cap-go/capgo-skills/capacitor-offline-first

Offline-First Capacitor Apps

When to Use

TRIGGER when:

• The app must read and write while offline and sync later
• Choosing local storage (Preferences vs SQLite vs IndexedDB) for user data
• Writes are duplicated, lost, or overwritten during sync
• Cached data vanishes on iOS after low-storage events
• A service worker was added and native plugins stopped working (Plugin not implemented)

Do not use:

• Replacing @capacitor-community/sqlite / Ionic secure storage: sqlite-to-fast-sql
• Getting new web code to devices (OTA): capgo-live-updates
• Encryption key storage, SSL pinning: capacitor-security
• Query speed / bridge overhead tuning: capacitor-performance

Decisions (make these before writing code)

Workflow

1. Inspect: current storage (localStorage, @capacitor/preferences, other SQLite plugins), API shape (does it support updated_since / version fields / idempotency keys?), auth refresh flow, existing service worker registration.
2. Report the plan (schema, outbox, conflict policy) to the user before invasive changes. For large apps, list each entity as a TODO.
3. Install: npm install @capgo/capacitor-fast-sql @capacitor/network && npx cap sync, then the platform setup in references/fast-sql-setup.md (required: iOS ATS local networking, Android localhost cleartext).
4. Schema + outbox: references/sync-engine.md.
5. Triggers for sync: app start, networkStatusChange -> connected, App resume, after each local write (debounced), optional background task.
6. UI: show pending count / last synced time; never block a write on the network.
7. Verify (below).

Only load a reference when its topic is in play.

Minimal shape

import { FastSQL } from '@capgo/capacitor-fast-sql';
import { Network } from '@capacitor/network';
import { App } from '@capacitor/app';
const db = await FastSQL.connect({ database: 'app' });
await db.execute(`CREATE TABLE IF NOT EXISTS outbox (
  id TEXT PRIMARY KEY, entity TEXT NOT NULL, op TEXT NOT NULL,
  payload TEXT NOT NULL, created_at INTEGER NOT NULL, attempts INTEGER NOT NULL DEFAULT 0
)`);
export async function saveTodo(todo: Todo) {
  await db.transaction(async (tx) => {
    await tx.run('INSERT OR REPLACE INTO todos (id, text, done, updated_at, dirty) VALUES (?, ?, ?, ?, 1)',
      [todo.id, todo.text, todo.done ? 1 : 0, Date.now()]);
    await tx.run('INSERT INTO outbox (id, entity, op, payload, created_at) VALUES (?, ?, ?, ?, ?)',
      [crypto.randomUUID(), 'todo', 'upsert', JSON.stringify(todo), Date.now()]);
  });
  void sync(); // fire and forget; sync() is single-flight
}
Network.addListener('networkStatusChange', (s) => { if (s.connected) void sync(); });
App.addListener('resume', () => void sync());

The data row and its outbox entry are written in one transaction so a crash cannot leave one without the other.

Traps

• Network.getStatus().connected === true does not mean the API is reachable (captive portals, VPN, server down). Treat fetch failures as offline; use connected only as a hint to retry.
• networkStatusChange can fire several times in a row; make sync() single-flight.
• Send an idempotency key (the outbox row id) with every mutation; retries after a timeout otherwise duplicate records.
• Process the outbox in order per entity; a delete must not overtake its create.
• Advance the pull cursor using the server's timestamp/version from the response, never the device clock.
• Auth tokens expire while offline: refresh before draining the outbox, and keep entries on 401 instead of dropping them.
• Do not store large blobs in SQLite rows; save files with @capacitor/filesystem and keep the path in the DB.
• iOS background execution is opportunistic. Background tasks help but cannot guarantee sync timing; foreground sync on resume is the reliable path.

Verification

1. Airplane mode on device: create, edit, delete records; kill and relaunch the app; data is still there.
2. Turn network back on: outbox drains to 0, server reflects all changes once (no duplicates).
3. Conflict test: edit the same record on two devices offline, reconnect both; result matches the chosen policy.
4. Flaky network: Android emulator network throttling / iOS Network Link Conditioner (Settings -> Developer); retries back off, no data loss.
5. Native checks:

grep -rn "serviceWorker.register" src/ | grep -v "isNativePlatform"   # should be empty or web-guarded
/usr/libexec/PlistBuddy -c "Print :NSAppTransportSecurity:NSAllowsLocalNetworking" ios/App/App/Info.plist
grep -n "networkSecurityConfig\|usesCleartextTraffic" android/app/src/main/AndroidManifest.xml

1. Inspect the DB: SELECT COUNT(*) FROM outbox exposed in a debug screen, or pull the file (adb exec-out run-as com.example.app ls databases/).

Error Handling

Resources

• Fast SQL plugin: https://capgo.app/docs/plugins/fast-sql/
• Capacitor storage guide: https://capacitorjs.com/docs/guides/storage
• Network API: https://capacitorjs.com/docs/apis/network
• Background task plugin: https://capgo.app/docs/plugins/background-task/
• iOS troubleshooting (WKAppBoundDomains): https://capacitorjs.com/docs/ios/troubleshooting