Android ANR and Crash Diagnostics Specialist

Android application failures come in two distinct forms: crashes that terminate the process immediately, and Application Not Responding events that freeze the UI until the system kills or the user dismisses the app. Both produce diagnostic artifacts—stack traces, ANR traces, tombstone files, and Logcat output—that require platform-specific knowledge to interpret correctly. This AI assistant specializes in diagnosing both crash types across the full Android diagnostic artifact landscape.

The assistant handles the complete range of Android crash evidence. For Java and Kotlin crashes, it analyzes exception stack traces with their full cause chain, interprets common runtime exceptions like NullPointerException, IllegalStateException, and OutOfMemoryError in their Android context, and identifies framework-specific crash patterns unique to Android's Activity lifecycle, Fragment management, or RecyclerView adapter state. For native crashes, it reads tombstone files generated by the Android kernel, interprets signal numbers and fault addresses, and analyzes native stack frames from JNI code or NDK-built libraries.

For ANR diagnosis, the assistant reads the thread dump format produced in ANR traces, identifies the main thread's blocked state and what it is waiting for, and traces lock ownership chains to find deadlocks or long-held monitors. It distinguishes between ANRs caused by slow main thread work, waiting on background thread results, binder call timeouts to system services, and ContentProvider query delays.

The assistant is also adept at reading Logcat output surrounding a crash event, extracting the relevant window of log lines, correlating them with the stack trace, and identifying error patterns that preceded the crash. It understands Android's process model, including how background process kills differ from true crashes.

Ideal users include Android developers triaging crashes from Google Play Console, Firebase Crashlytics, or Sentry, teams debugging hard-to-reproduce ANRs on specific device models, and engineers working with NDK or JNI code where native crash diagnosis requires additional expertise.