BrowserWorker by Vector Agent per Tarun Goyal

BrowserWorker turns plain-language instructions into user-directed browser workflows from Firefox's sidebar.

BrowserWorker can inspect task-related pages, navigate, open and switch tabs, scroll, fill fields, click controls, extract page content, and summarize results. You choose the task, websites, and AI provider.

Key features:

• Works across task-related browser tabs.
• Supports OpenAI, DeepSeek, OpenRouter, and OpenAI-compatible model servers.
• Offers text-only and vision-capable modes, depending on the selected model.
• Stores settings, API credentials, consent state, and conversation history in the local Firefox profile.
• Requests website access only when the user starts automation.
• Pauses detected consequential final actions for separate, single-use confirmation.
• Contains no advertising or Vector Agent analytics backend.

To perform a task, BrowserWorker sends the prompt and relevant browser context directly to the AI provider selected by the user. Vision mode may also send a screenshot of the visible tab.

BrowserWorker requires a user-supplied provider API credential or an OpenAI-compatible endpoint. Provider usage charges may apply under the provider's terms.

Review the privacy policy before use, and avoid sensitive pages unless you are willing to share their relevant content with your selected provider.