Browser companions

Attribute website time from the browser itself

The ActivityPulse desktop agent can read your active tab’s address through operating-system accessibility APIs, but on Chromium-based browsers that read is slow and frequently returns nothing. The browser companion reports the tab directly instead. It sends the address and title of your active tab to the agent on your own machine over local port 5321, and reads nothing else on the page.

Availability

Google Chrome
Coming soon
Microsoft Edge
Coming soon
Mozilla Firefox
Not supported
Apple Safari
Not supported

Chrome and Edge share one build and are awaiting store review. Until the listings are live, the companion can be loaded unpacked from the repository for testing; Firefox and Safari need their own manifests and are not built yet.

What it does and does not reach
The companion is deliberately small: two browser events, one local request, no server of its own.

It reads tab.url and tab.title of the active tab. There is no content script, so page contents, form fields and key events are not reachable from it. Browser-internal pages are skipped, and private windows are excluded by the manifest rather than by a preference — the browser will not run the companion in one.

Everything it reads goes to the desktop agent on the same machine, over loopback. The agent applies your organisation’s resolved capture policy to it exactly as it does to anything else it records.

Questions

What does the browser companion read?
The address and title of your active tab, and nothing else. The companion has no content script, so page contents, form fields and key events are not reachable from it. It sends what it reads to the ActivityPulse desktop agent on your own machine over local port 5321 — it has no server of its own to send anything to.
Does it run in private or incognito windows?
No. The companion's manifest sets "incognito": "not_allowed", so the browser will not run it in a private window at all — the option to allow it is not offered. When a private window is focused the companion tells the agent that nothing is attributable, without being able to see what is in it.
Why is a browser companion needed at all?
Without it the agent reads the address bar through operating-system accessibility APIs. That works, but Chromium keeps its accessibility tree lazily disabled, so the read is slow and often returns nothing. The companion reports the tab directly from the browser, which is both faster and reliable. The agent falls back to the accessibility read whenever the companion is absent.
Can my organisation install it for me?
On Windows, an administrator can have the desktop agent add the companion to Chrome's and Edge's enterprise policy for your user account, so it installs on the next browser launch. This is off unless your organisation turns it on. On macOS the same result requires an MDM configuration profile, because Chrome only reads policy from a managed preferences file.

See where the hours actually went

Free for up to three people. Windows and macOS. No keystroke content on any plan, and screenshots off unless an administrator turns them on.