buluslan/gpt-image2-ecommerce

What it solves

This project provides a streamlined way to generate professional e-commerce product imagery without needing a professional designer. It solves the problem of creating high-quality, consistent visual assets (such as hero images, lifestyle shots, and social media posts) by automating the prompt engineering required for AI image generation.

How it works

The tool acts as a "Skill" that can be integrated into AI Agents (like Claude Code, Cursor, or Windsurf) or used via the command line. It uses a library of 25 structured prompt templates for various e-commerce scenarios. When a user provides a natural language request, the system identifies the intent, matches it to the best template, assembles a refined prompt, and executes the image generation via the Codex CLI.

Who it’s for

• E-commerce sellers looking to create high-quality product photos quickly.
• Cross-border trade merchants needing professional-grade assets for platforms like Amazon or TikTok.
• AI Agent users who want to add image generation capabilities to their development environment.

Highlights

• 25 Specialized Templates: Covers everything from white-background hero images and lifestyle scenes to UGC-style buyer shows and technical exploded views.
• Reference Image Support: Allows users to upload a product photo to maintain brand and product appearance consistency.
• Anti-AI Look: Includes specific processing for UGC and social media scenes to add realistic imperfections and film textures (CCD retro style) to make images look less synthetic.
• Style Variants: Each scenario offers four distinct style variations to fine-tune the visual mood.
• Agent Integration: Compatible with any AI Agent capable of executing shell commands via Codex CLI.