Forking Around

Etsy Automation

From 'nice picture' to 'live on Etsy' without lifting more than one finger.

Queued up next

The Story Behind the Madness

How Etsy Automation Escaped The Idea Graveyard

A full-stack automation tool that turns listing print-on-demand products into a single guided flow. Upload an image, and a local AI pipeline takes over: 8x ESRGAN upscaling, Ollama vision for image analysis, and a second LLM pass for SEO-optimized Etsy metadata — titles, descriptions, and exactly 13 tags. Smart-crops to product-specific aspect ratios, submits to Printify with full variant and pricing config. Everything runs locally — no API bills, no cloud, one flow from image to draft product.

Genesis Spark

Born from the sheer audacity of Etsy expecting me to manually create 300 product listings like it's 2005. The dream was full automation — image in, live listing out. Reality? The Printify API had other plans and still forces parts of the flow to stay manual. But the hours of copy-paste-upload-repeat? Those got automated into oblivion.

Core Problem & Insight

Every Etsy listing is a 15-step obstacle course: upscale the image, crop for each format, write SEO copy, generate tags, set prices, map variants. Do it once — fine. Do it 300 times and you start questioning every life choice that led you to this moment. The pain wasn't the complexity; it was the soul-crushing repetition.

Known Chaos

Getting print-quality resolution from low-res source images without turning everything into an impressionist painting of artifacts.
Generating SEO-friendly metadata that actually follows Etsy's character limits and doesn't read like a keyword salad.
Pushing large image payloads (up to 100MB) through the full upscale-crop-upload pipeline without the server tapping out on memory.

Solutions

ESRGAN-slim 8x via Upscaler.js with patched inference (patch size 64, padding 6) handles the upscaling reliably on a Node.js backend. The images come out clean.
A prompt-engineering layer with strict formatting rules, shop-context injection, and product-type suffixes keeps the local LLM on a leash — consistently outputting Etsy-ready metadata instead of creative writing exercises.
Express with a 100MB body-parser limit and TensorFlow.js GPU acceleration handles the heavy lifting server-side. Big images, no drama.

Arsenal

NodeJS

A JavaScript runtime built on Chrome's V8 JavaScript engine.

Express

A minimal and flexible Node.js web application framework for building APIs and web apps.

TensorFlow.js

A library for machine learning in JavaScript, enabling training and deployment of ML models in the browser and Node.js.

Upscaler.js

A JavaScript library for upscaling images using neural network models.

Ollama

A tool for running large language models locally on your machine.

Vanilla JavaScript

Plain JavaScript without any frameworks or libraries, using the core language and browser APIs directly.

Astro

A web framework for building fast, content-driven websites using an islands architecture.

Thanks for Peeking Behind the Curtain!

This one's still brewing in the lab. Head back and explore more chaotic creations!