# EcomGen

> EcomGen is a self-hosted, open-source AI workbench for e-commerce sellers to generate product image suites from verified facts using OpenAI-compatible or Gemini provider.

- **Category:** E-commerce
- **Licence:** MIT
- **Latest version:** v0.2.0
- **Open source:** oss
- **Languages:** Typescript
- **Platforms:** Docker Compose, Build from Source
- **Tags:** SQLite, Redis, E-commerce, Vite, Local First, Gemini, Typescript, React, Fastify, Self Hosted, Ai Agent, Ai image generation, Bullmq, Openai compatible, Nano Banana, Pi Agent, Product Photography, Image Editing, Product Images, Seller Tools

## Description

EcomGen is a local-first AI-powered workbench for individual e-commerce sellers to create complete product image suites. It structures product facts, materials, target markets, and platform requirements into auditable storyboards, then uses a Pi Agent to generate final prompts for image generation providers. The system includes 25 built-in e-commerce templates covering hero, scene, infographic, packaging, comparison, and social media shots, plus over 1,100 reusable suite templates across 24 top-level categories. It supports OpenAI-compatible image APIs and Google Gemini native image generation (Nano Banana), with AI-powered layering that exports individual PNG layers, ZIP archives, and PSD files. The stack runs as three services (Fastify API, BullMQ Worker, React/Vite web UI) backed by Redis and SQLite, deployable via Docker Compose or manual Node.js/pnpm installation. All sensitive data including provider API keys are encrypted at rest. The project is MIT-licensed and actively developed.

## Links

- [Homepage](https://github.com/linbei0/ecomgen)
- [GitHub Repository](https://github.com/linbei0/EcomGen)

---
Source: https://selfhost.directory/project/ecomgen