Skip to content

SKILL.md

Latest commit

 

History

History
215 lines (166 loc) · 7.78 KB

SKILL.md

File metadata and controls

215 lines (166 loc) · 7.78 KB
name mobile-app-monitoring
description Add production application performance monitoring (APM) to a React Native/Expo or Flutter app. Covers Sentry Performance, Datadog RUM, and Instabug for error tracking, performance tracing, session replay, and release health. Includes OpenTelemetry spans, cold/warm start metrics, Apdex scoring, alerting, dashboards, and user impact analysis. Use when the user needs to monitor production errors, track app performance, measure launch times, or set up alerting.
standards-version 1.6.3

Mobile App Monitoring

Trigger

Use this skill when the user:

  • Wants production error tracking and crash reporting beyond basic Crashlytics
  • Asks about APM, performance monitoring, or production observability
  • Needs to track app launch times, screen load durations, or API latency
  • Mentions "Sentry Performance", "Datadog", "Instabug", "APM", "Apdex", "session replay", or "release health"
  • Wants dashboards, alerts, or user impact analysis for production issues
  • Is preparing for a production launch and needs monitoring infrastructure

Required Inputs

  • Framework: Expo (React Native) or Flutter
  • APM provider: Sentry Performance (recommended), Datadog RUM, or Instabug
  • Metrics: what to track (errors, performance, sessions, user actions)
  • Alert thresholds: error rate %, response time p95, crash-free session rate

Workflow

  1. Choose an APM provider:

    Provider Strengths Pricing Best for
    Sentry Performance Error grouping, breadcrumbs, source maps, release health Free tier available Most mobile apps
    Datadog RUM Full-stack observability, session replay, log correlation Per session Enterprise, full-stack teams
    Instabug Bug reporting, surveys, in-app feedback, session replay Per app User-facing feedback + monitoring

  2. Set up Sentry (recommended for Expo):

    npx expo install @sentry/react-native

    Use mobile_setupMonitoring to generate the monitoring module, then wrap your app:

    // app/_layout.tsx
import * as Sentry from "@sentry/react-native";
import { monitoring } from "../lib/monitoring";

export default Sentry.wrap(function RootLayout() {
  return <Stack />;
});

    Add the Sentry Expo plugin to app.config:

    export default {
  plugins: [
    [
      "@sentry/react-native/expo",
      {
        organization: "your-org",
        project: "your-project",
      },
    ],
  ],
};

  3. Set up Sentry for Flutter:

    dependencies:
  sentry_flutter: ^7.0.0
    Future<void> main() async {
  await SentryFlutter.init(
    (options) {
      options.dsn = const String.fromEnvironment('SENTRY_DSN');
      options.tracesSampleRate = 0.2;
    },
    appRunner: () => runApp(const MyApp()),
  );
}

  4. Add performance tracing. Measure what matters:

    import * as Sentry from "@sentry/react-native";

async function loadDashboard() {
  const span = Sentry.startInactiveSpan({
    name: "dashboard.load",
    op: "ui.load",
  });

  try {
    const data = await fetchDashboardData();
    renderDashboard(data);
    span?.setStatus({ code: 1, message: "ok" });
  } catch (error) {
    span?.setStatus({ code: 2, message: "error" });
    Sentry.captureException(error);
  } finally {
    span?.end();
  }
}

  5. Track app launch performance:

    import * as Sentry from "@sentry/react-native";

const appStartSpan = Sentry.startInactiveSpan({
  name: "app.start",
  op: "app.start.cold",
});

export default function App() {
  useEffect(() => {
    appStartSpan?.end();
  }, []);

  return <RootNavigator />;
}

  6. Set up alerting. Configure alerts for critical thresholds:

    • Crash-free session rate < 99.5% - immediate alert
    • Error rate > 1% - warning alert
    • API p95 latency > 3s - warning alert
    • New error type in release - notification

    Configure in Sentry Dashboard > Alerts > Create Alert Rule.

  7. Release health tracking. Tag every error with the release version:

    Sentry.init({
  dsn: "...",
  release: `com.example.app@${appVersion}`,
  dist: buildNumber,
});

    This enables:

    • Per-release error rate comparison
    • Regression detection (new errors in latest release)
    • Adoption tracking (how many users are on each version)

  8. Session replay (Datadog/Instabug). Record user sessions for debugging:

    // Datadog session replay
import { DdRum } from "@datadog/mobile-react-native";

// Sessions are recorded automatically when RUM is configured
// Mask sensitive views:
DdRum.addAttribute("view.name", "PaymentScreen");

    Session replay captures user interactions, network requests, and errors in context. Useful for reproducing bugs reported by users.

Key References

Example Interaction

User: "I'm launching next week. I need to know immediately if something breaks in production."

Agent:

  1. Runs mobile_setupMonitoring with provider=sentry to generate the monitoring module
  2. Installs @sentry/react-native via mobile_installDependency
  3. Wraps the root layout with Sentry.wrap()
  4. Adds the Sentry Expo config plugin for source map uploads
  5. Creates custom spans for key user flows (login, checkout, feed load)
  6. Configures alert rules: crash-free rate <99.5%, new error in release, p95 >3s
  7. Tags the release with version and build number for regression tracking
  8. Reminds the user to set EXPO_PUBLIC_SENTRY_DSN in the EAS Secrets dashboard

MCP Usage

Step MCP Tool Description
Set up APM mobile_setupMonitoring Generate monitoring module with tracing and error capture
Install packages mobile_installDependency Install @sentry/react-native or sentry_flutter
Check build mobile_checkBuildHealth Verify Sentry plugin is wired correctly
Security audit mobile_securityAudit Ensure DSN is in env vars, not hardcoded

Common Pitfalls

  1. 100% trace sampling in production - Sampling all transactions is expensive and can hit Sentry/Datadog quotas fast. Use 10-20% sample rate in production, 100% in dev.
  2. No source maps - Without source maps, stack traces show minified/bundled code. Configure the Sentry build plugin to upload source maps on every EAS Build.
  3. Missing release tagging - Without release and dist, errors cannot be grouped by version. Always set these from your app version and build number.
  4. Alert fatigue - Too many alerts cause the team to ignore them. Start with 2-3 critical alerts (crash rate, new errors) and expand gradually.
  5. Not tracking custom spans - Default instrumentation covers crashes but not business logic. Add spans for key flows: login, checkout, search, image upload.
  6. PII in error reports - Sentry captures breadcrumbs and context that may include user data. Configure beforeSend to scrub PII (emails, phone numbers, addresses).

See Also