Skip to content


Switch branches/tags

Latest commit


Git stats


Failed to load latest commit information.
Latest commit message
Commit time

User Idle Detection

Read the draft specification.

This proposed API allows developers to add an event listener for when the user becomes idle (e.g. they don’t interact with the keyboard, mouse or touchscreen, when a screensaver activates or when the screen is locked). Unlike solutions based on monitoring input events this capability extends beyond the site's content area (e.g. when users move to a different window or tab).

Native applications and browser extensions (e.g. Chrome apps, Android apps, Firefox extensions, Edge extensions) use idle detection to notify other users that the user is unreachable (e.g. in chat applications), to show timely alerts (e.g. "welcome back" when a user returns to their task) or to pause media (e.g. to save bandwidth when the user is not present).

The API should provide a means to detect the user's idle status (active, idle, locked), and a power-efficient way to be notified of changes to the status without polling from script.

Feedback: WICG Discourse ThreadIssues

Use cases

  • Chat application: presenting a user's status to other users and delivering notifications to the device where the user is active.
  • Showing timely notifications - e.g. deferring displaying feedback until the user returns to an active state.
  • Updating an outdated service worker when there's no unsaved state by triggering reloading of the tab.

Relationship with other APIs

  • As opposed to the requestIdleCallback, this is not about asynchronously scheduling work when the system is idle.
  • As opposed to the Page Visibility API, this API enables detecting idleness even after a page is no longer visible (e.g. after the page is no longer visible, is the user still around? if i showed a notification, would it be perceived?).


Currently, web apps (e.g. Dropbox’s idle.ts) are constrained to their own content area:

  1. costly polling for input events or
  2. listening to visibility changes

Script can't tell today when a user goes idle outside of its content area (e.g. whether a user is on a different tab or logged out of the computer altogether).


The API assumes that there is some level of engagement between the user, user agent, and operating system of the device in use. This is represented in two dimensions:

  1. The user idle state
  • active/idle - the user has / has not interacted with the user agent for some period of time
  1. The screen idle state
  • locked/unlocked - the system has an active screen lock preventing interaction with the user agent

Distinguishing "active" from "idle" requires heuristics that may differ across user, user agent, and operating system. It should also be a reasonably coarse threshold (See Privacy).

The model intentionally does not formally distinguish between interaction with particular content (i.e. the web page in a tab using the API), the user agent as a whole, or the operating system; this definition is left to the user agent.

Example: The user is interacting with an operating system providing multiple virtual desktops. The user may be actively interacting with one virtual desktop, but unable to see the content of another virtual desktop. A user agent presenting content on the second virtual desktop may report an "idle" state rather than an "active" state.

API Design

The API design is largely inspired by the Sensors API. You can find more about alternatives considered here.


dictionary IdleOptions {
  [EnforceRange] unsigned long threshold;
  AbortSignal signal;

enum UserIdleState {

enum ScreenIdleState {

] interface IdleDetector : EventTarget {
  readonly attribute UserIdleState? userState;
  readonly attribute ScreenIdleState? screenState;
  attribute EventHandler onchange;
  [Exposed=Window] static Promise<PermissionState> requestPermission();
  Promise<void> start(optional IdleOptions options = {});


Here is an example of how to use it (more detailed instructions here):

const main = async () => {
  // Feature detection.
  if (!('IdleDetector' in window)) {
    return console.log('IdleDetector is not available.');
  // Request permission to use the feature.
  if ((await IdleDetector.requestPermission() !== 'granted') {
    return console.log('Idle detection permission not granted.');
  try {
    const controller = new AbortController();
    const signal = controller.signal;
    const idleDetector = new IdleDetector();
    idleDetector.addEventListener('change', () => {
      console.log(`Idle change: ${idleDetector.userState}, ${idleDetector.screenState}.`);
    await idleDetector.start({
      threshold: 60000,
    console.log('IdleDetector is active.');
    window.setTimeout(() => {
      console.log('IdleDetector is stopped.');
    }, 120000);
  } catch (err) {
    // Deal with initialization errors like permission denied,
    // running outside of top-level frame, etc.
    console.error(, err.message);



All platforms (Linux, Windows, macOS, Android, iOS and Chrome OS) support some form of idle detection.

On desktop devices (Chrome OS, Linux, macOS and Windows), a screen saver (from a time when monitors were damaged if the same pixels were lit for an extended period of time) activates after a user-configurable period of inactivity. The operating system may optionally require the user to reauthenticate (i.e. lock the screen) after the screen saver has been activated for a period of time. Both of these events are observable by engines.

On mobile devices (Android and iOS), the screen is dimmed after a few seconds of inactivity (to save battery, not pixels) but this isn't observable by engines (on Android). The screen is eventually turned off (to save further battery) if the user remains inactive for a configurable amount of time (typically 30 seconds), and that is observable by engines. When the screen goes off, the screen is also typically locked (unlockable by Swipe, Pattern, PIN or Password), although it can be configured to be left off but unlocked.


The ability to use this API will be controlled by the new ["idle-detection" permission].

Security and Privacy

See answers to the W3C TAG's security & privacy self-review questionnaire in

The idle state is a global system property and so care must be taken to prevent this from being used as a cross-origin communication or identification channel. This is similar to other APIs which provide access to system events such as Generic Sensors and Geolocation.

A short idle threshold could be used to identify user behavior in another tab. With a short enough threshold an idle state change could be used to measure typing cadence when the user is in another application and thus leak sensitive data such as passwords.

Users with physical or cognitive impairments may require more time to interact with user agents and content. The API should not allow distinguishing such users, or limiting their ability to interact with content any more than existing observation of UI events.

If an implementation restricts the detection threshold it should also restrict how quickly responses to start() are delivered or ensure that the response is cached or otherwise provide a guarantee that rapid polling does not bypass the restriction on data granularity.

To mitigate the exposure of this global state the API should be restricted to top-level frames with a new "idle-detection" permission. This permission can be delegated to sub-frames via Permissions Policy or the sandbox attribute. The top-level frame requirement significantly reduces the number of cross-origin contexts which can observe the state event and thus identify the user through backend communication channels.

Requiring a permission does not completely mitigate the cross-origin identification issue but further reduces the number of sites which are able to participate in such an attack.

A new permission informs the user about the permission the page is requesting.

Screenshot of a permission request dialog
Example icon and text for an "idle-detection" permission request.

Rather than expanding the definition of an existing permission, it is clear to users which sites have access to additional capabilities.

Screenshot of Chromium's page information dialog
Example page information dialog box showing a site with both "notifications" and "idle-detection" permissions granted.

This capability changes the web privacy model by allowing the site to observe a limited amount of information about how the user interacts with their device outside the border of the site's content area. Sites using this permission should present the request in a context which explains the value to the user of being granted this capability.

Implementations that provide a "privacy browsing" mode should not enable this capability in such a mode. This may be satisfied incidentally by not allowing notification permission to be granted. Care should be taken to avoid allowing sites to detect this mode by, for example, randomly delaying the automatic disapproval of the notification permission so that it appears to have been denied by the user.

Alternative Permissions Models

Event Timing Fuzzing

The current idle state and the timing of transitions between states are global state which could be used to identify a user across origin boundaries. For example, two cooperating sites could compare the timestamps at which they observed clients transitioning from "active" to "idle". Two clients which appear to consistently make transitions at the same time have a high likelyhood of being the same user.

A mitigation with the potential to directly address this attack is to add a random delay between when the specified threshold is passed and when the "change" event is fired.

This mitigation was considered and determined to be unacceptable because:

  • Delaying the event by even a small amount drastically reduces the usefulness of the API in chat applications which want to ensure that messages are delivered to the correct device.
  • Even with a large fuzzing factor (e.g. 30 seconds) data collected over a long period of time is still sufficient to identify the user.

Combined with Notification Permission

Rather than defining a new "idle-detection" permission, the definition of the existing "notifications" permission could be expanded to control access to this capability. The most compelling use cases for this capability involve messaging applications, so the "notifications" permission seems appropriate.

The advantage of not defining a new permission type is that it helps to avoid "consent fatigue", in which users are presented with an endlessly increasing number of choices by operating systems, browsers, and web sites. Adding this capability to a related permission such as "notifications" can maintain a user's control while reducing the number of decisions that need to be made.

The disadvantage of not defining a new permission is that it changes the meaning of permission decisions the user had made in the past. While notifications are related to signals of user presence the ability to monitor this state may not be something the user considers when granting this permission.

There is a middle-ground in which the permission prompt text for the "notifications" permission is updated to explain this new capability but both capabilities are still controlled by a single permission. Implementations could internally track whether the user has seen the updated prompt and require a site re-request the "notifications" permission before it has access to the new capability.

Prior Work

  • Chrome's chrome.idle API for apps/extensions, which is a direct inspiration for this proposal.
    • Also exposed to Extensions in Firefox MDN
    • And Edge
    • That API has a global (per-execution-context) threshold and one source of events. This makes it difficult for two components on the same page to implement different thresholds.
  • Attempts to do this from JS running on the page:


A proposal for an idle detection and notification API for the web







No releases published


No packages published