riverpod-offline
Persist Riverpod notifier state offline with Storage and persist(); riverpod_sqflite, JsonPersist, key, destroyKey, cache duration, testing with in-memory storage. Use when saving state across app restarts or offline. Use this skill when the user asks about offline persistence, persisting state, or Riverpod storage.
How do I install this agent skill?
npx skills add https://github.com/serverpod/skills-registry --skill riverpod-offlineIs this agent skill safe to install?
- Gen Agent Trust Hubpass
The skill provides instructions and code examples for implementing offline data persistence in Flutter applications using Riverpod. It is safe for use and follows standard mobile development practices.
- Socketpass
No alerts
- Snykpass
Risk: LOW · No issues
- ZeroLeakspass
Score: 93/100 · 2 sections analyzed
What does this agent skill do?
Riverpod — Offline persistence (experimental)
Instructions
Offline persistence stores provider state on device so it survives restarts and works offline. Riverpod is storage-agnostic; packages like riverpod_sqflite provide a Storage implementation. Only Notifier-based providers can be persisted. The feature is experimental.
Creating a Storage
Install a package (e.g. riverpod_sqflite + sqflite) and create a Storage. With SQFlite:
final storageProvider = FutureProvider<Storage<String, String>>((ref) async {
return JsonSqFliteStorage.open(
join(await getDatabasesPath(), 'riverpod.db'),
);
});
Persisting a notifier
Inside the notifier's build, call persist with: the Storage (e.g. ref.watch(storageProvider.future)), a unique key, and encode/ decode for your state. Do not await persist; Riverpod handles it.
class TodoList extends AsyncNotifier<List<Todo>> {
@override
Future<List<Todo>> build() async {
persist(
ref.watch(storageProvider.future),
key: 'todo_list',
encode: (todos) => todos.map((todo) => {'task': todo.task}).toList(),
decode: (json) => (json as List).map((todo) => Todo(task: todo['task'] as String)).toList(),
);
return fetchTodosFromServer();
}
}
Keys
- Unique across all persisted providers (same key = same row, risk of corruption).
- Stable across restarts (changing the key loses restored state).
- For family providers, include the parameter in the key.
JsonPersist (code generation)
With riverpod_sqflite and codegen, use @JsonPersist() so key/encode/decode are generated:
@riverpod
@JsonPersist()
class TodoList extends _$TodoList {
@override
Future<List<Todo>> build() async {
persist(ref.watch(storageProvider.future));
return fetchTodosFromServer();
}
}
Cache duration
By default state is cached for a short time (e.g. 2 days). For long-lived data (e.g. user preferences), set StorageOptions:
persist(
ref.watch(storageProvider.future),
options: const StorageOptions(cacheTime: StorageCacheTime.unsafe_forever),
// ...
);
If using forever, plan to delete or migrate data when the app changes; Riverpod does not do migrations.
Destroy key (simple migration)
When the data shape changes, use destroyKey so old data is discarded:
options: const StorageOptions(destroyKey: '1.0'),
Bump the string in new releases; old persisted state is then ignored and the provider starts fresh.
Waiting for decode
To initialize from persisted state instead of a network call, await the persist future:
await persist(ref.watch(storageProvider.future), key: 'todo_list', ...).future;
return state.value ?? <Todo>[];
Testing
Override the storage provider with Storage.inMemory() so tests don't need a real database:
ProviderScope(
overrides: [
storageProvider.overrideWith((ref) => Storage<String, String>.inMemory()),
],
child: const MyApp(),
)
For advanced migrations or custom storage strategies, you may still need to work with the database directly.
How can the creator link this skill?
Add the canonical catalog link to the repository README so users can inspect current installs and available audits. The publishing guide covers the complete discovery path.
<a href="https://skillzs.dev/skills/serverpod/skills-registry/riverpod-offline">View riverpod-offline on skillZs</a>