams show the final execution form for this skill: main image visual interception, multi-angle product view, scene detail page, and textured selling point expression. The display images are used to illustrate the form of visual delivery and do not constitute legal opinions, platform review commitments, or commercial data proof for the corresponding products. - Write the final copy directly; only rework if the direction is wrong. - The selling point sounds strong, but it encounters advertising laws, efficacy claims, or platform review red lines. - Output only the copy, without visuals, layering, material lighting, disclaimers, or design execution instructions. This skill breaks down e-commerce visual copy into a controlled SOP: first identify the conversion driver, lock in unified visual style standards, then output main image and detail page storyboard scripts; Only after user confirmation do you proceed to the final image, in-image copy, design description, and raw image prompts. If your Agent runtime does not yet support npx skills add, you can manually copy it: - Chinese: Use SKILL.md. - English version: View README.en.md and SKILL.en.md. - At runtime, you can directly say to the Agent: "Output in English" or "Output in English." - "Help me create a set of Taobao main image and detail page copy" - "Give me the main image storyboard script first, don't write the final copy directly" - "How do you extract the selling points of this product to make the main image?" - "Create a visual script in the style of a Douyin store" - "Does this detail page carry advertising law risks?" - "Designer needs to get started, give me images + in-image copy + design description" - "Output English version of e-commerce main image planning"
Overview (machine translated)
E-commerce Visual Planning · Main Image / Details Page / Listing / A+ · Image Copy · Raw Image Prompt · Compliance Review
**Turn scattered product information into e-commerce visual solutions that designers and AI can directly execute. **
Chinese · English
Get Started in 30 Seconds · What Does It Solve · Workflow · Supported Platforms · Effect Display · Security Boundaries
It’s not about “writing another piece of ad copy.” It first determines **why users will buy **, then pinpoints visual style and product authenticity, establishes an evidence ledger, and finally outputs executable images for each image, in-image copy, design descriptions, and raw image prompts.
✨ What does it solve?
The common problem with ordinary AI for e-commerce visuals isn’t “not knowing how to write,” but rather:
| Frequently Asked Questions | How to handle this Skill |
|---|---|
| Directly stacking selling points, 5 images tell the same thing | Each chart only solves one purchase decision problem |
| Many parameters, but users don’t know what they have to do with themselves | Feature → Advantage → Benefit → Evidence |
| The reference image is changed more and less like the original product | Reference Fidelity + Negative Constraints Lock packaging, logo, structure, specifications |
| After “writing a bit more aggressively,” risks of efficacy, absoluteization, and false evidence arise | First establish the evidence ledger, then classify claims |
| Users only want to modify existing detail pages but are redone from scratch | Automatically enter review/revision mode, making the minimum necessary changes |
| Cross-border pages are just literal Chinese translations | Localized rewriting based on market, organization, scenario, and expression habits |
| Platform rules have changed, but AI still relies on old experience | Current rules prioritize verifying official latest sources when needed |
| The output is only text, and the designer still has to guess a second time | Deliver each image + copy + design description + prompt + prohibited items |
🚀 Get it started in 30 seconds
1. Installation
2. Just state your needs
It can also be simpler:
🧭 Workflow
Four Execution Modes
| Pattern | When to use | Behavior |
|---|---|---|
| Standard Mode | Unclear direction, high cost, user requirements confirm first | First strategy → Storyboard → Confirm then execute |
| One-time delivery | Users must clearly “complete it directly” or have sufficient documentation | No mechanical pause, complete delivery |
| Review / Revise | Existing main image, detail page, copy, prompt | Find the problem + Minimum necessary revisions |
| Designer Handover | Execution draft that requires immediate start | Output per-map task cards using a fixed structure |
Core Capabilities of the New V3
- Evidence Ledger: Each key selling point is marked as “Confirmed / Supported / Pending Evidence / Prohibited.”
This page was translated automatically. Technical details may differ from the original.
Recommended Tools
Try a different keyword or remove a filter.
Install
npx skillfish add feichanggege/ecommerce-visual-copywriting-skill