embed_flutter 0.4.0
embed_flutter: ^0.4.0 copied to clipboard
A powerful Flutter SDK that provides a powerful voice-enabled AI-agent functionality with real-time widget tree monitoring for screen context fetching and realtime voice communication.
Embed Flutter SDK #
A powerful Flutter SDK that provides a powerful voice-enabled AI-agent functionality with real-time widget tree monitoring for screen context fetching and realtime voice communication.
Table of Contents #
Installation #
Add the following dependency to your pubspec.yaml:
dependencies:
embed_flutter: ^0.4.0
Then run:
flutter pub get
Android Configuration
Nothing to add. The SDK declares the permissions a call needs in its own
manifest, and they are merged into your app at build time: INTERNET,
RECORD_AUDIO, MODIFY_AUDIO_SETTINGS, ACCESS_NETWORK_STATE, WAKE_LOCK,
BLUETOOTH_CONNECT, and BLUETOOTH (Android 11 and below). It also asks for
them at runtime itself: the microphone when a call starts, and Bluetooth
(Android 12+) only when a Bluetooth headset is connected. Listing any of them
in your own manifest as well is harmless.
iOS Configuration
1. Add microphone permission to Info.plist
Add the following permissions to your ios/Runner/Info.plist:
<key>NSMicrophoneUsageDescription</key>
<string>This app needs access to microphone for voice calls.</string>
2. Configure Podfile for microphone permission
The SDK requires explicit configuration in your ios/Podfile to enable microphone permission requests. Add the following configuration to your post_install block:
post_install do |installer|
installer.pods_project.targets.each do |target|
flutter_additional_ios_build_settings(target)
target.build_configurations.each do |config|
config.build_settings['GCC_PREPROCESSOR_DEFINITIONS'] ||= [
'$(inherited)',
## dart: PermissionGroup.microphone
'PERMISSION_MICROPHONE=1',
# ...other permissions...
]
end
end
end
Important: Without the
PERMISSION_MICROPHONE=1macro in the Podfile, the microphone permission dialog will not appear on iOS and the permission will be reported as permanently denied even if it hasn't been requested before.
Getting Started #
Initialization #
Before using any EmbedWidget, you must initialize the configuration once in your main.dart:
void main() {
embedInitialize(
apiKey: 'your-api-key', // Required
appVersion: '1.0.0', // Required
embedUrl: 'https://embed.revrag.ai', // Optional
);
runApp(const MyApp());
}
Required: Navigator Observer
For route-based activation to work properly, you must add EmbedNavigatorObserver to your app's navigator observers. This enables the SDK to track route changes and activate/deactivate the EmbedWidget accordingly.
For GoRouter:
import 'package:go_router/go_router.dart';
import 'package:embed_flutter/embed_flutter.dart';
final GoRouter router = GoRouter(
observers: [EmbedNavigatorObserver()], // Required for route tracking
routes: [
// Your routes here
],
);
For MaterialApp:
import 'package:embed_flutter/embed_flutter.dart';
MaterialApp(
navigatorObservers: [EmbedNavigatorObserver()], // Required for route tracking
// Other MaterialApp properties
)
Required: USER_DATA Event
To activate the EmbedWidget on the screen, you must call the USER_DATA embedEvent with a user ID. This is typically done after user authentication or when you have a user identifier:
import 'package:embed_flutter/embed_flutter.dart';
// Call this after user authentication or when you have a user ID
embedEvent(
EventKeys.USER_DATA,
UserEventPayload(
app_user_id: 'user_12345', // Required: Unique user identifier
data: {
'name': 'John Doe', // Optional: Additional user data
'email': 'john@example.com',
'phone': '+1234567890',
},
),
);
Important: The EmbedWidget will not be fully functional until the
USER_DATAevent is sent. This event initializes the voice communication features.
Optional: SCREEN_STATE Event
You can optionally send SCREEN_STATE events to provide additional context about screen changes or user navigation:
// Send screen state when navigating to a new screen
embedEvent(
EventKeys.SCREEN_STATE,
ScreenEventPayload(
screen: 'product_selection_screen', // Required: Screen identifier
data: {
'category': 'banking', // Optional: Additional screen context
'user_type': 'premium',
'flow_step': 'selection',
},
),
);
Optional: CUSTOM_EVENT
CUSTOM_EVENT lets you capture bespoke analytics or interaction events that matter to your flows (for example, plan selections, button taps, or offer views). Supply any key/value metadata you want to track:
embedEvent(
EventKeys.CUSTOM_EVENT,
CustomEventPayload(
data: {
'context': 'pricing_screen',
'event': 'option_selected',
'option_id': 'plan_12_months',
'price_per_month': 249,
'currency': 'INR',
},
),
);
Tip: Combine
CUSTOM_EVENTwith your flow configuration to build rich analytics around user journeys.
Optional: ANALYTICS_DATA Event
ANALYTICS_DATA is similar to CUSTOM_EVENT but requires an explicit event name. Use this when you need to track structured analytics events with a specific event identifier. This is useful for tracking user actions, conversions, permission status changes or other analytics events that need to be categorized by name:
embedEvent(
EventKeys.ANALYTICS_DATA,
AnalyticsDataEventPayload(
event_name: 'purchase_completed', // Required: Event identifier
data: {
'product_id': 'prod_12345',
'amount': 99.99,
'currency': 'USD',
'payment_method': 'credit_card',
},
),
);
Tip: Use
ANALYTICS_DATAwhen you need to track events with a specific name for analytics purposes, whileCUSTOM_EVENTis better for general-purpose custom tracking without a named event.
Available Events
| Event | Purpose | Required | When to Use |
|---|---|---|---|
USER_DATA |
Initialize user context and activate EmbedWidget | Required | After user authentication or when user ID is available |
SCREEN_STATE |
Provide screen context and navigation info | Optional | When navigating between screens or when screen context changes |
CUSTOM_EVENT |
Capture bespoke user interactions or analytics | Optional | Whenever you need additional tracking for specific actions |
ANALYTICS_DATA |
Track structured analytics events with event name | Optional | When tracking named analytics events (e.g., conversions, purchases, permission status changes) |
Basic Usage #
Here's a complete example showing how to use the SDK with route-based activation:
import 'package:flutter/material.dart';
import 'package:embed_flutter/embed_flutter.dart';
void main() {
// 1. Initialize the SDK
embedInitialize(apiKey: 'your-api-key', appVersion: '1.0.0');
runApp(const MyApp());
}
class MyApp extends StatelessWidget {
const MyApp({super.key});
@override
Widget build(BuildContext context) {
return EmbedWidget(
showEmbedWidget: true,
// Configure route-based activation
enabledRoutes: const ['home', 'product_selection_screen', 'checkout', 'payment'],
child: MaterialApp(
navigatorObservers: [EmbedNavigatorObserver()], // Required for route tracking
routes: [
'home': (context) => const MyHomePage(),
'product_selection_screen': (context) => const ProductSelectionScreen(),
'checkout': (context) => const CheckoutScreen(),
'payment': (context) => const PaymentScreen(),
]
home: MyHomePage(),
),
);
}
}
class MyHomePage extends StatefulWidget {
@override
State<MyHomePage> createState() => _MyHomePageState();
}
class _MyHomePageState extends State<MyHomePage> {
@override
void initState() {
super.initState();
// 2. Send USER_DATA event to activate EmbedWidget
// This should typically be called after user authentication
embedEvent(
EventKeys.USER_DATA,
UserEventPayload(
app_user_id: 'user_12345', // Replace with actual user ID
data: {
'name': 'John Doe',
'email': 'john@example.com',
'plan': 'premium',
},
),
);
}
void _navigateToProductScreen() {
// 3. Send SCREEN_STATE event for additional context (Optional)
embedEvent(
EventKeys.SCREEN_STATE,
ScreenEventPayload(
screen: 'product_selection_screen',
data: {
'category': 'banking_products',
'user_segment': 'new_customer',
},
),
);
// Navigate to product screen
Navigator.pushNamed(context, 'product_selection');
}
@override
Widget build(BuildContext context) {
return Scaffold(
appBar: AppBar(title: const Text('My App')),
body: Center(
child: Column(
mainAxisAlignment: MainAxisAlignment.center,
children: [
const Text('Welcome to My App!'),
const SizedBox(height: 20),
ElevatedButton(
onPressed: _navigateToProductScreen,
child: const Text('Go to Product Selection'),
),
],
),
),
);
}
}
Required Parameters #
child(required): The main content of your app
Route-Based Activation #
The agent button shows only on the routes you list in enabledRoutes. When the
user navigates, EmbedNavigatorObserver checks the new route against that list
and shows or hides the button — no per-route code needed. (One integration is
one agent; there is no per-flow/multi-agent scoping.)
Basic route configuration
class MyApp extends StatelessWidget {
const MyApp({super.key});
@override
Widget build(BuildContext context) {
return EmbedWidget(
enabledRoutes: const [
'personal_info',
'personal_details',
'product_selection_screen',
'review_confirm_screen',
'aadhar_verification_screen',
],
child: MaterialApp.router(
routerConfig: appRouter,
),
);
}
}
Route configuration options
enabledRoutes:List<String>— route names the agent button may show on.showOnAllRoutes: Show the agent button on all routes (default:false).disabledRoutes:List<String>— route names where the button is always hidden.routeMatchMode: How route names are matched:RouteMatchMode.exact— exact match (default)RouteMatchMode.startsWith— route starts with patternRouteMatchMode.contains— route contains pattern
Optional Parameters #
embedUrl(optional): Custom base URL for API endpoints — set viaembedInitialize()showEmbedWidget(default:true): Controls whether the embedded widget is visibleenabledRoutes(optional):List<String>- route names where the agent button appearsshowOnAllRoutes(default:false): Show EmbedWidget on all routesdisabledRoutes(optional): List of route names where EmbedWidget should be disabledrouteMatchMode(default:RouteMatchMode.exact): How to match routescontinuity(optional):EmbedButtonContinuity(keepVisible/reset) — whether the FAB stays up or re-shows when navigating between enabled routesembedButtonDelayMs(default:0): Global pre-show delay (ms) before the FAB appears on an enabled routebuttonDelay(optional):EmbedButtonDelay— a delay +EmbedButtonDelayPolicy(perScreen/oncePerGroupEntry/oncePerAppSession)buttonVisibility(optional):EmbedButtonVisibilityConfig— per-screen visibility groups, each with its own continuity / delay / policy (React Native parity)onBackRequested(optional):Future<bool> Function()?— first-chance handler for the agent's "go back"onPermissionStatusChanged(optional): A callback function that receives a boolean indicating whether microphone permission is denied or permanently denied. If the microphone permission is not given, the initial permission request is prompted automatically when the call is started. If the user denies it, the result can be handled accordingly. Called when a call is initiated:true- Microphone permission was grantedfalse- Microphone permission was denied or permanently denied
Embed Widget Position
The agent button's position (and colors) come from the backend widget_config
(widget_position / right_padding / bottom_padding), configured from the
Revrag dashboard — there are no host position props on EmbedWidget.
Using onPermissionStatusChanged
The onPermissionStatusChanged callback allows you to handle microphone permission status changes when a user attempts to start a voice call. This is useful for showing custom UI feedback or handling permission denial gracefully:
import 'package:flutter/material.dart';
import 'package:embed_flutter/embed_flutter.dart';
class MyApp extends StatelessWidget {
const MyApp({super.key});
// Global key for ScaffoldMessenger to show SnackBars from callbacks
static final GlobalKey<ScaffoldMessengerState> scaffoldMessengerKey =
GlobalKey<ScaffoldMessengerState>();
@override
Widget build(BuildContext context) {
return EmbedWidget(
showEmbedWidget: true,
// Handle permission status changes
onPermissionStatusChanged: (bool isGranted) {
if (isGranted) {
// Permission granted - call can proceed
print('Microphone permission granted');
// You can show a success message or update UI here
} else {
// Permission denied - handle accordingly
print('Microphone permission denied');
// You can show an error message or guide user to settings
scaffoldMessengerKey.currentState?.showSnackBar(
const SnackBar(
content: Text('Microphone permission is required for voice calls. Please enable it in settings and re-open the app.'),
duration: Duration(seconds: 3),
),
);
}
},
enabledRoutes: const ['home', 'product_selection_screen'],
child: MaterialApp(
scaffoldMessengerKey: scaffoldMessengerKey,
navigatorObservers: [EmbedNavigatorObserver()],
home: MyHomePage(),
),
);
}
}
When is it called?
- The callback is triggered when a user attempts to start a call
- It's called after the permission request dialog is shown (if needed)
- It's called before the call connection is established
Examples #
Basic Integration #
import 'package:flutter/material.dart';
import 'package:embed_flutter/embed_flutter.dart';
void main() {
// Initialize the SDK
embedInitialize(apiKey: 'your-api-key-here', appVersion: '1.0.0');
runApp(const MyApp());
}
class MyApp extends StatelessWidget {
const MyApp({super.key});
@override
Widget build(BuildContext context) {
return EmbedWidget(
showEmbedWidget: true,
child: MaterialApp(
navigatorObservers: [EmbedNavigatorObserver()], // Required for route tracking
home: MyHomePage(),
),
);
}
}
class MyHomePage extends StatefulWidget {
@override
State<MyHomePage> createState() => _MyHomePageState();
}
class _MyHomePageState extends State<MyHomePage> {
@override
void initState() {
super.initState();
// Send USER_DATA event to activate EmbedWidget
// This should typically be called after user authentication
embedEvent(
EventKeys.USER_DATA,
UserEventPayload(
app_user_id: 'user_12345', // Replace with actual user ID
data: {
'name': 'John Doe',
'email': 'john@example.com',
},
),
);
}
@override
Widget build(BuildContext context) {
return Scaffold(
appBar: AppBar(title: const Text('My App')),
body: const Center(
child: Column(
mainAxisAlignment: MainAxisAlignment.center,
children: [
Text('Welcome to My App!'),
SizedBox(height: 20),
Text('The embedded agent is monitoring this screen.'),
],
),
),
);
}
}
Form with Monitoring #
import 'package:flutter/material.dart';
import 'package:embed_flutter/embed_flutter.dart';
void main() {
// Initialize the SDK
embedInitialize(apiKey: 'your-api-key-here', appVersion: '1.0.0');
runApp(const MyApp());
}
class MyApp extends StatelessWidget {
const MyApp({super.key});
@override
Widget build(BuildContext context) {
return EmbedWidget(
showEmbedWidget: true,
child: MaterialApp(
navigatorObservers: [EmbedNavigatorObserver()], // Required for route tracking
home: const MyFormPage(),
),
);
}
}
class MyFormPage extends StatefulWidget {
const MyFormPage({super.key});
@override
State<MyFormPage> createState() => _MyFormPageState();
}
class _MyFormPageState extends State<MyFormPage> {
final _formKey = GlobalKey<FormState>();
final _nameController = TextEditingController();
final _emailController = TextEditingController();
@override
void initState() {
super.initState();
// Send USER_DATA event to activate EmbedWidget
// This should typically be called after user authentication
embedEvent(
EventKeys.USER_DATA,
UserEventPayload(
app_user_id: 'user_12345', // Replace with actual user ID
data: {
'name': 'John Doe',
'email': 'john@example.com',
},
),
);
// Send SCREEN_STATE event for additional context
embedEvent(
EventKeys.SCREEN_STATE,
ScreenEventPayload(
screen: 'form_page',
data: {
'form_type': 'user_registration',
'step': 'personal_info',
},
),
);
}
@override
Widget build(BuildContext context) {
return Scaffold(
appBar: AppBar(title: const Text('My Form')),
body: Form(
key: _formKey,
child: Padding(
padding: const EdgeInsets.all(16.0),
child: Column(
children: [
TextFormField(
controller: _nameController,
decoration: const InputDecoration(labelText: 'Name'),
validator: (value) {
if (value == null || value.isEmpty) {
return 'Please enter your name';
}
return null;
},
),
const SizedBox(height: 16),
TextFormField(
controller: _emailController,
decoration: const InputDecoration(labelText: 'Email'),
validator: (value) {
if (value == null || value.isEmpty) {
return 'Please enter your email';
}
return null;
},
),
const SizedBox(height: 20),
ElevatedButton(
onPressed: () {
if (_formKey.currentState!.validate()) {
// Form is valid - EmbedWidget will track the submission
ScaffoldMessenger.of(context).showSnackBar(
const SnackBar(content: Text('Form submitted successfully!')),
);
}
},
child: const Text('Submit'),
),
],
),
),
),
);
}
@override
void dispose() {
_nameController.dispose();
_emailController.dispose();
super.dispose();
}
}
Route-Based Configuration #
Configure the EmbedWidget with a flat list of routes where the agent button should appear:
import 'package:flutter/material.dart';
import 'package:embed_flutter/embed_flutter.dart';
void main() {
embedInitialize(apiKey: 'your-api-key', appVersion: '1.0.0');
runApp(const MyApp());
}
class MyApp extends StatelessWidget {
const MyApp({super.key});
@override
Widget build(BuildContext context) {
return EmbedWidget(
enabledRoutes: const [
'welcome',
'personal-info',
'personal-details',
'product-selection',
'payment',
'review-confirm',
],
child: MaterialApp(
navigatorObservers: [EmbedNavigatorObserver()], // Required for route tracking
initialRoute: '/',
routes: {
'/': (context) => const SplashScreen(),
'welcome': (context) => const WelcomeScreen(),
'personal-info': (context) => const PersonalInfoScreen(),
'personal-details': (context) => const PersonalDetailsScreen(),
'product-selection': (context) => const ProductSelectionScreen(),
},
),
);
}
}
Route behavior
- On an enabled route: the agent button is visible; navigating between
enabled routes keeps it up (subject to
continuity). - Off the enabled routes: the button is hidden and its resources cleaned up; any active call ends.
Custom Router Integration #
An Example to show Full integration with Custom Routers we can take example of GoRouter for declarative navigation:
import 'package:go_router/go_router.dart';
import 'package:embed_flutter/embed_flutter.dart';
final appRouter = GoRouter(
initialLocation: '/splash',
routes: [
GoRoute(
path: '/splash',
name: 'splash_screen',
builder: (context, state) => const SplashScreen(),
),
GoRoute(
path: '/welcome',
name: 'welcome_screen',
builder: (context, state) => const WelcomeScreen(),
),
GoRoute(
path: '/personal-info',
name: 'personal_info',
builder: (context, state) => const PersonalInfoScreen(),
),
// ... more routes
],
// Add EmbedNavigatorObserver to the list of observers
observers: [
EmbedNavigatorObserver(),
// ... other observers
],
);
void main() {
WidgetsBinding widgetsBinding = WidgetsFlutterBinding.ensureInitialized();
widgetsBinding = EmbedWidgetsBinding();
embedInitialize(apiKey: 'your-api-key', appVersion: '1.0.0');
runApp(const MyApp());
}
class MyApp extends StatelessWidget {
const MyApp({super.key});
@override
Widget build(BuildContext context) {
return EmbedWidget(
enabledRoutes: const [
'welcome_screen', 'personal_info', 'product_selection_screen',
'checkout', 'payment',
],
child: MaterialApp.router(
routerConfig: appRouter,
),
);
}
}
Configuration #
UI capture and screen scope #
During a call, one semantics capture feeds the live agent and the on-device screen cache. The cache receives the capture before live node and byte budgets are applied. Outside a call, the cache captures independently.
The live feed sends a full baseline on connection, screen changes, and
request_ui_tree; ordinary updates use diffs (an app foreground/background
flush pushes a diff, or a baseline only when the agent has none yet). Every
tree frame carries app_user_id (once USER_DATA has been sent),
app_version, and sdk {sdk_name, sdk_version, platform} — the same values
/ui-graph sync sends. Each serialized LiveKit payload is limited to 15,000
UTF-8 bytes. Snapshots prioritize visible controls and trim to fit, envelope
included. Mission exploration audits trim to fit as well (per-node paths
first, then older screens, then long text, then the least useful nodes of the
current screen) and are marked truncated: true. find_candidates_response
sheds its lowest-ranked candidates and info_query_response shortens its
value; a response that still cannot be sent goes out as an explicit failure
(failure_reason) so the agent never waits for a reply that will not come.
Two questions are kept apart:
- May the agent look? Missions the agent runs (find, tap, scroll, read a
field, verify) are blocked only by the host's explicit exclusions:
disabledRoutes(matching the current route or the detected screen name) and an open named popup listed there.enabledRoutes, the platform allowlist and visibility-group routes decide where the button shows and what telemetry is sent, but they do not blind a mission — a mission that navigates to a screen outside the allowlist still sees it on its next step (Android parity: absence from an include list is a default, an exclusion is a decision).action_config.flags.dataCaptureand widget visibility do not blind a mission either — data capture off means "no telemetry", not "agent cannot act". To keep the agent away from a screen, list it indisabledRoutes. A response the host refuses is replaced by a failure frame, and a mission phase update is sent without its screen text. - May the SDK send telemetry? Live tree pushes (
ui_diff) and themission_explorationaudit follow the full scope (disabledRoutesAND the enabled-route list / allowlist) and additionally requiredataCapture(default on). While the widget is hidden (e.g. during the per-screen button delay) they continue by default; a tenant can opt out with an explicitaction_config.flags.transmitOnHiddenWidget: false(ortreePush.transmitOnHiddenWidget: false, Android's spelling). This flag never overrides a route exclusion.
Queued content is checked again against its original route before delivery. The cache may continue on allowed screens while the widget is hidden. Previously persisted files are not deleted by changing route rules.
Keep EmbedNavigatorObserver attached to identify popup routes such as dialogs,
bottom sheets, and menus. These remain part of their underlying screen (they
never change the page route, FAB visibility, or the page's permissions) and
have separate cache records; dismissed overlays retain present: false. A
named popup can be excluded like a screen: list its route name (its
RouteSettings.name, or what your nameExtractor returns) in
disabledRoutes, and nothing is captured or sent — for missions or telemetry —
while it is open. Excluding a popup blocks capture only: it does not hide
the agent button (the button follows the page route underneath):
showDialog(
context: context,
routeSettings: const RouteSettings(name: 'otp_dialog'),
builder: (_) => const OtpDialog(),
);
// EmbedWidget(disabledRoutes: const ['otp_dialog'], …)
Named popups also emit SCREEN_STATE analytics on open/close. Custom
OverlayEntry content without modal semantics and native platform windows are
not automatically grouped and cannot be excluded by name — exclude the
underlying screen instead.
Stable screen names
Every capture is filed under a screen name, which keys the on-device cache,
prefixes element stable IDs, and tells the agent that a tap changed screens.
The SDK uses, in order: the current route name (RouteSettings.name, what
your nameExtractor returns, or a name set through EmbedRouteListener),
then stable signals from the widget tree (a ValueKey<String> on a tab root,
Semantics(identifier:), or the page's first short heading), then unknown_flow.
Class names are never used, so debug, release and obfuscated builds name every
screen identically. Name your routes (and wrap in-page tabs with
EmbedRouteListener) for the clearest names — an unnamed page with no stable
signal is named after the route below it as route/page_N.
Remote effect and fallback switches
action_config.flags.nativeBridge: falseturns off the synthetic-pointer tap fallback (the last rung of the tap ladder). Default on; the primary accessibility-action path is never affected.action_config.effects.*Msvalues of0are accepted but floored at 1 ms, so a tenant asking for "no animation" gets an imperceptible one instead of a crash.
WireLog.enabled prints full wire bodies for debugging and is forced off in
release builds.
Icon-only buttons and custom tab bars
The agent finds controls by their accessibility label. For an icon-only
tappable with no label or tooltip, the SDK falls back to a name derived from
what it draws: the asset file name (assets/nav/ic_home_active.png →
"home"), a common Material/Cupertino icon (Icons.pie_chart → "pie chart"),
or, for an unnamed bottom-nav item, its position ("Tab 3 of 4"). For a custom
bottom bar with no Semantics(selected:), the active item is inferred only
when it alone is drawn differently, or from tab taps the session has seen
lead to the current screen. For best results, label icon-only nav items and
mark the active one yourself:
Semantics(
label: 'Portfolio',
selected: currentIndex == 2,
child: GestureDetector(onTap: () => onTab(2), child: const PortfolioIcon()),
)
Disabled controls (onPressed: null, TextField(enabled: false)) are still
found, but a tap or set_text on them fails at once with target_disabled.
Custom Base URL #
You can set a custom base URL globally during initialization:
void main() {
embedInitialize(
apiKey: 'your-api-key',
embedUrl: 'https://your-custom-domain.com',
appVersion: '1.0.0',
);
runApp(const MyApp());
}
Troubleshooting #
Common Issues #
-
Widget Not Visible
- Check
showEmbedWidgetparameter - Verify API key is valid
- Check network connectivity
- Check
-
Voice Call Issues / Microphone Permission Not Being Requested
-
Ensure microphone permission is granted and not permanently denied.
-
Check device audio settings
-
iOS: Verify that
NSMicrophoneUsageDescriptionis added toios/Runner/Info.plist -
iOS: Ensure the Podfile includes the permission handler configuration. Add the following to your
ios/Podfileif it's missing:post_install do |installer| installer.pods_project.targets.each do |target| target.build_configurations.each do |config| config.build_settings['GCC_PREPROCESSOR_DEFINITIONS'] ||= ['$(inherited)'] config.build_settings['GCC_PREPROCESSOR_DEFINITIONS'] << 'PERMISSION_MICROPHONE=1' end end endAfter adding this configuration, run:
cd ios pod install cd ..
-
-
Widget Tree Not Updating
- Check widget tree handler initialization
- Verify context is available
- Check for widget disposal issues
-
Route-Based Activation Not Working
- Verify route names match exactly (case-sensitive)
- Check
EmbedNavigatorObserveris added to router observers - Ensure GoRouter is properly configured with named routes
- Verify
enabledRoutescontains the correct route names - Verify the current route is listed in
enabledRoutes
-
Click Events Missing or Badly Named
- Clicks are sent only after
USER_DATA, while the widget is shown on the route, and whileaction_config.flags.clickTrackingisn'tfalse - A control is named by its
ValueKey<String>, thenSemantics(identifier:), then its tooltip / visible text, then the image / SVG file name or common icon it draws. Give it aValueKey<String>for a stable name (ValueKey('apply_loan')→apply_loan_pressed) - A tap needs semantics: a custom tappable with no
onTapsemantics (e.g. a rawListener) isn't reported. Wrap it inSemantics(button: true, onTap: …) - See INTEGRATION.md → Click analytics for the full naming rules
- Clicks are sent only after
-
EmbedWidget Not Appearing on Expected Screens
- Check
enabledRoutesconfiguration (a flat list of route names) - Check
disabledRoutesconfiguration - Verify route matching mode (
exactvsstartsWithvscontains) - Check if
showEmbedWidgetis set totrue - Verify the route is listed in
enabledRoutes
- Check
License #
This project is licensed under the Non-Commercial License License - see the LICENSE file for details.
Support #
- Documentation: https://docs.revrag.ai/embed/integration/flutter
- Email Support: contact@revrag.ai
- GitHub Issues: https://github.com/RevRag-ai/embed-react-native/issues
Changelog #
See CHANGELOG.md for a complete list of changes and version history.