For the complete documentation index, see llms.txt. This page is also available as Markdown.

Scan a page

Capture the elements of a page so the Agent can read them: scan, interact to uncover more, review the list, and save.


Click Scan a page in the Interface tab, then open the page you want the Agent to learn. The scan runs directly on your product.

Where the scan runs

The scan works like the In-app Builder: Jimo opens your product and layers itself on top of it, this time as a side panel rather than a full editor. Everything you see and click is your real product.

Any page where the Jimo snippet is installed can be scanned. The browser extension is not required, but it helps: it keeps the panel in sync with the dashboard and forces the load on pages that would otherwise not pick it up.

The scan

Jimo analyzes the page and picks out what your Agent can read: navigation links, buttons, inputs, data displays, text content. Each detected element appears in the side panel with a name and a zone label (Sidebar, Header, Main).

Interact to uncover more

A first scan sees the page at rest. Menus, dropdowns and modals only exist once opened, so the panel invites you to interact with the page like a user: click through your tabs, open your menus, trigger your dialogs. Jimo keeps detecting as you go.

A floating toolbar lets you switch between two modes:

Mode
What your click does

Select

Adds the element you click to the catalog

Interact

Actually clicks, so you can open menus and navigate states without adding anything

Add manually covers anything the detection missed: click it, then point at the element yourself.

Save

Save stores the catalog. The page appears in the Interface list with its element count, and you can now curate each element and attach Skills: see Page details.

Last updated