Chrome intent

Definition

A Chrome intent is a URL using the `intent://` scheme, which encodes an Android Intent — its target package, action, extras and a fallback URL — inside a link so that a web page can launch a specific app and specify what should happen when that app is absent.

This is the mechanism that predates App Links and still outlives them in practice, because it does the one thing a verified https:// link cannot: it names the package it wants and carries an explicit instruction for the not-installed case. It is also the single most misapplied piece of Android deep linking, because it works flawlessly in the browser it is named after and does nothing at all in several places teams routinely ship it.

The syntax, and the anchor it has to live in

<!-- Chrome only launches an intent from a user gesture.
     A redirect on page load, a meta refresh, or a JS
     navigation without a click will be blocked silently. -->
<a href="intent://product/42
  #Intent
  ;scheme=myapp
  ;package=com.example.app
  ;S.utm_source=newsletter
  ;S.click_id=abc123
  ;S.browser_fallback_url=https%3A%2F%2Fexample.com%2Fproduct%2F42
  ;end">Open in the app</a>

<!-- Unwrapped, with no whitespace — this is the real value: -->
<!-- intent://product/42#Intent;scheme=myapp;package=com.example.app;S.browser_fallback_url=https%3A%2F%2Fexample.com%2Fproduct%2F42;end -->

Everything before #Intent is the data URI the app receives, with scheme= supplying the scheme that the intent:// placeholder stands in for — so the app above is handed myapp://product/42. Everything after it is the Intent itself, as semicolon-separated key-value pairs terminated by ;end. Android parses this with Intent.parseUri(), which is why the grammar is a framework detail rather than a Chrome invention.

What Chrome strips, and what happens with no app

Chrome does not hand your fragment to Android verbatim. Before launching, it sanitises the parsed Intent: it adds the BROWSABLE category, and it discards the component and selector fields. That is deliberate — without it, any web page could target a non-exported activity inside any installed app by naming its class directly. The practical rule is that package= is honoured and component= is a waste of bytes.

The fourth row is the failure teams report as "the link does nothing". Chrome has correctly determined that no activity can handle the intent, found no instruction for that case, and stopped. Always set a fallback, and prefer a real web page over a store listing so that desktop, iOS and every unsupported browser land somewhere useful — the same reasoning as a link fallback chain.

<!-- The BROWSABLE category is not optional. Chrome adds it to the
     intent, so a filter without it will not match. android:exported
     must be true, and is required explicitly from Android 12. -->
<activity android:name=".DeepLinkActivity" android:exported="true">
    <intent-filter>
        <action android:name="android.intent.action.VIEW" />
        <category android:name="android.intent.category.DEFAULT" />
        <category android:name="android.intent.category.BROWSABLE" />
        <data android:scheme="myapp" android:host="product" />
    </intent-filter>
</activity>

Note that the intent above targets a URI scheme, not a verified domain, so the filter is a scheme filter and no assetlinks.json is involved. That is a feature when you need a link to reach a package you name, and a liability in every other respect — see intent filters for why scheme-only filters are hijackable.

Where it works, and what to use instead

A raw WebView does not handle intent:// at all unless the hosting app overrides shouldOverrideUrlLoading, parses the URL and starts the activity itself. Since the hosting app is usually not yours, treat every embedded browser as a surface where the intent will fail — which is the same conclusion reached from the other direction in deep links in in-app browsers.

# The scheme the intent resolves to, sent straight to the package.
adb shell am start -W -a android.intent.action.VIEW   -c android.intent.category.BROWSABLE   -d "myapp://product/42" com.example.app

# What actually resolves an intent URL on the device. Chrome's
# sanitisation is not reproduced here, so this is the upper bound
# of what will work, not a guarantee that Chrome will do it.
adb shell am start -a android.intent.action.VIEW   -d "intent://product/42#Intent;scheme=myapp;package=com.example.app;end"

# Is the package's domain actually verified? If it is, drop the
# intent:// from the happy path and keep it only as a fallback.
adb shell pm get-app-links com.example.app