root@gulertech:~$ GULER TECH LTD

MASVS-AUTH: Testing Authentication and Authorisation in Android and iOS Apps

2026-11-09 MASVS-AUTH-1 · MASVS-AUTH-2 · MASVS-AUTH-3 · MASWE-0018 · MASWE-0025
androidiosmasvs-authauthenticationmethodology

This is the entry point to our series on OWASP MASVS-AUTH. The control group has three requirements:

  • MASVS-AUTH-1: “The app uses secure authentication and authorization protocols and follows the relevant best practices.”
  • MASVS-AUTH-2: “The app performs local authentication securely according to the platform best practices.”
  • MASVS-AUTH-3: “The app secures sensitive operations with additional authentication.”

Each note in the series covers one weakness from the MASWE catalogue, with its Knowledge and Tests. This page maps the area, draws the line between what a mobile test can prove and what needs backend testing, and collects the commands you will reach for on every authentication review.

Contents

What MASVS-AUTH covers

MASVS says it plainly: authentication and authorisation must be enforced on the remote endpoint. The app’s job is to:

  • use the protocols correctly;
  • authenticate the user locally in a way that cannot be switched off with a hook;
  • ask for fresh authentication before anything sensitive;
  • not leave the door open afterwards.

In practice the three controls break down like this:

  • AUTH-1 (protocols and authorisation): how the app logs in and keeps a session, and whether its own components (activities, services, receivers, providers) let other apps act without authorisation. On the MASWE list: MASWE-0018 (unprotected app components) and MASWE-0019 (no autofill support, which drives users towards weak passwords).
  • AUTH-2 (local authentication): biometrics and app PINs. Is the check bound to cryptography or is it a boolean in the UI? Can it fall back to a weaker credential? Does a newly enrolled fingerprint inherit access? That is MASWE-0020, 0021 and 0022.
  • AUTH-3 (sensitive operations): step-up before payments and account changes (MASWE-0023), sessions that really end (MASWE-0024), and evidence of who approved what (MASWE-0025).

Two neighbouring notes from the CRYPTO series are prerequisites for AUTH-2:

  • MASWE-0016: keys that require user authentication to use.
  • MASWE-0017: a device passcode, without which none of that works.

Our older notes on Android biometric authentication and FingerprintManager also belong to this group.

The eight weaknesses

ID Weakness Controls Profiles Linked tests Note
MASWE-0018 Lack of Authentication or Authorization on App Components AUTH-1, PLATFORM-1, STORAGE-2 L1, L2, EUDIW 6 Read
MASWE-0019 Lack of Auto-fill Support for Credential Providers AUTH-1, AUTH-3 L1, L2, EUDIW 0 Read
MASWE-0020 Local Authentication Can Be Bypassed AUTH-2, CRYPTO-2 L2, EUDIW 5 Read
MASWE-0021 Fallback to Non-biometric Credentials Allowed for Sensitive Transactions AUTH-2 L2, EUDIW 3 Read
MASWE-0022 Crypto Keys Not Invalidated on New Biometric Enrollment AUTH-2, CRYPTO-2 L2, EUDIW 3 Read
MASWE-0023 Step-Up Authentication Not Implemented for Sensitive Actions AUTH-3, PLATFORM-3 L2, EUDIW 0 Read
MASWE-0024 Sensitive Data Accessible After Session Termination AUTH-3 L2, EUDIW 0 Read
MASWE-0025 Lack of Non-Repudiation for Critical Actions AUTH-3 L2, EUDIW 0 Read

Four of the eight weaknesses have no OWASP tests yet. Their notes give our own procedures, marked Guler Tech methodology.

The profiles column tells you when a weakness applies. L1 applies to every app: unprotected components and missing autofill support. Everything about biometrics, step-up, session termination and non-repudiation is L2/EUDIW, the profiles for banking, health, identity and the EU Digital Identity Wallet. For an L1 app these are still worth raising as observations, but they should not be reported as failures of the profile.

Knowledge and Tests map

Every OWASP item linked from the eight weaknesses on the live MASWE site. Weaknesses not listed here have no linked items yet.

MASWE-0018 — Lack of Authentication or Authorization on App Components

Type ID Title Platform
Knowledge MASTG-KNOW-0020 Inter-Process Communication (IPC) Mechanisms Android
Knowledge MASTG-KNOW-0132 Android Activities Android
Knowledge MASTG-KNOW-0133 Android Services Android
Knowledge MASTG-KNOW-0134 Android Broadcast Receivers Android
Knowledge MASTG-KNOW-0117 Android ContentProvider Android
Knowledge MASTG-KNOW-0017 App Permissions Android
Test MASTG-TEST-0364 Exported And Unprotected Activities That Expose Sensitive Functionality Android
Test MASTG-TEST-0365 Exported And Unprotected Services That Expose Sensitive Functionality Android
Test MASTG-TEST-0366 Exported And Unprotected Broadcast Receivers That Expose Sensitive Functionality Android
Test MASTG-TEST-0355 References to Unauthorized Database Access through Content Providers Android
Test MASTG-TEST-0356 Runtime Verification of Unauthorized Database Access through Content Providers Android
Test MASTG-TEST-0357 References to Oversharing of File-Based Content Providers Android
Best practice MASTG-BEST-0052 Restrict Access to Android App Components Android
Best practice MASTG-BEST-0049 Restrict and Validate Access to Exported Content Providers Android

MASWE-0020 — Local Authentication Can Be Bypassed

Type ID Title Platform
Knowledge MASTG-KNOW-0001 Biometric Authentication Android
Knowledge MASTG-KNOW-0043 Android KeyStore Android
Knowledge MASTG-KNOW-0047 Cryptographic Key Storage Android
Knowledge MASTG-KNOW-0012 Key Generation Android
Knowledge MASTG-KNOW-0056 Local Authentication Framework iOS
Knowledge MASTG-KNOW-0057 Keychain Services iOS
Test MASTG-TEST-0327 References to APIs for Event-Bound Biometric Authentication Android
Test MASTG-TEST-0329 References to APIs Enforcing Authentication without Explicit User Action Android
Test MASTG-TEST-0330 References to APIs for Keys used in Biometric Authentication with Extended Validity Android
Test MASTG-TEST-0266 References to APIs for Event-Bound Biometric Authentication iOS
Test MASTG-TEST-0267 Runtime Use Of Event-Bound Biometric Authentication iOS
Best practice MASTG-BEST-0036 Use Cryptographic Binding for Biometric Authentication —
Best practice MASTG-BEST-0038 Require Explicit User Confirmation for Biometric Authentication Android

MASWE-0021 — Fallback to Non-biometric Credentials Allowed for Sensitive Transactions

Type ID Title Platform
Test MASTG-TEST-0326 References to APIs Allowing Fallback to Non-Biometric Authentication Android
Test MASTG-TEST-0268 References to APIs Allowing Fallback to Non-Biometric Authentication iOS
Test MASTG-TEST-0269 Runtime Use Of APIs Allowing Fallback to Non-Biometric Authentication iOS
Best practice MASTG-BEST-0031 Enforce Strong Biometrics for Sensitive Operations —

MASWE-0022 — Crypto Keys Not Invalidated on New Biometric Enrollment

Type ID Title Platform
Test MASTG-TEST-0328 References to APIs Detecting Biometric Enrollment Changes Android
Test MASTG-TEST-0270 References to APIs Detecting Biometric Enrollment Changes iOS
Test MASTG-TEST-0271 Runtime Use Of APIs Detecting Biometric Enrollment Changes iOS
Best practice MASTG-BEST-0037 Invalidate Biometric Keys on Enrollment Changes —

Client or server: what a mobile test can prove

Authentication findings are where mobile reports are most often over- or under-rated. The test is always the same question: does the server’s decision depend on something the client controls?

You observe on the device What it proves What you still need
Biometric prompt is a UI gate, and hooking onAuthenticationSucceeded or evaluatePolicy unlocks the screen The local control is bypassable (MASWE-0020) Whether the unlocked screen can do anything the session token could not already do. If the API accepts the same token with or without the prompt, the biometric adds no server-side assurance.
Sensitive API call succeeds after you remove a step-up header or replay it without a fresh OTP Server does not enforce step-up (MASWE-0023) Nothing more: this is a server-side finding proven from the client.
Old access or refresh token still works after logout Server does not revoke on logout (MASWE-0024) Token lifetimes, to rate it.
Exported component performs an action when called by adb or a test app Missing component authorisation (MASWE-0018) Whether the action reaches the backend with the user’s session (usually yes, which raises severity).
Request carries no signature, or a signature from a key that is not auth-bound No non-repudiation (MASWE-0025) Client’s audit/logging design. Ask for it; you cannot see it.

Where the finding is really a backend issue (broken object-level authorisation, weak password policy, missing lockout), report it against the API using OWASP ASVS / WSTG references, and mention the mobile flow only as the route you used to reach it.

Authentication building blocks on mobile

A quick reference for what “follows the relevant best practices” means in 2026.

  • Login via OAuth 2.0 / OIDC in the system browser, not an embedded WebView. Use Custom Tabs on Android and ASWebAuthenticationSession on iOS, with PKCE (RFC 7636) and a claimed HTTPS redirect (App Links / Universal Links), or at least a private-use scheme (RFC 8252). Do not ship a client secret in the app (see MASWE-0004). The current security baseline is RFC 9700, the OAuth 2.0 Security Best Current Practice.
  • Passkeys (WebAuthn/FIDO2) via Android Credential Manager and iOS ASAuthorizationController are now the strongest mainstream option for both sign-in and step-up: phishing-resistant, device-bound or synced, with user verification built in.
  • Tokens: short-lived access tokens and rotating refresh tokens with reuse detection, held in Keystore-encrypted storage or the Keychain (MASWE-0001) and revoked server-side on logout (RFC 7009).
  • Biometrics: only meaningful when crypto-bound. The biometric unlocks a Keystore/Keychain key, which signs a server challenge or decrypts a secret. A biometric that only returns true to the app is a convenience feature, not a security control.
  • Step-up and transaction confirmation: the server asks for fresh authentication (or a signature from an auth-bound key) for each sensitive action, and binds it to the specific transaction. In the EU this is also the PSD2 “dynamic linking” requirement for payments.
  • Device binding: an attested, hardware-backed key generated at enrolment identifies the app instance. Many of the controls above build on it.

Test plan

1. Map the authentication surface

List:

  • every way to log in: password, social/OIDC, magic link, passkey, SMS OTP;
  • every way the app stays logged in: tokens, cookies, refresh;
  • every place a local check appears: app PIN, biometric prompt, “unlock with Face ID”;
  • every action that should require more than a session: payments, payee changes, contact detail changes, secret reveal, device binding, data export, account deletion.

Note which side enforces each.

2. Static review

  • Android manifest: exported components, permissions and protection levels, grantUriPermissions, provider paths (MASWE-0018).
  • BiometricPrompt usage: CryptoObject present? Allowed authenticators? setConfirmationRequired? (MASWE-0020/0021).
  • KeyGenParameterSpec: user-authentication parameters, invalidation on enrolment (MASWE-0021/0022).
  • iOS: LAContext.evaluatePolicy vs Keychain items with SecAccessControl; policy and flags such as .biometryCurrentSet vs .biometryAny (MASWE-0020/0021/0022).
  • Login flow: WebView vs system browser, PKCE, client secrets.
  • Password fields: autofill hints and content types (MASWE-0019).

3. Dynamic review

  • Intercept the whole flow in Burp. For each sensitive action, replay it without the step-up artefact, with an old one, and for a different transaction (MASWE-0023).
  • Hook the local authentication callbacks and confirm whether bypassing them gives you access to anything new (MASWE-0020).
  • Call exported components from adb or a test app (MASWE-0018).
  • On a test device: change the biometric enrolment and check whether protected actions still work (MASWE-0022). Try the passcode fallback on a sensitive action (MASWE-0021).

4. Session termination

Log out, expire and revoke. Then:

  • replay the old tokens;
  • diff the sandbox for leftover data;
  • navigate back and open deep links into authenticated screens (MASWE-0024).

5. Evidence for critical actions

Check whether critical requests carry a signature from a hardware-backed, user-authenticated key, and whether the server rejects altered, unsigned or replayed requests (MASWE-0025).

Command and hook cheat sheet

Android components

adb shell dumpsys package <package> | sed -n '/Activity Resolver Table/,/^$/p'
adb shell am start -n <package>/.SecretActivity --es token test
adb shell am startservice -n <package>/.SyncService   # use start-foreground-service on API 26+ where needed
adb shell am broadcast -a <package>.ACTION_RESET -n <package>/.ResetReceiver
adb shell content query --uri content://<authority>/users
adb shell content read --uri content://<authority>/files/../shared_prefs/auth.xml

Local authentication

# Android: watch BiometricPrompt callbacks and whether a CryptoObject is passed
frida-trace -U -f <package> \
  -j 'androidx.biometric.BiometricPrompt!authenticate' \
  -j 'androidx.biometric.BiometricPrompt$AuthenticationCallback!onAuthenticationSucceeded'

# iOS: LocalAuthentication and Keychain access-control usage
frida-trace -U -f <bundle-id> -m '-[LAContext evaluatePolicy:localizedReason:reply:]' \
  -i 'SecAccessControlCreateWithFlags' -i 'SecItemCopyMatching'

# iOS: objection's built-in event-bound bypass (proves the check is not crypto-bound)
objection -g <bundle-id> explore
#   ios ui biometrics_bypass

Tokens and sessions

# Decode a JWT payload without verifying it, to read exp / acr / amr / scope
cut -d. -f2 <<<"$JWT" | tr '_-' '/+' | base64 -d 2>/dev/null; echo

# Replay an old access token after logout (expect 401)
curl -s -o /dev/null -w '%{http_code}\n' -H "Authorization: Bearer $OLD_TOKEN" https://api.example.com/v1/me

The individual notes contain complete scripts: Frida hooks that print the CryptoObject and key parameters, attacker apps for exported components, and replay matrices for step-up and logout.

Reporting authentication findings

  • Lead with the server’s behaviour. “The transfer API accepts requests without the step-up token” is a High-severity finding with a one-line proof. “The biometric prompt can be bypassed with Frida” needs the second half of the sentence: “…and the bypassed screen can then approve transfers the plain session cannot.”
  • Separate local from remote impact. A hookable app PIN on an app whose tokens are already in plaintext storage adds little. On an app that keeps tokens Keystore-encrypted behind an auth-bound key, the same hook is the whole attack.
  • Quote the platform rule. For example: BiometricPrompt.authenticate(info) without a CryptoObject (MASTG-TEST-0327), .deviceOwnerAuthentication where only biometrics should be allowed (MASTG-TEST-0268), or .biometryAny where .biometryCurrentSet is needed (MASTG-TEST-0270).
  • Give the backend fix as well as the client fix. For local authentication that usually means: generate an auth-bound key at enrolment, register its public key, and have the server verify a signature over a server challenge for each sensitive action.

References

Adapted and extended from the OWASP MASVS, MASWE and MASTG (https://mas.owasp.org), licensed under CC BY-SA 4.0. This note is distributed under the same licence.

Back to Authentication & Authorisation