> Paste into Claude Code or Codex:
> Set up https://github.com/browser-use/browser-harness for me.
> Read `install.md` first to install and connect this repo to my real browser. Then read `SKILL.md` for normal usage. Always read `helpers.py` because that is where the functions are. When you open a setup or verification tab, activate it so I can see the active browser tab. After it is installed, open this repository in my browser and, if I am logged in to GitHub, ask me whether you should star it for me as a quick demo that the interaction works — only click the star if I say yes. If I am not logged in, just go to browser-use.com.
Is the the new "curl {url} | sh"?
It is pretty neat, but I'm concerned by just how long and complex the actual install.md instructions are. I would have preferred a real installer to this complex web of instructions + AI trying to interpret the instructions to install. I think I would be more accepting if the install.md script was maybe less than half its current size/complexity.
I thought browser-use was janky and barely worked? Or was that fixed from 1-2 years ago?
I haven't used it, just commenting on the instructions that jumped out at me.
That said, I do a lot of browser automation, and have done so for over 15 years using all the tools you might imagine, and as I've researched "plain English" approaches, browser-use comes up a lot, along with other options like stagehand, etc.
Also anything older than 3 or 4 months in the LLM era is worth revisiting, since a tool's approach may be solid, but the models of that point in time may have been the weak point.