agentleFS
Cap-goCap-go/capgo-skills/skills/capacitor-performance/SKILL.md

capacitor-performance

Skill75 starsChanged 32 days agoMITView on GitHub

Diagnoses and fixes performance problems specific to Capacitor apps - slow cold start or long splash, white screen after camera/backgrounding (WKWebView "WebView process terminated" reload, Android onRenderProcessGone), large base64 payloads over the native bridge (Camera, Filesystem.readFile, CapacitorHttp), jank in the WebView, memory growth from unremoved plugin listeners, oversized Preferences storage, slow Android WebView versions, and debug logging left on in release. Covers profiling with Safari Web Inspector, chrome://inspect, Xcode Instruments, Android Studio Profiler, and adb timing commands. Do not use for crash/stack-trace debugging (use debugging-capacitor or capacitor-native-observability), device log capture (use ios-android-logs), offline data architecture (use capacitor-offline-first / sqlite-to-fast-sql), splash configuration details (use capacitor-splash-screen), or generic web performance unrelated to Capacitor.

1 other project uses a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

2 rules. About 0 are common, 2 are distinctive, and 0 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈2

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 0

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Base64 over the bridge: Every plugin call/response is JSON-serialized across the bridge.

    line 49
  2. Distinctive:

    Only load when needed:

    line 81

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Capacitor Performance

Find the bottleneck first (WebView JS, bridge, native, or network), then fix the specific cause. Generic web advice (code splitting, image compression, virtual lists) still applies; this skill focuses on what is different inside Capacitor.

When to Use

TRIGGER when:

  • Cold start is slow, splash stays long, or the first screen is blank for seconds.
  • App "restarts" or shows a white screen after taking a photo, opening a big file, or returning from background.
  • Native calls feel slow, especially with images, files, or large HTTP responses.
  • Scrolling/animations jank only on device (often older Android WebView).
  • Memory grows over time; listeners pile up.

Do not use:

  • Crashes, native stack traces -> debugging-capacitor, capacitor-native-observability.
  • Getting logs off a device -> ios-android-logs.
  • Offline storage design / SQLite migration -> capacitor-offline-first, sqlite-to-fast-sql.
  • Splash screen setup -> capacitor-splash-screen; keyboard resize jank -> capacitor-keyboard.
  • OTA bundle size and delta updates -> capgo-live-updates.

Procedure

  1. Reproduce on a real device in a release-like build. Debug builds log every bridge call and disable optimizations; simulators hide memory and GPU limits.
  2. Classify with the profiling table below: JS main thread busy, bridge payloads large, native thread busy, or waiting on network.
  3. Check the traps list. Most Capacitor-specific regressions are one of these.
  4. Fix one thing, measure again with the same command/trace. Report before/after numbers to the user.

Profiling Quick Reference

QuestionTool
JS long tasks, layout, paint (iOS)Safari > Develop > device > app > Timelines. Requires ios.webContentsDebuggingEnabled: true for release builds (debug builds are inspectable by default on iOS 16.4+).
JS long tasks, memory heap (Android)chrome://inspect > Performance / Memory.
Native CPU, allocations, leaks (iOS)Xcode > Product > Profile > Time Profiler / Allocations / Leaks; field data in Xcode Organizer > Metrics.
Native CPU/memory (Android)Android Studio > Profiler.
Android cold start timeadb shell am force-stop <appId> && adb shell am start -W -n <appId>/.MainActivity (read TotalTime).
Android frame statsadb shell dumpsys gfxinfo <appId> (janky frames %).
Android WebView versionadb shell dumpsys webviewupdate (current provider and version).

Details and recipes: references/profiling.md.

Capacitor-Specific Traps

  1. Base64 over the bridge. Every plugin call/response is JSON-serialized across the bridge. Multi-MB base64 strings stall both threads and can kill the WebView process.
    • Camera (@capacitor/camera 8.1+ API): use takePhoto / chooseFromGallery and display result.webPath; set targetWidth/targetHeight/quality. Avoid reading full-resolution base64 unless uploading. The older getPhoto / pickImages are deprecated.
    • Files: pass paths, use Capacitor.convertFileSrc(uri) for <img>/<video> src, Filesystem.readFileInChunks for large reads, @capacitor/file-transfer for uploads/downloads.
  2. WebView process killed under memory pressure. iOS logs WebView process terminated; Capacitor resets the bridge and reloads the page, so in-memory JS state is lost (looks like a restart). Android calls onRenderProcessGone; unless handled, the app is killed. Reduce peak memory (trap 1, large canvases, decoded images), persist critical state, restore on load.
  3. CapacitorHttp for large responses. With plugins.CapacitorHttp.enabled: true, fetch/XMLHttpRequest are routed through native and bodies cross the bridge. Great for CORS/cookies, slower for big downloads or streaming. Use file-transfer for big files.
  4. Preferences as a database. @capacitor/preferences (UserDefaults / SharedPreferences) is for small key/value data. Large JSON blobs are read/written whole and slow startup; use SQLite (for example @capgo/capacitor-fast-sql).
  5. Logging in release. loggingBehavior defaults to debug (logs only in debug builds). If it was set to production, every bridge call is logged in release; set it back to debug or none.
  6. Dev server left in config. server.url pointing to a LAN/live-reload server in a shipped build loads over network and fails offline. Remove before release; in Capacitor 9 use npx cap run <platform> --url for live reload instead of editing config.
  7. Old Android System WebView. Performance and CSS support depend on the updatable WebView, not the OS version. android.minWebViewVersion (default 60) only shows an error/server.errorPath; it does not update anything. Expensive CSS (backdrop-filter, large box-shadow/filter on scrolling content) janks on low-end WebViews.
  8. Listener leaks. addListener returns Promise<PluginListenerHandle>; await it and call remove() on unmount. Re-registering on every route change multiplies native callbacks (appStateChange, keyboardWillShow, geolocation watchPosition - clear with clearWatch).
  9. Startup sequencing. Hide the splash only after first meaningful render (launchAutoHide: false + SplashScreen.hide()), and defer non-critical plugin calls (analytics init, permission checks) until after first paint. Note Capacitor 9 changes Android launchFadeOutDuration default from 200 to 0.
  10. Plugin imports are not the startup cost. Importing @capacitor/* packages creates lightweight proxies; their web implementations are already lazy-loaded via registerPlugin. Lazy-load heavy JS libraries (charts, PDF, maps SDK JS) and routes instead.

Verification

  • Same device, same build type, before/after numbers: cold start TotalTime (Android) or Instruments App Launch / Organizer launch time (iOS), janky frame %, peak memory in Allocations/Profiler.
  • Grep the release config: grep -nE '"url"|loggingBehavior|webContentsDebuggingEnabled' capacitor.config.* and confirm no dev server.url.
  • For trap 1, grep for full-size base64 usage: grep -rnE "resultType: *CameraResultType.Base64|readFile\(|getPhoto\(" src/.
  • Background the app, open the camera 10 times, return: the page must not reload (iOS) or crash (Android).

Error Handling

Message / symptomMeaningFix
WebView process terminated (Xcode console)WKWebView content process killed (memory)Trap 1/2: lower peak memory, persist state.
Android logcat Render process ... gone / app exits after heavy pageRenderer crashed or killedReduce memory; implement handling via a WebViewListener onRenderProcessGone if recovery is required.
ANR in <appId> (logcat)Main thread blocked > 5 s, often heavy plugin work on UI threadMove native work off main thread in custom plugins; profile with Android Studio.
Message from debugger: Terminated due to memory issueiOS jetsam killed the whole appAllocations instrument; reduce image decoding and caches.
Logcat error about the WebView version being below minWebViewVersionDevice WebView too oldAsk user to update Android System WebView; show server.errorPath page.

References

Only load when needed:

  • references/profiling.md: step-by-step traces (Safari Timelines, Chrome Performance, Instruments templates, Android Profiler, adb commands) and how to read them.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.