Mobile Application Security Audit (OWASP Mobile Top 10)
After a mobile app update on Google Play or App Store, competitors can gain access to your user base. Or an attacker exploits a vulnerable deep link to reset an admin password. These scenarios stem from a lack of systematic security audits. Our engineers, with 5+ years of hands-on experience and over 100 completed audits, find these weaknesses before attackers do.
According to OWASP, more than 80% of mobile applications contain at least one vulnerability from the Mobile Top 10. Fixing a vulnerability during development costs 30 times less than after release. Therefore, an audit before publication is a smart investment. We combine static and dynamic analysis with manual penetration testing to cover all attack vectors. Here is how it works.
Why OWASP Mobile Top 10 is the benchmark for testing
The OWASP Mobile Top 10 is not a formal checklist but a structured approach to active testing. Each of the 10 items is adapted to the application’s architecture: we do not just check off a list—we reproduce attacks in real conditions. Here is how we do it.
How we find vulnerabilities: details of the process
Let us start with the most common source of risk: improper secret storage.
M1 and M2: Secrets, dependencies, and supply chain
We look for hardcoded credentials: API keys in code, passwords in configuration files, tokens in git history. Tools: jadx + grep, truffleHog for repositories, analysis of AndroidManifest.xml and Info.plist. We verify storage: credentials in SharedPreferences/UserDefaults are a vulnerability. They should be in Android Keystore / iOS Keychain. On a jailbroken device we read Keychain using objection keychain dump—we check what is stored and with what access attributes.
Third-party dependencies are often the weakest link. We check library versions against known CVEs (OWASP Dependency-Check, gradle dependencyInsight, pod-outdated), usage of untrusted libraries, permissions requested by analytics and advertising SDKs. Separately, we examine the CI/CD pipeline: secret scanning in the repository, artifact signing, dependency integrity via hash verification.
M3 and M4: Authentication, authorization, and input validation
We test bypassing the login screen via deep links (passing parameters in URL that should only be accessible to authenticated users), horizontal privilege escalation (authenticated user A accesses user B’s data by modifying user_id in the request), and lack of session revalidation for critical operations.
In practice, we often find: a deep link myapp://reset-password?token=XXX is handled without checking the source of the intent—any app can send such an intent and trigger a password reset. Or changing an email in the profile does not require the current password.
On the mobile client especially relevant: SQL injections via deep link parameters or WebView URLs, XSS in WebView with setJavaScriptEnabled(true), path traversal when working with files (URL like ../../etc/passwd in file upload parameters), and unsafe deserialization in Intent extras.
// vulnerable code – takes Intent extras without validation String fileName = getIntent().getStringExtra("file_name"); File file = new File(getExternalFilesDir(null), fileName); // fileName = "../../../../../../data/data/com.other.app/secret.db" M5 and M8: Communications and configuration
We check using Burp Suite proxy: HTTPS enabled for all endpoints, certificate pinning (bypassed with Frida ssl-unpinning.js), sensitive data in GET parameters (logged by servers, proxies, CDNs), insecure WebSocket connections, leakage of sensitive data in request headers. network_security_config.xml on Android—check cleartextTrafficPermitted, presence of user-supplied CAs in trust-anchors.
Safety Misconfiguration: android:debuggable="true" in the production manifest opens debug access. android:allowBackup="true" enables adb backup on Android < 12—from the backup we read SharedPreferences, databases. exported="true" on components without intent validation. On iOS—ATS (App Transport Security) disabled via NSAllowsArbitraryLoads. Entitlements: excessive capabilities (e.g., com.apple.developer.icloud-container-identifiers in an app that does not use iCloud).
M6, M7 and M9: Data, binary protection, and storage
Permissions: the app requests ACCESS_FINE_LOCATION constantly, but geolocation is only needed in a specific scenario? Or READ_CONTACTS without any visible contact-related functionality? We analyze the alignment of requested permissions with declared functionality. Logs: adb logcat often reveals PII in production builds. We check for sensitive data in logcat, Crashlytics/Sentry messages, and analytics events.
Decompile APK with jadx, IPA with Ghidra. Assess: readability of business logic after decompilation, presence and quality of obfuscation (R8/ProGuard/DexGuard), plaintext string constants, debug flags in production builds (BuildConfig.DEBUG, debuggable in manifest), and anti-tampering checks.
Full audit of device storage:
| Storage | What we look for | Tool |
|---|---|---|
| SQLite DB | Sensitive data, lack of encryption | objection, sqlite3 |
| SharedPreferences / UserDefaults | Passwords, tokens, keys | objection data storage |
| Keychain (iOS) | Access attributes, what is stored | objection keychain dump |
| File system | Unencrypted documents, API response cache | objection files ls |
| Clipboard | Auto-copying sensitive data | Manual testing |
M10: Weak cryptography
Weak algorithms: DES, 3DES, RC4, MD5 for passwords, ECB mode for block ciphers, predictable seed in java.util.Random instead of SecureRandom, zero or fixed IV, missing MAC (using AES-CBC without HMAC). Custom cryptography implementations instead of standard libraries are a red flag. “Homegrown crypto” is almost always broken.
What the audit delivers: results and prioritization
For each of the 10 categories, we record: found/not found, specific vulnerability instances with CVSS scores, reproduction steps, and remediation recommendations with code examples. Priorities: Critical (exploitable without root/jailbreak, direct access to data) → High → Medium → Low (informational findings).
How to prepare for the audit: 3 steps
- Collect the current binary files (APK/IPA) and the source code if available.
- Provide documentation: architecture description, list of third-party libraries, API endpoints.
- Identify critical scenarios: authentication, payments, handling of personal data.
Comparison of testing methods
| Type of analysis | Time required | Coverage depth | Vulnerability detection rate |
|---|---|---|---|
| Static | 1–2 days | Code + dependencies | 60–70% |
| Dynamic | 2–3 days | Runtime + network | 40–50% |
| Manual pentest | 3–5 days | Logic + business | 80–95% |
Manual testing detects up to 95% of vulnerabilities—30% more than automated screening. That is why our audit combines all three types.
In one project, we discovered that the app stored OAuth tokens in SharedPreferences with no encryption. After our recommendation, the development team moved them to the Android Keystore. The fix took one day and prevented unauthorized account access on rooted devices.
An audit of a typical medium-scale application against the OWASP Top 10 takes 3–5 working days. It includes static analysis, dynamic testing on a rooted Android and jailbroken iOS devices, and traffic analysis. The deliverable is a report according to the client’s specifications. Request a consultation to evaluate your app—our experts will contact you within an hour. Schedule a mobile app security audit today.