What it does

Autotab is a browser automation tool that watches you click around in Chrome, then generates Python code using Selenium to replay those actions. You run autotab record, perform your task, copy the generated code into a boilerplate file, and later run autotab play to execute it. The pitch is auditable automations — you get to see and edit the code rather than trusting a black box.

The interesting bit

The “record to code” loop is the hook, but the real friction is in the setup: ChromeDriver, a YAML config file, an autotab API key, and a browser extension sidepanel toggled with keyboard shortcuts. The generated code is explicitly meant to be pasted and formatted by hand — no hidden pipeline, which is either refreshing or tedious depending on your patience.

Key highlights

• Records clicks, typing, hovers, and clipboard operations via a Chrome sidepanel extension
• Outputs plain Python/Selenium scripts to an agents/ folder for manual cleanup
• Requires autotab API key (free) and uses the project’s OpenAI API key, not yours
• Stores Google login cookies locally to skip 2FA on subsequent runs
• Alpha software; the README warns of breaking changes and “exciting new features regularly”

Caveats

• Deprecated as of December 2023: the repo is no longer supported while the team builds a “no dependencies” V1
• Interactions often fail if the Chrome window loses focus; headless/cloud version is promised but not shipped
• By default logs DOM state and model responses during recording; you must opt out via dashboard settings
• Cookie storage (google_cookies.json) contains sensitive data and needs manual safeguarding

Verdict

Worth a look if you want inspectable browser automation and don’t mind wiring together Selenium, API keys, and a Chrome extension. Skip it if you need something maintained, headless, or turnkey — this starter repo is explicitly end-of-life.