Autotab

Stale
GitHub Python MIT

Description

Open-source framework for building browser agents for real-world tasks, learning from user demonstrations to automate web interactions.

Key Features

  • Point-and-click browser automation recording that converts user actions into auditable Python code
  • Chrome browser integration via Selenium for real-world web interaction automation
  • Record and playback workflow: record browser actions then replay them as automated scripts
  • Chrome extension sidepanel for recording clicks, typing, and element selection
  • Agent-based architecture with configurable agent profiles and reusable automation scripts
  • Cookie management for handling Google login and 2FA across automation runs

Use Cases

💡 Automating repetitive browser-based tasks without writing code manually
💡 Creating auditable browser automation scripts from point-and-click demonstrations
💡 Web scraping and data collection from real-world websites with login support
💡 Building browser agents that can execute complex multi-step web workflows
💡 Rapid prototyping of browser automation for testing and quality assurance

Quick Start

Clone the repo, run `make install`, install chromedriver via Homebrew, configure credentials in `.autotab.yaml`, then run `autotab record` to start recording browser actions and `autotab play --agent <name>` to replay them.

Related Projects