Report Abuse

Basic Information

autotab-starter is a developer-focused starter project for creating auditable browser automations with AI. It lets users demonstrate interactions in a Chrome browser and convert those point-and-click or selection-based demonstrations into runnable Python automation code. The repo provides a CLI with commands to record browser sessions and to play back generated agents, generates agent boilerplate files in an agents/ folder, and uses a browser extension sidepanel during recording to capture actions, DOM state, and model responses. Configuration is managed via a .autotab.yaml file and the project stores cookies to simplify repeated logins. The README notes the project is an alpha release, that the repository is transitioning to a newer V1 with no dependencies and that the repo itself will no longer be actively supported, while existing generated Python code remains usable.

Links

Categorization

App Details

Features
The README documents key features such as a recording mode and a select mode that capture clicks, typing, element selection, hover, clipboard copy and injected text. It provides CLI commands autotab record and autotab play to record new automations and run existing agent scripts. The tool launches a Chrome session controlled by Selenium and opens an extension sidepanel for interactive recording. It scaffolds Python agent files in the agents/ directory, requires a .autotab.yaml configuration and Chrome/Chromedriver, and offers keyboard shortcuts for recording and selection. By default it logs application state, DOM, and model responses for auditable runs, and it supports disabling DOM and state logging via settings. The README includes setup steps, dependency installation via make install, and notes about handling cookies and 2FA.
Use Cases
This repository accelerates building repeatable, auditable browser automations for developers and technical users by converting manual browser demonstrations into executable Python scripts. It reduces time to create automation by capturing user interactions and producing ready-to-run agent boilerplate that can be edited and debugged like any Python program. Built-in logging of DOM and model responses aids auditing and troubleshooting. Cookie storage helps avoid repeated 2FA prompts during repeated runs. The CLI workflow lets users iterate quickly between recording and playback. The README also warns about legal and ethical considerations for scraping and automation and emphasizes the project is alpha and transitioning to a newer version, so users should review generated code and configuration before production use.

Please fill the required fields*