Planetary-Computers/autotab-starter ? reverse-engineered prompt

Build me a Python tool that lets me record browser actions and then play them back as an automation script.

I want to open Chrome, log in with my autotab API key from a local config file, and then record clicks, typing, and element selections from a side panel. When I finish recording, it should generate a Python agent file I can edit and run later. I also want a simple play command that runs one of those saved agents in Chrome, with a default agent if I do not pass a name.

Please make it easy to set up with a quick install step, a sample config file, and clear instructions for Chrome and chromedriver. If you need to check current docs or best practices online, go ahead. Keep it focused on making browser automations that are easy to audit and reuse, and make sure the code is organized so someone can add more agents later.