android/skills

android-intent-security

Best practices for Android Intent security.

Voir la source

Mode d’emploi

  1. Copiez la commande et examinez le paquet avant de l’installer dans votre environnement agent.
  2. Consultez le document original pour les outils, identifiants et environnements requis.
  3. Décrivez les entrées et le livrable souhaité ; comparez la sortie aux sources.
npx skills add https://github.com/android/skills --skill android-intent-security

Avant de commencer

  • L’installation n’inclut ni abonnement au modèle, ni quota API, ni accès aux fichiers.
  • Étoiles et installations reflètent l’activité, pas un test indépendant de qualité ou de sécurité.
Document Skill original

Rendu depuis le dépôt source en conservant titres, exemples, code, tableaux, liens et images.

This skill provides guidelines and patterns to secure Android components (Activities, Services, Broadcast Receivers, Content Providers) and handle Intents safely, preventing privilege escalation and unauthorized access.

Glossary

  • Intent: An asynchronous messaging object used to request an action from another app component.
  • Exported component: A component (android:exported="true") that can be launched by other apps on the device.
  • Sticky intent: A broadcast intent that remains in the system cache after it's sent, allowing any app to retrieve its contents.
  • Signature permission: A permission whose protection level is set to signature, granted only to apps signed with the same developer key.
  • onNewIntent: An activity lifecycle callback invoked when an activity is launched with FLAG_ACTIVITY_SINGLE_TOP and is already running at the top of the history stack.
  • PendingIntent: A token granted to a foreign application (for example, system services) allowing it to execute a predefined Intent with the creator's permissions.
  • Mutable PendingIntent: A PendingIntent whose underlying Intent parameters can be modified by the receiving application.
  • ContentProvider: A component that encapsulates data and provides it to other applications via standard query/insert interfaces.
  • IntentSanitizer: A utility class in AndroidX Core used to build a safe, sanitized copy of an incoming Intent by filtering out unauthorized components, actions, or extras.
  • Intent redirection (forwarding): A vulnerability where an application receives an intent from an untrusted source and uses it to launch a private, non-exported component.

Prerequisites

  • The agent MUST be able to describe the function and security implications of onCreate, onNewIntent, and the singleTop launch mode.
  • The agent MUST be able to declare <activity>, <service>, <receiver>, and <provider> tags in AndroidManifest.xml and define their android:exported and android:permission attributes.
  • The agent MUST be able to implement signature verification checks using PackageManager.

Limitations

  • This skill focuses on local inter-component and inter-app communication security on the Android platform.
  • This skill doesn't cover network security, web integration, or host-to-server security.

Setup and dependencies

  • Android SDK: Minimum API Level 23 (Android 6.0) is required for standard hardware-backed keystore operations and component validation.
  • AndroidX Core Library: androidx.core:core:1.9.0 or higher is mandatory to leverage IntentSanitizer.
  • Standard API access: Standard Android PackageManager APIs are required for runtime component verification.

Intent security logic and decisions

1. Intent routing comparison

Evaluate the security features of different intent delivery methods:

Intent Delivery MethodScopeRecommended Use Case
Explicit Intent (Internal)App PrivateLaunching internal activities/services
Implicit IntentSystem WideLaunching system camera, dialer, or sharing
Local Broadcasts (LocalBroadcastManager) (DEPRECATED)App PrivateInternal asynchronous event routing. Deprecated: Use in-app observers like Kotlin Flows/SharedFlow, LiveData, or reactive patterns instead.
System BroadcastsSystem WideReceiving system events (NFC, Bluetooth)

2. PendingIntent mutability flag options

Evaluate the security implications of PendingIntent mutability flags:

Flag NameMutabilityRecommended Use Case
PendingIntent.FLAG_IMMUTABLEImmutableDefault for almost all PendingIntents, such as alarms and notifications
PendingIntent.FLAG_MUTABLEMutableInline notifications replies, slice actions (requires explicit target intent)

3. Intent handling and redirection logic

IF (the component receives a nested Intent as an extra) { IF (AndroidX Core 1.9.0+ and higher is available) { MUST construct an IntentSanitizer to explicitly allowlist components, actions, data, and extras. MUST call sanitizeByThrowing() or sanitizeByFiltering() before launching. } ELSE { MUST reject or strip URI permission grant flags (FLAG_GRANT_READ_URI_PERMISSION, FLAG_GRANT_WRITE_URI_PERMISSION, FLAG_GRANT_PERSISTABLE_URI_PERMISSION, FLAG_GRANT_PREFIX_URI_PERMISSION). MUST resolve the target component using resolveActivity() and verify that both its packageName and className match an explicitly allowlisted component. } NEVER launch the nested Intent directly without validation. } ELSE IF (the component handles broadcasts) { MUST rely on the system's Protected Broadcast mechanism for system events (which guarantees the sender is the system framework). MUST protect custom receivers with signature-level permissions or use RECEIVER_NOT_EXPORTED for dynamic receivers to restrict the sender. IF (using ContextCompat.registerReceiver with RECEIVER_NOT_EXPORTED AND minSdk < 33) { MUST verify <permission android:name="${applicationId}.DYNAMIC_RECEIVER_NOT_EXPORTED_PERMISSION" android:protectionLevel="signature" /> and <uses-permission android:name="${applicationId}.DYNAMIC_RECEIVER_NOT_EXPORTED_PERMISSION" /> are declared in AndroidManifest.xml. } }

4. PendingIntent security logic

IF (a PendingIntent is created for delivery to another application) { MUST use PendingIntent.FLAG_IMMUTABLE by default. IF (the PendingIntent must be mutable) { MUST set the explicit target component or package name on the base Intent. NEVER create an implicit, mutable PendingIntent. } }

5. ContentProvider security logic

IF (the ContentProvider is only for internal app use) { MUST set android:exported="false". } ELSE { MUST protect it with android:readPermission and android:writePermission. MUST set android:grantUriPermissions="false" unless temporary URL access is strictly required. }

6. Service caller verification logic

IF (an exported service communicates with trusted sister/partner apps) { MUST expose a cross-process interface (such as an AIDL Stub or Messenger) rather than a local Binder subclass, which only works within the same process. MUST verify the caller on every IPC request---using Binder.getCallingUid() inside an AIDL Stub method (or onTransact()) or Message.sendingUid in a Messenger Handler---and verify that the calling package signature fingerprint matches your trusted certificate hash. }

Code and configuration patterns

1. Safe intent redirection (manual verification)

Validate the target of a nested intent before launching it when modern sanitization libraries are unavailable.

  • Expected Inputs:
  • An incoming Intent containing a nested Intent extra named EXTRA_NESTED_INTENT.
  • Expected Outputs:
  • Launches the target component if safe; throws SecurityException if validation fails.
kotlin
fun safeIntentRedirectionManual() {
    val nestedIntent = IntentCompat.getParcelableExtra(intent, "EXTRA_NESTED_INTENT", Intent::class.java)
    if (nestedIntent != null) {
        // 1. Check for URI permission grants to prevent URI permission bypass
        val hasUriPermissionGrants = (
            nestedIntent.flags and (
                Intent.FLAG_GRANT_READ_URI_PERMISSION or
                    Intent.FLAG_GRANT_WRITE_URI_PERMISSION or
                    Intent.FLAG_GRANT_PERSISTABLE_URI_PERMISSION or
                    Intent.FLAG_GRANT_PREFIX_URI_PERMISSION
                )
            ) != 0
        if (hasUriPermissionGrants) {
            throw SecurityException("Nested intent contains forbidden URI permission grant flags!")
        }

        // 2. Clear any selector intent that could override resolution
        nestedIntent.selector = null

        // 3. Resolve the target activity and verify both packageName and className
        //    match the expected allowlisted destination
        val target = nestedIntent.resolveActivity(packageManager)
        if (target != null &&
            target.packageName == packageName &&
            target.className == TargetActivity::class.java.name
        ) {
            // 4. Explicitly pin the resolved component to prevent intent interception
            nestedIntent.component = target
            startActivity(nestedIntent)
        } else {
            throw SecurityException("Invalid or disallowed redirect target: $target")
        }
    }
}

<br />

2. Safe intent redirection using IntentSanitizer

Filter or reject dynamic intents using AndroidX IntentSanitizer (AndroidX Core 1.9.0+).

  • Expected Inputs:
  • An untrusted incoming Intent.
  • Expected Outputs:
  • Intent: A sanitized copy containing only allowlisted components, categories, and actions. Throws SecurityException on violations if using sanitizeByThrowing().
kotlin
fun safeIntentRedirectionSanitizer() {
    // Allowlist: anything NOT listed is rejected (only component is required; other fields are optional).
    // No allowFlags(): every flag, including FLAG_GRANT_*_URI_PERMISSION, is rejected.
    val sanitizer = IntentSanitizer.Builder()
        .allowComponent(ComponentName(this, TargetActivity::class.java))
        .allowAction(Intent.ACTION_VIEW)
        .allowData { it.scheme == ContentResolver.SCHEME_CONTENT && it.authority == "$packageName.provider" }
        .allowType("text/plain")
        .allowExtra("user_display_name", String::class.java) { it.length <= 100 } // type and content
        .build()

    val safeIntent = try {
        val untrusted = IntentCompat.getParcelableExtra(intent, "EXTRA_NESTED_INTENT", Intent::class.java)
            ?: return
        sanitizer.sanitizeByThrowing(untrusted)
    } catch (_: SecurityException) {
        Log.w("SECURITY_ALERT", "Blocked non-allowlisted nested intent")
        return
    } catch (_: BadParcelableException) { // e.g. an extra naming an unknown Parcelable/Serializable
        Log.w("SECURITY_ALERT", "Malformed nested intent")
        return
    }

    // The sanitizer doesn't enforce presence; check what the target actually requires.
    if (safeIntent.getStringExtra("user_display_name") == null) {
        Log.w("SECURITY_ALERT", "Nested intent missing user_display_name")
        return
    }
    startActivity(safeIntent)
}

<br />

3. Custom signature permission protection

Declare a custom signature-level permission in the manifest to secure family app communication.

  • Expected Inputs: Manifest configuration.
  • Expected Outputs: An activity that can only be launched by apps signed with the same developer certificate.
xml
<permission
    android:name="com.example.snippets.permission.INTERNAL_COMMUNICATION"
    android:protectionLevel="signature" />

<br />

xml
<activity
    android:name=".intents.InternalSharingActivity"
    android:exported="true"
    android:permission="com.example.snippets.permission.INTERNAL_COMMUNICATION">
    <intent-filter>
        <action android:name="com.example.snippets.ACTION_SHARE" />
        <category android:name="android.intent.category.DEFAULT" />
    </intent-filter>
</activity>

<br />

4. Safe onNewIntent lifecycle verification (warm boot protection)

Ensure that activities reusing dynamic intents (for example, in background launch paths) apply the same strict security filters inside onNewIntent.

  • Expected Inputs:
  • newIntent (Intent): The newly delivered intent.
  • Expected Outputs:
  • Executes processing logic only if the new intent passes security validation.
kotlin
override fun onNewIntent(newIntent: Intent) {
    super.onNewIntent(newIntent)

    // Set the intent to ensure intent returns the new one
    intent = newIntent

    // Validate the intent payload
    if (validateIntent(newIntent)) {
        processIntentPayload(newIntent)
    } else {
        Log.w("SECURITY_ALERT", "Received invalid or insecure intent during warm boot")
    }
}

private fun validateIntent(intent: Intent): Boolean {
    return intent.hasExtra("VALID_PAYLOAD_MARKER")
}

<br />

5. Secure PendingIntent creation

Enforce immutability unless mutability is explicitly required.

  • Expected Inputs (Immutable): An intent target.
  • Expected Outputs (Immutable): A PendingIntent that cannot be altered by the receiver.
  • Expected Inputs (Mutable): An intent with an explicit component set.
  • Expected Outputs (Mutable): A mutable PendingIntent locked to a specific receiver component to prevent hijacking.
kotlin
fun createPendingIntents(context: Context) {
    // 1. Secure Immutable PendingIntent (Default)
    val intent = Intent(context, TargetActivity::class.java)
    val pendingIntent = PendingIntent.getActivity(
        context,
        0,
        intent,
        PendingIntent.FLAG_IMMUTABLE or PendingIntent.FLAG_UPDATE_CURRENT
    )

    // 2. Secure Mutable PendingIntent (e.g., Notification Direct Reply)
    val mutableIntent = Intent().apply {
        // MUST set explicit target component to prevent redirection hijacking
        component = ComponentName(context, ReplyReceiver::class.java)
    }
    val mutablePendingIntent = PendingIntent.getBroadcast(
        context,
        0,
        mutableIntent,
        PendingIntent.FLAG_MUTABLE or PendingIntent.FLAG_UPDATE_CURRENT
    )
}

<br />

6. Secure ContentProvider configuration and queries

Expose a ContentProvider securely and parameterize queries to prevent SQL injection.

  • Expected Inputs:
  • uri (Uri): The query URI.
  • projection (String[]): Columns to retrieve.
  • selection (String): Query criteria.
  • selectionArgs (String[]): Values mapping to selection placeholders (?).
  • Expected Outputs:
  • Cursor: Filtered query results, strictly bound to projection maps.
xml
<provider
    android:name=".intents.SecureDataProvider"
    android:authorities="com.example.snippets.provider"
    android:exported="true"
    android:readPermission="com.example.snippets.permission.READ_DATA"
    android:writePermission="com.example.snippets.permission.WRITE_DATA"
    android:grantUriPermissions="false" />

<br />

kotlin
override fun query(
    uri: Uri,
    projection: Array<String>?,
    selection: String?,
    selectionArgs: Array<String>?,
    sortOrder: String?
): Cursor? {
    val queryBuilder = SQLiteQueryBuilder()
    queryBuilder.tables = tableName
    // Strict projection map to prevent querying unauthorized columns
    queryBuilder.projectionMap = mapOf(
        "_id" to "_id",
        "display_name" to "display_name"
    )
    // Enable strict validation (always available since minSdk is 36)
    queryBuilder.setStrict(true)
    queryBuilder.setStrictColumns(true)
    queryBuilder.setStrictGrammar(true)

    // MUST parameterize selection criteria; NEVER append selection strings directly
    val db = dbHelper.readableDatabase
    return queryBuilder.query(db, projection, selection, selectionArgs, null, null, sortOrder)
}

<br />

7. Service caller signature verification

Verify the calling application's signing certificate on incoming cross-process IPC calls in an exported bound service.

  • Expected Inputs:
  • An incoming cross-process Binder transaction (for example, an AIDL Stub method call or onTransact()).
  • Expected Outputs:
  • Executes the requested service operation if the calling UID's signing certificate matches the trusted partner hash; throws SecurityException otherwise.
kotlin
class SecureBoundService : Service() {
    companion object {
        // Expected SHA-256 hash of the trusted app's signing certificate (Base64 encoded)
        private const val TRUSTED_PARTNER_SHA256 = "A1B2C3D4E5F6G7H8I9J0K1L2M3N4O5P6Q7R8S9T0U1V="
    }

    override fun onBind(intent: Intent): IBinder {
        // Return the binder. Do NOT perform signature verification in onBind() because
        // the binder connection is cached by Android, which can bypass checks on subsequent binds.
        return LocalBinder()
    }

    private fun enforceTrustedCaller() {
        val callingUid = Binder.getCallingUid()
        // Allow calls from the same application
        if (callingUid == Process.myUid()) {
            return
        }
        val pm = packageManager
        val packages = pm.getPackagesForUid(callingUid)

        if (packages.isNullOrEmpty() || !verifySignature(pm, packages[0])) {
            throw SecurityException("Access Denied: Caller signature is untrusted.")
        }
    }

    private fun verifySignature(pm: PackageManager, packageName: String): Boolean {
        try {
            val trustedSha256Raw = Base64.decode(TRUSTED_PARTNER_SHA256, Base64.DEFAULT)
            // API 28+ handles rotated certificates and avoids manual hashing.
            // Since minSdk is 36, this is always available.
            return pm.hasSigningCertificate(packageName, trustedSha256Raw, PackageManager.CERT_INPUT_SHA256)
        } catch (e: Exception) {
            Log.e("SECURITY_ERROR", "Verification failed for package: $packageName", e)
        }
        return false
    }

    inner class LocalBinder : Binder() {
        fun doSecureWork() {
            // Verify caller identity on every transaction method call
            enforceTrustedCaller()
            // Safe to proceed with sensitive operations
        }
    }
}

<br />

Error handling

Handle component binding, database queries, and intent redirection failures securely to avoid exposing internal structures.

kotlin
fun safeErrorHandling(callingPackage: String?) {
    try {
        val payload = intent.getStringExtra("DATA_EXTRA") ?: throw IllegalArgumentException("Payload parameter missing.")
        // Create a specific target intent using the validated payload
        val targetIntent = Intent(this, TargetActivity::class.java).apply {
            putExtra("SECURE_PAYLOAD", payload)
        }
        startActivity(targetIntent)
    } catch (e: SecurityException) {
        // MUST log security violations for audit, but NEVER expose exception details to the user.
        Log.e("SECURITY_ERROR", "Unauthorized component transition blocked. Calling Package: ${callingPackage ?: "Unknown"}", e)
        // MUST provide generic user feedback.
        showFeedbackToUser("Process request failed: Access Denied.")
    } catch (e: IllegalArgumentException) {
        Log.w("INTEGRITY_WARNING", "Missing intent parameter", e)
    }
}

<br />

// Secure handling of ContentProvider queries on the client side: try { val cursor = contentResolver.query(providerUri, projection, selection, selectionArgs, null) } catch (e: SQLiteException) { Log.e("PROVIDER_ERROR", "ContentProvider database query failed", e) // Secure handling: prevent raw query syntax details from leaking to UI }

Reporting guidelines

When this skill is executed to apply security hardening updates to a codebase, the agent MUST generate a structured "Best Practices and Security Alignment Update" report for the developer. The report must be written to the session artifact folder (or printed in the final response) and include:

  1. Security alignment area: The category of improvement applied (for example, Safe Intent Redirection, Secure PendingIntent Configuration, ContentProvider Data Guarding).
  2. Impact and priority: The potential safety risk addressed by the update (for example, Component Hijacking Prevention, Private Data Isolation).
  3. Scope of changes: A list of all modified classes, XML files, and dependencies.
  4. Implementation summary: Concrete details of the solution (for example, "Updated nested intent parsing to use the IntentSanitizer API with a strict component allowlist").
  5. Code diff: Standard unified diffs showing the exact modifications.

Best practices and security alignment update template

Use the following markdown template when reporting changes to developers:

Best practices and security alignment update: [Security Alignment Area]

  • Improvement Description: [Brief description of the hardening update and why it's recommended]
  • Priority Level: [High / Medium / Low]
  • Alignment Action: [Summary of updates, for example, converted to FLAG_IMMUTABLE]

Files modified

  • [Relative path to File 1]
  • [Relative path to File 2]

Implementation diff

diff
    // Insert Unified Diff here

#### Testing and verification

1. \[Step 1 to verify the component behaves correctly, for example, run component unit test\]
2. \[Step 2 to verify regression safety\] \`\`\`

*** ** * ** ***

## Antipatterns

- **NEVER** launch a nested `Intent` received from an untrusted source without verifying that its resolved component (`packageName` and `className`) matches an explicit allowlist and rejecting or stripping URI permission grant flags.
- **NEVER** use sticky broadcasts (`sendStickyBroadcast`).
- **NEVER** assume an exported component is safe because it runs in a background thread or performs internal checks.
- **NEVER** expose sensitive functionalities (like SSO authentication or payment processors) to components without signature-level permission restrictions.
- **NEVER** process incoming intents in `onNewIntent` without applying the same security controls as `onCreate`.
- **NEVER** create a mutable `PendingIntent` without setting an explicit target component in the base `Intent`.
- **NEVER** use dynamic string concatenation to construct selection blocks inside a `ContentProvider` query.
- **NEVER** use `Binder.getCallingUid` inside a `BroadcastReceiver.onReceive` to identify the sender of a broadcast, as it returns the receiver's own UID, not the sender's.

## Best Practices

- **MUST** explicitly set `android:exported="false"` for all components that don't need external communication.
- **MUST** protect all exported components with custom permissions utilizing `android:protectionLevel="signature"` when communicating between family apps.
- **MUST** validate all incoming intent extras and handle missing parameters gracefully to prevent crashes.
- **MUST** rely on the system's **Protected Broadcast** mechanism for system events (for example, boot completed, package changes), as the system prevents untrusted apps from spoofing these actions.
- **MUST** protect custom broadcasts with signature-level permissions or use `RECEIVER_NOT_EXPORTED` for dynamic receivers to restrict the sender identity (and when `minSdk < 33`, **MUST** verify that `<permission android:name="${applicationId}.DYNAMIC_RECEIVER_NOT_EXPORTED_PERMISSION" android:protectionLevel="signature" />` and `<uses-permission android:name="${applicationId}.DYNAMIC_RECEIVER_NOT_EXPORTED_PERMISSION" />` are present in `AndroidManifest.xml` when using `ContextCompat.registerReceiver`).
- **MUST** call `setIntent(newIntent)` inside `onNewIntent()` before processing payloads to keep active references updated.
- **MUST** use `PendingIntent.FLAG_IMMUTABLE` by default when constructing `PendingIntent` instances.
- **MUST** protect exported `ContentProviders` with `readPermission` and `writePermission`.
- **MUST** enforce parameterized selection structures in `ContentProvider` query/update methods.
- **MUST** verify the calling application's signing certificate on incoming IPC transactions at runtime inside exported bound services using a cross-process interface (such as an AIDL `Stub` or `Messenger`).
- **MUST** use `androidx.core.content.IntentSanitizer` to sanitize incoming dynamic intents before redirection, if AndroidX Core 1.9.0+ is imported in the project.
du même dépôt

Autres Skills

Tous les Skills
android
Communauté

android-cli

Provides instructions for installing and using the android CLI. The android command-line tool is a critical tool for Android development and helps you create new Android projects, install APKs onto devices and emulators and launch and run apps on them, manage and interact with Android virtual devices (including screenshots and UI inspection), manage Android SDK components, look up official Android documentation, and discover and install official Android skills.

Installations / 1h
1
GitHub Stars
10,3 k
Mis à jour
4 nov.
android
Communauté

android-for-cars-parked-apps

Adapt existing Android video streaming apps, games, and web browsers to run as parked apps on Android Auto and Android Automotive OS. Covers manifest configuration, Google Play hardware feature declarations, compatibility mode, back navigation affordances, orientation and display continuity, controllable system bar insets, irregular display cutouts, driver distraction lifecycle handling, MediaSession playback restrictions with CarUxRestrictionsManager, background audio while driving, and browser

Installations / 1h
1
GitHub Stars
10,3 k
Mis à jour
4 nov.
android
Communauté

android-profiler

'Records and analyzes Android performance data and diagnoses performance bottlenecks across user apps, system apps, and system services, even if the user does not explicitly mention profiling or traces. Use when: recording system traces, CPU call stacks, Vulkan GPU/DRM memory, or Java/native heap profiles; analyzing Perfetto traces for UI jank, frame drops, ANRs, or app startup latency issues; investigating CPU bottlenecks, IPC/Binder storms, or thread scheduling stalls; diagnosing battery drain

Installations / 1h
1
GitHub Stars
10,3 k
Mis à jour
4 nov.
android
Communauté

anr-audit-and-fix

Audits Android codebases for ANR risks and diagnoses ANR stack dumps. Use this skill when scanning Android code for ANR risks or triaging ANR stack dumps. Don't use for non-ANR crashes, dropped frames, general CPU profiling, or Perfetto system traces.

Installations / 1h
1
GitHub Stars
10,3 k
Mis à jour
4 nov.