What it does
Capture the visible tab, send the image to an AI provider, get the answer in place. It works with Gemini, GPT-4o, Claude, Groq, Mistral and OpenRouter, so you're not locked to whichever one is cheapest or best this month.
You bring your own API key. That keeps it a one-off $5 rather than a subscription, and it means I never handle your keys or your screenshots — but it does mean there is a two-minute setup before it works, and I would rather say so here than have you find out after paying.
What was actually hard
Chrome's captureVisibleTab permission can't be called from just anywhere in an extension — the call has to be routed through a background service worker, and getting that message-passing right took longer than the entire user interface did.
The other genuinely hard part had nothing to do with code: pricing it, writing the sales copy, and setting up license key validation so the product couldn't simply be copied. That's the part nobody warns you about when you decide to sell software instead of open-sourcing it.