finds.dev← search

// the find

bamlab/flashlight

★ 1,584 · TypeScript · MIT · updated Jun 2026

📱⚡️ Lighthouse for Mobile - audits your app and gives a performance score to your Android apps (native, React Native, Flutter..). Measure performance on CLI, E2E tests, CI...

Flashlight is a performance auditing tool for Android apps that collects CPU, FPS, and RAM metrics via ADB without requiring any changes to the app being tested. Works on production APKs, supports React Native, Flutter, and native Android. Think Lighthouse but for Android runtime performance, with a numeric score and HTML report output.

No instrumentation required in the target app — it reads `dumpsys gfxinfo` and a custom C++ profiler pushed via ADB, so you can profile apps you don't own. The scoring model aggregates multiple signals (frame drops, CPU spikes, RAM) into a single comparable number, which makes CI regression detection straightforward. The `flashlight test` command runs multiple iterations and averages them, which matters because single-run mobile perf numbers are noisy. The cloud mode lets you run against real devices without maintaining a device farm yourself.

Android-only for now — iOS support has been 'in the works' for a long time based on the open issue, so don't plan around it. The C++ profiler ships as prebuilt binaries checked into the repo (`BAMPerfProfiler-arm64-v8a` etc.), which means you're trusting binaries you can't easily audit, and rebuilding them requires a proper NDK setup that isn't documented in the main README. The performance score weighting is opaque — the formula is in `getScore.ts` but there's no explanation of why the weights are what they are, making it hard to know if the number actually maps to user-perceived quality. No Windows device support for ADB-based testing: the install script works, but ADB device workflows assume macOS/Linux.

View on GitHub → Homepage ↗

// want more like this?

We dig through GitHub every week and send a few repos picked for what you actually care about — each with an honest take like this one.

Get finds in your inbox → Search again →