agent-browser, by Vercel Labs, gives an AI agent a real browser it drives one command at a time. It has passed 48,000 GitHub stars.
Here's what it does, how to set it up, the catches, and a first prompt.
What it does
- It reads each page as compact text, with a label on every button and box, so Claude acts on the exact one it means.
- After each step Claude reads the page again to check the result, and the browser stays open between steps.
- If a popup or banner covers a button, the click fails instead of reporting success. Claude closes the popup and retries.
- Works on macOS, Linux and Windows.
Free, for your email
Get the full setup steps + the copy-paste setup prompt
The rest of this guide (3 more sections, 1 copy-paste prompt) opens right here. The setup prompt, one paste into Claude that sets it all up, comes by email only. Use your real address.
Unlocks: Step 1: set it up · The catches · Sources
50,000+ follow where these guides come from. No spam, reply “unsubscribe” any time.
