I Built a Chrome Extension in 24 Hours With Zero Coding Experience

I had an idea for a Chrome extension, no coding background, and 24 hours. With Claude and ChatGPT doing the heavy lifting, I shipped it to the Chrome Web Store. Here's exactly how it went — including the parts that almost broke me.

It started with frustration. I kept losing track of AI prompts I'd written — the good ones, the ones that actually worked. I wanted a simple browser extension that would let me save, tag, and recall prompts from any tab. Seemed simple enough. Problem: I've never written a line of JavaScript in my life.

I opened Claude and typed something like: "I want to build a Chrome extension that saves text I highlight on any page. I have zero coding experience. Walk me through it step by step." What came back wasn't a tutorial — it was actual code. A manifest.json, a content script, a popup UI, the whole thing. Claude explained what each file did as it generated it. I copy-pasted into a folder, loaded it as an unpacked extension in Chrome's developer mode, and it... mostly worked. The first version had a bug where saved prompts would disappear on browser restart because I hadn't implemented localStorage correctly. I pasted the broken behavior back into Claude: "When I reopen the browser, all my saved prompts are gone. Here's my current code." Fixed in one response.

The iteration loop was the real unlock. I'd break something, describe the break, get a fix. I'd want a new feature — drag to reorder, a search bar, export to CSV — describe it in plain English, get working code. ChatGPT handled a few of the trickier UI components; Claude was better at explaining the Chrome extension architecture when I got confused about message passing between scripts. Total build time across both sessions: about 14 hours of actual work, spread across 24 hours. The Chrome Web Store review took another three days, but the extension passed on the first submission.

The skill being demonstrated here isn't coding. It's specification. The ability to clearly describe what you want, observe what breaks, and articulate the gap between expectation and reality. That's it. If you can do that — and most intelligent adults can — you can now build functional software. Not enterprise-grade systems, not apps that need to scale to millions of users. But real, working tools that solve real problems.

The Chrome extension now has 47 installs, mostly from people I've shared it with directly. That's not a startup. But it's a thing that exists in the world because I described a problem clearly enough for an AI to help me solve it. The barrier between "I wish someone would build that" and "I built that" has collapsed for a huge category of software. This is one of the most underreported practical shifts in the current AI wave.