What it is
The browser extension is a lightweight LoneAct entry point. You can hand the current page or selected text to an Agent so it can summarize, compare, and extract information in the page context, or perform browser actions once authorized.
Chrome / Edge
Install the extension and pair it with your LoneAct account or the LoneAct service on your computer.
Selections and pages
Reads only the text you select, the current page, or a screenshot; it does not silently crawl your browsing history.
Browser control
When needed, the Agent can click, fill forms, and scroll, with every step happening in the visible page.
Install and pair
Install the extension
Install the LoneAct extension from your browser's extension store or an organization distribution package, then pin it to the toolbar.
Connect your device
Open the extension, choose a local or cloud device, and complete pairing or sign-in as prompted.
Choose context
Select text on the page, or click the extension to add the current page or a screenshot to the conversation.
State the goal
Say what to extract, compare, or do; approve page actions when prompted.
What the extension can see
| Data | When it is read | Notes |
|---|---|---|
| Current page text | When you click "Add to context" or ask for it explicitly | Used for the current task; other tabs are not crawled automatically |
| Selection | When you select and add it yourself | Good for quoting an article, ticket, or document |
| Screenshot | When you trigger one or a browser action needs it | Used for visual understanding or to verify page state |
| Page actions | When the Agent asks and you approve | Clicks, form fills, scrolling, and similar actions run on the current page |
Do not grant page actions casually on banking, email, admin, or personal-data pages. Check the current URL, site permissions, and the exact action the Agent wants to perform first.
Where it fits
Fits
Research, comparing page content, filling forms, assisting with admin workflows, and extracting structured information from web pages.
Does not fit
Tasks that need long-term project files and a full terminal; use Local Web or the IDE for those.