Allow google.com and, if you continue to YouTube later, youtube.com.
Google changes often, so this demo is intentionally best-effort, not a fake “works forever” promise.
What this demo shows
This page is a guided scenario using RTOForm and named command actions.
It opens or reuses the controlled tabKey="google" lane, checks consent, shows the remote overlay cursor,
types a query, submits the search, reads the first result, then returns the story to this master tab.
The script does not execute arbitrary JS in the target tab. It only calls predefined safe actions
implemented by the extension content script.
Tour steps
Wake up RTO
Detect the extension and register this page as the master return point.
Open or reuse Google
One stable tabKey="google", one logical lane, no duplicate-tab confetti.
Handle consent if Google asks nicely
Look for a common consent button and move on if there is nothing to do.Scanning consent UI…
Move the remote cursor and type
Use overlays to guide the eye, then type the query and verify the actual field value.
Launch the search
Click the submit control or fall back to navigation when the button plays hide-and-seek.
Read the first visible result
Collect title, link, and thumbnail when available, then mirror them here.
First Google result
Return to the master tab
Replace this step with the extracted result and ask RTO to bring you back home.
Log output
Why this demo is fun to watch
The master tab registers itself as a return target.
RTO opens or reuses the controlled Google tab with the same tabKey.
The script uses overlays and labels so the action is visible instead of silently teleporting values.
The query is typed and verified before the search is launched.
The first result is read back and rendered here like a tiny post-mission report.
Plugin responsibilities
The background owns tab lifecycle, allowlist checks, and master-tab routing.
The content script executes named DOM actions such as waitForSelector, typeText, getText, and overlay actions.