Update (Dec 12, 2023)

Thanks for trying out autotab! Over the last few weeks we’ve learned from so many of you and are excited to start rolling out a smoother and more user-friendly autotab V1 (no dependencies!).

As we transition, this repo will no longer be supported. The good news is that any Python code you’ve generated with autotab will of course work as usual.

Join our Discord channel to follow along and for a chance to be one of the first to try out new updates!

autotab

Welcome to autotab! autotab makes it easy to create auditable browser automations using AI. Go from a point & click demonstration in the browser to live code for those actions in seconds.

Note: This project is alpha release and actively being developed. Expect breaking changes and exciting new features regularly!

Quickstart

It usually takes 5-10 minutes to get everything set up (including gathering passwords and installing dependencies). You must have the Chrome browser installed, and we recommend setting up a Python virtual environment:

git clone https://github.com/Planetary-Computers/autotab-starter.git
cd autotab-starter
# Recommended: Setup a Python virtual environment
make install
brew install --cask chromedriver

Configuration

Configure your credentials: Create a .autotab.yaml file following the example in .example.autotab.yaml. (~3 minutes)

Run

Run autotab record to open a new browser window where you can start recording your actions.

Note: When you run autotab record, an automation will first try to log you in to autotab using the autotab_api_key from your .autotab.yaml file. You need to be logged in to autotab to use the extension (and our Open AI API key). You log in to autotab record using your autotab API key which you can get for free at autotab.com/dashboard.

Usage

Recording an automation

To record a new automation, run autotab record. You can optionally add a --agent argument. This will launch a Chrome session controlled by Selenium and then log you in to Google and open the autotab extension in the sidepanel.

If the sidepanel does not open, type Command - Shift - Y to open the sidepanel.

Once the sidepanel is open, you can use record mode to record clicks and typing (Command - E) or select mode (Command I) to select an element to be hovered, copied to clipboard or to inject text into.

At the end of recording make sure to copy all the code. autotab will have created a .py file in the agents/ folder with boilerplate code. Paste the code in there, format it and then your agent is ready to run!

Running an automation