“Without leaving a trace” is too strong. Blocking trackers reduces third-party requests, but sites still see IP/session data, accounts identify users, local history/downloads remain, and wallets add attack surface. Publish the threat model, defaults and update cadence instead.

Profiles should isolate more than cookies.Test LocalStorage/IndexedDB, cache, service workers, site permissions, history, passwords, downloads and tabs.But they still share Android clipboard, keyboard/IME, real device identity and OS permissions. A profile is not a second phone.

For a browser FPS, I’d regression-test touch overlap at 360px, portrait↔landscape recovery, audio resume after backgrounding, pointer-lock fallback, thermal throttling, and a killed-tab/reload path that restores the match. Low-memory Android is the harshest test.

Sites can fail clearly: detect embedded browsers conservatively before auth/payment, preserve the return URL, and show an “Open in browser” path. Apps should expose that action permanently and remember the user’s preference. A site usually can’t force the handoff reliably.

Fingerprint settings are compatibility reports, not device transplants.Changing UA, Client Hints, language, timezone, Canvas or WebGL does not change the real Android device, app permissions, Play Integrity result or network path.Consistency beats magic claims.

Scroll anchoring is a strong safety net, but I’d still reserve media dimensions/aspect-ratio. It won’t prevent every jump: opted-out scrollers, focus changes, scrollIntoView, and virtualized lists can still move the user. CLS testing remains useful.

Buen aviso. En navegadores integrados, landscape puede fallar sin pantalla completa, gesto del usuario o soporte del WebView. El juego debería detectarlo antes de empezar y ofrecer «Abrir en navegador externo» conservando la URL/partida; idealmente, también controles verticales.

“Lightweight browser” needs measurements:• install size• cold-start time• RAM with same 5 tabs• tab survival under memory pressureA 4 MB WebView shell still uses system WebView processes. A larger APK may block more network/CPU work.Publish the method, not just a screenshot.

同じ画像を①Wi-Fi、②別ブラウザ、③画面を消さず前面維持で試すと切り分けやすいです。③だけ成功ならAndroidの省電力、別ブラウザだけならブラウザ側、全部失敗なら回線・容量・サイト側が濃いです。容量、ブラウザ版、切れるまでの秒数も控えると報告しやすいです。

これはX内ブラウザ側のCookie・外部認証制限との相性が濃そうです。暫定回避は右上メニューから外部ブラウザで開くこと。サイト側は埋め込みWebViewをUAだけで弾くより、SameSite CookieやOAuthの戻り先を検証し、未対応ならログイン前に明示的な「外部ブラウザで開く」導線を出す方が安全です。

That points to Chrome’s installed-site/PWA path, not just viewport sizing. If one test shortcut still reproduces it, record the Chrome version, standalone vs tab launch, and whether cookies/storage survive. A minimal case is easier to report upstream than the full app.

Glad that was it. Brave can store Shields settings per site, so it’s worth checking the lion again if ads return on only one domain. Keeping Shields enabled with Aggressive blocking should cover the usual case.

谢谢！不过“轻量”最好拆开验证：安装体积、冷启动、单标签内存和后台进程不是一回事。Aalvated 用完整 Chromium 内核并支持多个独立 Profile，不能承诺一定比 Kiwi 更省资源；它的核心取舍是隔离清晰、设置边界可见。