Call Kit for ZEGO

Real-time audio/video calling for Flutter, built on ZEGOCLOUD, with native CallKit on iOS and ConnectionService on Android for the incoming-call UI.

Unofficial. Not affiliated with or endorsed by ZEGOCLOUD — it builds on their zego_express_engine SDK.

Pub Version License: MIT

Features

  • Audio & Video Calling - High-quality real-time communication using ZEGOCLOUD
  • In-Session Chat - Text messaging with typing indicators and delivery/read receipts
  • Flexible CallKit Integration - Choose between native CallKit UI or custom app UI
    • Native mode: Full-screen incoming call UI on locked screens (iOS/Android)
    • Custom mode: Complete control over call UI with your own design
  • Socket Agnostic - Works with any WebSocket/Socket.IO implementation
  • State Machine - Robust call state management with transition guards
  • Network Resilience - Automatic reconnection with exponential backoff
  • Plugin Architecture - Optional analytics, crash reporting, and call storage
  • WhatsApp-like UX - 30-second ringing timeout, proper call lifecycle
  • Demo signaling server included - a working Node server in demo_server/ so you can place a real call before writing any backend

Table of Contents

Demo Server — place a real call in 5 minutes

This repo ships a complete signaling server in demo_server/, so you can make a working call before writing any backend code.

Calling needs two halves: media (ZEGOCLOUD handles that) and signaling — telling the other person "you have a call". This package is signaling-agnostic on purpose, so you provide the second half. The demo server is a working implementation you can call against today and copy from later.

It is a test harness, not a backend: no auth, no database, everything lives in memory and dies with the process.

1. Start the server

cd demo_server
npm install
npm start

It listens on port 3000. Check it and see who is connected:

curl localhost:3000/health

2. Point the app at it

The example reads the URL from a --dart-define:

cd example
flutter run --dart-define=SIGNALING_URL=http://192.168.1.20:3000
Where the app runs URL to use
Android emulator http://10.0.2.2:3000 (the default)
iOS simulator http://localhost:3000
Real device http://<your-machine-LAN-IP>:3000

Find your IP with ipconfig getifaddr en0 (macOS) or hostname -I (Linux). A real device must be on the same Wi-Fi as your machine.

10.0.2.2 is an emulator-only alias for your machine's localhost. On a physical device it resolves to nothing and you get [SocketAdapter] Connection error: timeout.

3. Make the call

Add your ZEGOCLOUD appId / appSign in example/lib/main.dart — signaling works without them, but there will be no audio or video. Then launch the app on two devices, pick a different user on each, and call.

Accept the microphone and camera prompts on both. A denied mic does not fail the call — it publishes an empty stream, so you get a connected call with silence.

Verify signaling without any devices

npm test in demo_server/ drives two fake clients through a full call — ring, accept, chat, hang up, plus the reject and offline paths — asserting the exact field names the mapper reads:

cd demo_server && npm test

Writing your own server

demo_server/README.md documents the full protocol — every event in both directions and its payload — plus the three things a real backend must get right. Each of them fails silently, as a call that connects and looks fine:

  1. sessionStarted must reach both parties, and it is what starts media. The controller deliberately does not join the ZEGO room on the offer or the accept; it waits for your authoritative roomId here.
  2. Both parties need the identical roomId. Two different ones means two people alone in two rooms, with the UI showing "connected" throughout.
  3. Never infer direction from userId / listenerId. The mapper puts whoever is emitting in userId, so the callee occupies that field on accept-request. Look the call up by requestId instead.

Installation

Add to your pubspec.yaml:

dependencies:
  call_kit_for_zego: ^1.0.0

Then run:

flutter pub get

Quick Start

1. Initialize the Service

import 'package:call_kit_for_zego/call_kit_for_zego.dart';

// Initialize the service
await RtcCommunicationService.initialize(
  config: CommunicationConfig(
    zegoConfig: ZegoConfig(
      appId: YOUR_ZEGO_APP_ID,      // From ZEGOCLOUD console
      appSign: 'YOUR_ZEGO_APP_SIGN', // From ZEGOCLOUD console
    ),
    socketConfig: SocketConfig(
      url: 'https://your-server.com',
    ),
    callKitConfig: CallKitConfig(
      appName: 'Your App Name',
      iconName: 'app_icon',  // Optional: custom icon for incoming call
    ),
    useCallKit: true,  // Optional: Use native CallKit (default: true)
  ),
  socketAdapter: YourSocketAdapter(),  // Your socket implementation
);

// Get the service instance
final rtcService = RtcCommunicationService.instance;

💡 Note: Set useCallKit: false to use your own custom incoming call UI instead of native CallKit. See CallKit Options.

2. Set Current User

rtcService.setCurrentUser(UserInfo(
  userId: 'user_123',
  userName: 'John Doe',
  userImage: 'https://example.com/avatar.jpg',
));

3. Listen to Events

// Listen to call events
rtcService.callEvents.listen((event) {
  switch (event.type) {
    case CallEventType.incomingCall:
      // Show incoming call UI (or let native CallKit handle it)
      break;
    case CallEventType.callAccepted:
      // Navigate to call screen
      break;
    case CallEventType.callEnded:
      // Clean up and navigate back
      break;
    case CallEventType.callDeclined:
      // Show "call declined" message
      break;
  }
});

4. Make a Call

await rtcService.startCall(
  calleeId: 'user_456',
  calleeName: 'Jane Smith',
  calleeImage: 'https://example.com/jane.jpg',
  callType: CallType.video,
);

5. Handle Incoming Calls

// Accept call
await rtcService.acceptCall();

// Reject call
await rtcService.rejectCall();

6. End Call

await rtcService.endCall();

CallKit Options

The package offers two modes for handling incoming calls:

Native CallKit Mode (Default)

When useCallKit = true (default), the package uses native iOS CallKit and Android ConnectionService to display full-screen incoming call UI, even on locked screens.

Perfect for:

  • WhatsApp/Telegram-like call experience
  • System-level call integration
  • Calls that work on locked screens
  • Less implementation work

Configuration:

await RtcCommunicationService.initialize(
  config: CommunicationConfig(
    zegoConfig: ZegoConfig(
      appId: YOUR_ZEGO_APP_ID,
      appSign: 'YOUR_ZEGO_APP_SIGN',
    ),
    socketConfig: SocketConfig(url: 'https://your-server.com'),
    callKitConfig: CallKitConfig(
      appName: 'My App',
      iconName: 'app_icon',
    ),
    useCallKit: true, // ✅ Native CallKit UI (default)
  ),
  socketAdapter: socketAdapter,
);

What you get:

  • ✅ Full-screen incoming call on locked screen
  • ✅ System ringtone
  • ✅ Native accept/decline buttons
  • ✅ iOS CallKit & Android ConnectionService integration
  • ✅ Your callbacks still fire for navigation

Custom UI Mode

When useCallKit = false, you have complete control over the incoming call UI using your own custom screens.

Perfect for:

  • Fully custom call UI design
  • Unique UX requirements
  • Showing additional context on call screen
  • Testing without native UI complexity

Configuration:

await RtcCommunicationService.initialize(
  config: CommunicationConfig(
    zegoConfig: ZegoConfig(
      appId: YOUR_ZEGO_APP_ID,
      appSign: 'YOUR_ZEGO_APP_SIGN',
    ),
    socketConfig: SocketConfig(url: 'https://your-server.com'),
    useCallKit: false, // ❌ No native CallKit - you handle UI
  ),
  socketAdapter: socketAdapter,
);

// Listen to events and show your custom UI
rtcService.callEvents.listen((event) {
  if (event.type == CallEventType.incomingCall) {
    // Show YOUR custom incoming call screen
    Navigator.push(
      context,
      MaterialPageRoute(
        builder: (_) => MyCustomIncomingCallScreen(
          callerName: event.callData?.caller.userName,
          callerImage: event.callData?.caller.avatarUrl,
          onAccept: () async {
            await rtcService.acceptCall();
            // Navigate to call screen
          },
          onReject: () async {
            await rtcService.rejectCall();
            Navigator.pop(context);
          },
        ),
      ),
    );
  }
});

What you get:

  • ✅ Complete control over UI design
  • ✅ Show custom info (caller bio, mutual friends, etc.)
  • ✅ Custom animations and styling
  • ✅ Easier testing and debugging
  • ⚠️ Requires more implementation work
  • ⚠️ Won't work on locked screen without push notifications

Comparison:

Feature Native CallKit (true) Custom UI (false)
Full-screen on locked screen ✅ ❌
System integration ✅ ❌
Custom UI design Limited ✅ Full control
Implementation complexity ⭐ Easy ⭐⭐⭐ More work
Works with system UI ✅ ❌

💡 Tip: Start with useCallKit = false during development for easier testing, then switch to true for production.

📖 Full guide: See CallKit usage below for detailed examples and a migration guide.

Architecture

┌─────────────────────────────────────────────────────────────┐
│                    Your Flutter App                          │
├─────────────────────────────────────────────────────────────┤
│                                                             │
│  ┌─────────────────┐       ┌─────────────────────────────┐ │
│  │  Your Socket    │       │   RtcCommunicationService   │ │
│  │  (Socket.IO,    │──────▶│   (Main Entry Point)        │ │
│  │   WebSocket)    │       └─────────────────────────────┘ │
│  └─────────────────┘                    │                   │
│                                         │                   │
│          ┌──────────────────────────────┼──────────────┐   │
│          │                              │              │   │
│          ▼                              ▼              ▼   │
│  ┌───────────────┐          ┌───────────────┐  ┌──────────┐│
│  │ CallManager   │          │ SessionManager│  │ ChatMgr  ││
│  │ (Call Events) │          │ (User State)  │  │ (Text)   ││
│  └───────────────┘          └───────────────┘  └──────────┘│
│          │                                                  │
│          ▼                                                  │
│  ┌───────────────────────────────────────────────────────┐ │
│  │              RtcCallController                         │ │
│  │  ┌─────────────┐  ┌─────────────┐  ┌────────────────┐ │ │
│  │  │ ZegoService │  │ StateManager│  │ NativeCallKit  │ │ │
│  │  │ (A/V Engine)│  │ (State M/C) │  │ (iOS/Android)  │ │ │
│  │  └─────────────┘  └─────────────┘  └────────────────┘ │ │
│  └───────────────────────────────────────────────────────┘ │
└─────────────────────────────────────────────────────────────┘

Configuration

ZegoConfig

ZegoConfig(
  appId: 123456789,           // Required: Your ZEGOCLOUD App ID
  appSign: 'your_app_sign',   // Required: Your ZEGOCLOUD App Sign
)

CallKitConfig

CallKitConfig(
  appName: 'My App',              // Shown on incoming call screen
  iconName: 'app_icon',           // Icon resource name (optional)
  ringtoneSound: 'ringtone.mp3',  // Custom ringtone (optional)
  duration: 30000,                // Ring duration in ms (default: 30s)
  textAccept: 'Accept',           // Accept button text
  textDecline: 'Decline',         // Decline button text
  missedCallNotification: true,   // Show missed call notification
  android: AndroidCallKitConfig(
    isShowLogo: true,
    isShowFullLockedScreen: true,
  ),
  ios: IOSCallKitConfig(
    supportsVideo: true,
    maximumCallGroups: 1,
  ),
)

CommunicationConfig

CommunicationConfig(
  zegoConfig: ZegoConfig(...),       // Required: ZEGO configuration
  socketConfig: SocketConfig(...),   // Required: Socket configuration
  callKitConfig: CallKitConfig(...), // Optional: CallKit configuration
  useCallKit: true,                  // Optional: Enable/disable native CallKit (default: true)
  signalingMapper: MyCustomMapper(), // Optional: Custom signaling mapper
  enableLogging: true,               // Optional: Enable debug logs (default: true)
  logLevel: LogLevel.info,           // Optional: Log level (default: info)
)

useCallKit parameter:

  • true (default): Native CallKit handles incoming call UI
  • false: Your app handles UI using callbacks
  • See CallKit Options section for detailed comparison

RtcCallConfig (Advanced)

RtcCallConfig(
  // Call lifecycle
  ringingTimeout: Duration(seconds: 30),        // Auto-cancel if unanswered

  // Reconnection (exponential backoff: 2s, 4s, 8s, 16s, 30s)
  reconnectRetryCount: 5,
  reconnectBackoff: Duration(seconds: 2),
  maxReconnectBackoff: Duration(seconds: 30),

  // Audio
  audioBitrate: 48,                             // kbps
  enableEchoCancellation: true,                 // AEC
  enableNoiseSuppression: true,                 // ANS
  enableAutoGainControl: true,                  // AGC

  // Video (defaults are portrait HD, matching phone screens)
  videoBitrate: 2000,                           // kbps
  videoWidth: 720,
  videoHeight: 1280,

  // Diagnostics
  enableLogs: true,
  logLevel: RtcLogLevel.info,                   // verbose | info | warning | error
  enableAudioDebug: false,                      // Detailed audio pipeline logs
  audioDebugInProduction: false,                // Keep those logs in release builds
)

Two presets are provided:

RtcCallConfig.development()  // verbose logging, 60s ringing timeout
RtcCallConfig.production()   // info logging, 30s ringing timeout

Socket Integration

The package is socket-agnostic. You need to provide a SocketAdapter that wraps your socket implementation.

SocketAdapter Interface

abstract class SocketAdapter {
  /// Whether socket is currently connected
  bool get isConnected;

  /// Register event listener
  void on(String event, Function(dynamic) callback);

  /// Remove event listener
  void off(String event);

  /// Emit event with data
  void emit(String event, dynamic data);
}

Socket.IO Implementation Example

import 'package:socket_io_client/socket_io_client.dart' as IO;
import 'package:call_kit_for_zego/call_kit_for_zego.dart';

class SocketIOAdapter implements SocketAdapter {
  final IO.Socket _socket;

  SocketIOAdapter(this._socket);

  @override
  bool get isConnected => _socket.connected;

  @override
  void on(String event, Function(dynamic) callback) {
    _socket.on(event, callback);
  }

  @override
  void off(String event) {
    _socket.off(event);
  }

  @override
  void emit(String event, dynamic data) {
    _socket.emit(event, data);
  }
}

// Usage
final socket = IO.io('https://your-server.com', IO.OptionBuilder()
  .setTransports(['websocket'])
  .enableAutoConnect()
  .build());

final adapter = SocketIOAdapter(socket);
await rtcService.initialize(
  config: config,
  socketAdapter: adapter,
);

Custom Signaling Mapper

If your backend uses different event names/formats, implement SignalingMapper:

class MyBackendMapper implements SignalingMapper {
  @override
  List<String> get incomingEventNames => [
    'receiveChatRequest',    // Your incoming call event
    'requestAccepted',       // Your call accepted event
    'requestRejected',       // Your call rejected event
    'sessionStarted',        // Your session started event
    'sessionEnded',          // Your session ended event
  ];

  @override
  RtcSignalingEvent? mapIncomingEvent(String eventName, Map<String, dynamic> data) {
    switch (eventName) {
      case 'receiveChatRequest':
        return RtcSignalingEvent.callOffer(
          callId: data['requestId'] ?? _generateId(data),
          fromUserId: data['userId'],
          toUserId: data['listenerId'],
          callType: data['type'] == 'video' ? 'video' : 'audio',
          roomId: data['roomId'],
          metadata: {
            'senderName': data['userName'],
            'senderImage': data['userImage'],
          },
        );
      // ... handle other events
    }
    return null;
  }

  @override
  SignalingEmitData mapOutgoingEvent(RtcSignalingEvent event, {...}) {
    // Map outgoing events to your backend format
  }

  @override
  String getEventName(RtcSignalingEventType type) {
    switch (type) {
      case RtcSignalingEventType.callOffer:
        return 'chat-request';
      case RtcSignalingEventType.callAnswer:
        return 'accept-request';
      case RtcSignalingEventType.callReject:
        return 'reject-request';
      case RtcSignalingEventType.callEnd:
        return 'session-end';
      // ...
    }
  }
}

// Use custom mapper
await rtcService.initialize(
  config: config,
  socketAdapter: adapter,
  signalingMapper: MyBackendMapper(),
);

Handling Call Events

All call events are emitted through the callEvents stream, regardless of whether you're using native CallKit (useCallKit: true) or custom UI (useCallKit: false). This allows you to navigate to appropriate screens and handle UI updates.

💡 Important: With useCallKit: false, you must handle the incoming call UI yourself by listening to CallEventType.incomingCall and showing your custom screen.

CallEvent Types

rtcService.callEvents.listen((event) {
  switch (event.type) {
    case CallEventType.incomingCall:
      final callId = event.callId;
      final caller = event.remoteUser;
      print('Incoming call from ${caller?.userName}');

      // If useCallKit = false, show your custom incoming call screen here
      if (!usingCallKit) {
        showCustomIncomingCallScreen(event.callData);
      }
      break;

    case CallEventType.outgoingCall:
      print('Call initiated, waiting for answer...');
      break;

    case CallEventType.callAccepted:
      print('Call accepted, connecting...');
      // Navigate to call screen
      Navigator.push(context, MaterialPageRoute(
        builder: (_) => CallScreen(callId: event.callId),
      ));
      break;

    case CallEventType.callConnected:
      print('Call connected!');
      break;

    case CallEventType.callDeclined:
      print('Call was declined');
      break;

    case CallEventType.callEnded:
      print('Call ended: ${event.reason}');
      Navigator.pop(context);
      break;

    case CallEventType.callFailed:
      print('Call failed: ${event.error}');
      break;

    case CallEventType.callMissed:
      print('Missed call from ${event.remoteUser?.userName}');
      break;

    case CallEventType.remoteUserJoined:
      print('Remote user joined');
      break;

    case CallEventType.remoteUserLeft:
      print('Remote user left');
      break;
  }
});

Making and Receiving Calls

Start an Outgoing Call

try {
  await rtcService.startCall(
    calleeId: 'user_456',
    calleeName: 'Jane Smith',
    calleeImage: 'https://example.com/jane.jpg',
    callType: CallType.video,  // or CallType.audio
  );
  // Show outgoing call UI
} catch (e) {
  print('Failed to start call: $e');
}

Accept Incoming Call

// When user taps "Accept" (or handled automatically by CallKit)
await rtcService.acceptCall();

Reject Incoming Call

// When user taps "Decline"
await rtcService.rejectCall();

End Active Call

await rtcService.endCall();

In-Call Controls

// Toggle mute
await rtcService.toggleMute();
bool isMuted = rtcService.isMuted;

// Toggle camera
await rtcService.toggleCamera();
bool isCameraOn = rtcService.isCameraOn;

// Switch camera (front/back)
await rtcService.switchCamera();

// Toggle speaker
await rtcService.toggleSpeaker();

Video Widget

Display local and remote video:

// In your call screen widget
@override
Widget build(BuildContext context) {
  return Stack(
    children: [
      // Remote video (full screen)
      Positioned.fill(
        child: RtcVideoView(
          userId: remoteUserId,
          isLocal: false,
        ),
      ),

      // Local video (picture-in-picture)
      Positioned(
        top: 50,
        right: 20,
        width: 120,
        height: 160,
        child: ClipRRect(
          borderRadius: BorderRadius.circular(12),
          child: RtcVideoView(
            userId: localUserId,
            isLocal: true,
          ),
        ),
      ),
    ],
  );
}

In-Session Chat

The package includes a built-in chat system for sending text messages during active sessions. Chat works over the same socket connection used for call signaling.

Chat Features

  • Send and receive text messages
  • Typing indicators (is typing / stopped typing)
  • Message status tracking: sending → sent → delivered → read
  • Message deduplication (prevents duplicate processing)
  • Session-aware messaging (tied to call sessions)

Listening to Chat Events

rtcService.chatEvents.listen((event) {
  switch (event.type) {
    case ChatEventType.messageReceived:
      // New message from remote user
      final message = event.message!;
      print('${message.senderName}: ${message.content}');
      break;

    case ChatEventType.messageSent:
      // Message successfully sent
      print('Sent: ${event.message?.content}');
      break;

    case ChatEventType.messageDelivered:
      // Message delivered to recipient
      final messageId = event.metadata?['messageId'];
      print('Delivered: $messageId');
      break;

    case ChatEventType.messageRead:
      // Message read by recipient
      final messageIds = event.metadata?['messageIds'];
      print('Read: $messageIds');
      break;

    case ChatEventType.typingIndicator:
      // Remote user is typing
      final isTyping = event.metadata?['isTyping'] as bool;
      final userName = event.user?.userName;
      print('$userName is ${isTyping ? "typing..." : "idle"}');
      break;

    case ChatEventType.sessionStarted:
      print('Chat session started: ${event.sessionId}');
      break;

    case ChatEventType.sessionEnded:
      print('Chat session ended: ${event.sessionId}');
      break;
  }
});

Sending Messages

// Send a text message
final result = await rtcService.sendMessage(
  sessionId: 'session_123',
  content: 'Hello! How are you?',
);

if (result.success) {
  print('Message sent: ${result.message?.messageId}');
} else {
  print('Failed: ${result.error}');
}

Typing Indicators

// Notify that the user started typing
await rtcService.sendTypingIndicator(
  sessionId: 'session_123',
  isTyping: true,
);

// Notify that the user stopped typing
await rtcService.sendTypingIndicator(
  sessionId: 'session_123',
  isTyping: false,
);

Session Management

// Set the active chat session (auto-set when a call session starts)
rtcService.setCurrentChatSession('session_123');

// Get the current chat session ID
final sessionId = rtcService.currentChatSessionId;

Chat Socket Events

The chat system uses these socket events by default:

Direction Socket Event Description
Incoming receiveMessage New message from remote user
Incoming messageDelivered Delivery confirmation
Incoming messageRead Read receipt
Incoming typing Typing indicator
Outgoing sendMessage Send a message
Outgoing typing Send typing indicator

ChatEvent Types

Type Description
sessionStarted Chat session started
sessionEnded Chat session ended
messageReceived New message received
messageSent Message successfully sent
messageDelivered Message delivered to server
messageRead Message read by recipient
messageFailed Message failed to send
typingIndicator Remote user typing status

MessageStatus

Status Description
sending Message is being sent
sent Message sent to server
delivered Message delivered to recipient
read Message read by recipient
failed Message failed to send

📖 Widget: The package ships a ready-made chat UI in lib/src/widgets/chat_screen.dart — message bubbles, typing indicators and status icons included.

Platform Setup

call_kit_for_zego is a pure-Dart package — it has no android/ or ios/ folder of its own. Everything below is configuration your app owns. The example app under example/ is a working reference for all of it.

Android

Permissions are mostly automatic. flutter_callkit_incoming ships its own AndroidManifest.xml, and the Gradle manifest merger folds it into your app. You get these without doing anything:

INTERNET · RECORD_AUDIO · CAMERA · POST_NOTIFICATIONS · MANAGE_OWN_CALLS · USE_FULL_SCREEN_INTENT · FOREGROUND_SERVICE · FOREGROUND_SERVICE_PHONE_CALL · FOREGROUND_SERVICE_CAMERA · FOREGROUND_SERVICE_MICROPHONE · WAKE_LOCK · VIBRATE · DISABLE_KEYGUARD · TURN_SCREEN_ON · ACCESS_NOTIFICATION_POLICY

Add to android/app/src/main/AndroidManifest.xml only what the merger does not give you:

<manifest xmlns:android="http://schemas.android.com/apk/res/android">
    <!-- Not in the plugin manifest; needed for the speaker/earpiece toggle -->
    <uses-permission android:name="android.permission.MODIFY_AUDIO_SETTINGS" />

    <!-- Keeps audio-only devices eligible on the Play Store -->
    <uses-feature android:name="android.hardware.camera" android:required="false" />
    <uses-feature android:name="android.hardware.microphone" android:required="false" />

    <application>
        <activity
            android:name=".MainActivity"
            android:launchMode="singleTop"
            android:showWhenLocked="true"
            android:turnScreenOn="true">
        </activity>
    </application>
</manifest>

⚠️ Delete android:taskAffinity="" from MainActivity

flutter create puts it there by default, and it breaks accepting a call. flutter_callkit_incoming launches your MainActivity with FLAG_ACTIVITY_NEW_TASK when the user accepts. With an empty task affinity Android cannot reuse the running task, so it starts a second MainActivity in a second task: a brand-new route stack that shows a black screen while the real call screen sits in the original task, plus a duplicate entry in the recents switcher.

Removing the attribute lets the activity share the app's default affinity, so the running task is brought forward instead.

In android/app/build.gradle.kts:

android {
    defaultConfig {
        // The full-screen incoming-call UI needs API 23+
        minSdk = maxOf(flutter.minSdkVersion, 23)
    }
}

Required Gradle workaround. zego_express_engine 3.25.0 still declares compileSdkVersion 31, while its own AndroidX dependencies require 33+. Without this, any app depending on it fails to build with Dependency 'androidx.window.extensions.core:core:1.0.0' requires ... version 33 or later. Add to android/build.gradle.kts, above the existing subprojects { project.evaluationDependsOn(":app") } block:

subprojects {
    afterEvaluate {
        extensions.findByName("android")?.let { ext ->
            val android = ext as com.android.build.gradle.BaseExtension
            if (android.compileSdkVersion?.substringAfter("android-")?.toIntOrNull()
                    ?.let { it < 35 } == true) {
                android.compileSdkVersion(35)
            }
        }
    }
}

Order matters — placed after evaluationDependsOn, Gradle fails with Cannot run Project.afterEvaluate(Action) when the project is already evaluated.

iOS

Add to ios/Runner/Info.plist:

<key>NSMicrophoneUsageDescription</key>
<string>Microphone access is required to make and receive calls.</string>

<key>NSCameraUsageDescription</key>
<string>Camera access is required for video calls.</string>

<key>UIBackgroundModes</key>
<array>
    <string>audio</string>
    <string>voip</string>
    <string>remote-notification</string>
    <string>fetch</string>
</array>

Then in Xcode (ios/Runner.xcworkspace → Signing & Capabilities), add:

  • Push Notifications
  • Background Modes → Audio/AirPlay/PiP, Voice over IP, Background fetch, Remote notifications

⚠️ Runtime permissions are your app's job

This is the single most common cause of "the call connects but there is no audio." Declaring RECORD_AUDIO in the manifest is not the same as being granted it. Android 6+ requires a runtime grant, and neither ZEGOCLOUD nor flutter_callkit_incoming requests microphone or camera access for you.

If the permission is missing, the ZEGO engine still joins the room and still publishes a stream — a silent one. Both sides show a connected call and nothing in the logs looks wrong.

Request the permissions before calling startCall() or acceptCall(). Any permission plugin works; permission_handler is the usual choice — it is deliberately not a dependency of this package, so you are free to use whatever your app already has:

import 'package:permission_handler/permission_handler.dart';

Future<bool> ensureCallPermissions({required bool isVideo}) async {
  final needed = <Permission>[
    Permission.microphone,
    if (isVideo) Permission.camera,
    if (Platform.isAndroid) Permission.notification, // Android 13+
  ];

  final results = await needed.request();
  return results.values.every((s) => s.isGranted);
}

// Before starting or accepting a call:
if (!await ensureCallPermissions(isVideo: callType == CallType.video)) {
  // Show your own "permissions required" UI — do not start the call.
  return;
}
await rtcService.startCall(/* ... */);

On Android, also request the full-screen-intent permission once at startup so incoming calls can appear over the lock screen:

if (Platform.isAndroid) {
  await FlutterCallkitIncoming.requestFullIntentPermission();
}

The package verifies, even though it cannot request. If the engine reports that a device was never authorized, call_kit_for_zego surfaces a CallError.permissionDenied on its error stream rather than letting the call fail silently. Listen for it:

rtcService.errors.listen((error) {
  if (error.code == CallError.codePermissionDenied) {
    // Mic or camera was never granted, or was revoked mid-call.
  }
});

This also fires for a mic that is unavailable for other reasons — occupied by another app, or taken over by Siri mid-call.

Plugins

Analytics Plugin

Track call metrics with your analytics service:

class FirebaseAnalyticsPlugin implements AnalyticsPlugin {
  final FirebaseAnalytics _analytics = FirebaseAnalytics.instance;

  @override
  Future<void> logCallStarted({
    required String callId,
    required String callType,
    required String callerId,
    required String calleeId,
    bool isOutgoing = true,
  }) async {
    await _analytics.logEvent(
      name: 'call_started',
      parameters: {
        'call_id': callId,
        'call_type': callType,
        'is_outgoing': isOutgoing,
      },
    );
  }

  // Implement other methods...
}

// Register plugin
rtcService.registerPlugin(FirebaseAnalyticsPlugin());

Crash Reporting Plugin

Report call errors to your crash reporting service:

class FirebaseCrashlyticsPlugin implements CrashReportingPlugin {
  final FirebaseCrashlytics _crashlytics = FirebaseCrashlytics.instance;

  @override
  Future<void> recordError(
    dynamic error,
    StackTrace? stackTrace, {
    String? reason,
    bool fatal = false,
  }) async {
    await _crashlytics.recordError(
      error,
      stackTrace,
      reason: reason,
      fatal: fatal,
    );
  }

  // Implement other methods...
}

rtcService.registerPlugin(FirebaseCrashlyticsPlugin());

Call Storage Plugin

Persist call history:

class HiveCallStoragePlugin implements CallStoragePlugin {
  final Box<CallRecord> _callBox;

  @override
  Future<void> saveCallToHistory(CallRecord record) async {
    await _callBox.put(record.callId, record);
  }

  // Implement other methods...
}

rtcService.registerPlugin(HiveCallStoragePlugin(callBox));

Advanced Usage

Custom Call State Handling

rtcService.onStateChanged = (CallState oldState, CallState newState) {
  print('State changed: $oldState → $newState');

  // Custom handling based on state
  if (newState == CallState.connecting) {
    showConnectingOverlay();
  } else if (newState == CallState.connected) {
    hideConnectingOverlay();
    startCallTimer();
  }
};

Network Reconnection

The package automatically handles network reconnection. You can customize behavior:

rtcService.onReconnecting = () {
  showReconnectingBanner();
};

rtcService.onReconnected = () {
  hideReconnectingBanner();
};

rtcService.onReconnectFailed = () {
  showReconnectFailedDialog();
};

Call Duration Timer

// Access current call duration
final duration = rtcService.currentCallDuration;
print('Call duration: ${duration.inMinutes}:${duration.inSeconds % 60}');

// Or listen to duration updates
rtcService.durationStream.listen((duration) {
  updateDurationUI(duration);
});

Troubleshooting

Common Issues

1. Incoming call not showing on locked screen (Android)

  • Ensure USE_FULL_SCREEN_INTENT permission is declared
  • Call FlutterCallkitIncoming.requestFullIntentPermission() at app start
  • Check that showOnLockScreen and turnScreenOn are set in Activity

2. No audio after call connects

  • Check microphone permission is granted
  • Ensure ZEGO credentials are correct
  • Verify both users joined the same roomId

3. Call state stuck

  • Enable debug logging: RtcLogger.setLogLevel(LogLevel.debug)
  • Check socket connection status
  • Verify signaling events are being received

4. CallKit not dismissing when call is cancelled

  • Ensure endCall() is called with correct callId
  • Check that socket events are properly mapped

5. Want to test without native CallKit complexity

  • Set useCallKit: false in CommunicationConfig
  • Implement custom incoming call UI using onIncomingCall callback
  • Call rtcService.acceptCall() or rtcService.rejectCall() from your UI
  • See CallKit Options for complete example

6. Incoming call not showing (useCallKit: false)

  • Verify you're listening to rtcService.callEvents stream
  • Check that you show your custom UI in CallEventType.incomingCall handler
  • Ensure your incoming call screen calls acceptCall() or rejectCall()

Debug Logging

// Enable verbose logging
RtcLogger.setLogLevel(LogLevel.debug);

// Logs will show:
// [RtcCallKit] INFO: Initializing...
// [RtcCallKit] DEBUG: Socket event received: incoming_call
// [RtcCallKit] INFO: State transition: idle → ringing

API Reference

RtcCommunicationService

Method Description
initialize() Initialize the service with config
setCurrentUser() Set the current user
startCall() Initiate an outgoing call
acceptCall() Accept incoming call
rejectCall() Reject incoming call
endCall() End active call
toggleMute() Toggle microphone
toggleCamera() Toggle camera
switchCamera() Switch front/back camera
sendMessage() Send a chat message
sendTypingIndicator() Send typing indicator
setCurrentChatSession() Set active chat session
dispose() Clean up resources

CallEvent Types

Type Description
incomingCall Received incoming call
outgoingCall Outgoing call initiated
callAccepted Call was accepted
callConnected Both parties connected
callDeclined Call was declined
callEnded Call ended normally
callFailed Call failed with error
callMissed Call was missed (timeout)
remoteUserJoined Remote user joined room
remoteUserLeft Remote user left room

CallState

State Description
idle No active call
initiating Starting outgoing call
ringing Waiting for answer
connecting Call accepted, connecting
connected Call active
reconnecting Reconnecting after network loss
ended Call ended
failed Call failed

License

MIT License - see LICENSE file for details.

Contributing

Contributions are welcome — please open an issue before starting significant work.

Support

CallKit usage

Overview

The useCallKit flag allows you to control whether native CallKit UI is used or if your app handles the call UI manually.

Configuration

Option 1: Use Native CallKit (Default)

When useCallKit = true (default), the native CallKit UI handles incoming calls with full-screen notification even on locked screens.

await RtcCommunicationService.initialize(
  config: CommunicationConfig(
    zegoConfig: ZegoConfig(
      appId: YOUR_ZEGO_APP_ID,
      appSign: 'YOUR_ZEGO_APP_SIGN',
    ),
    socketConfig: SocketConfig(url: 'https://your-server.com'),
    callKitConfig: CallKitConfig(
      appName: 'My App',
      iconName: 'app_icon',
    ),
    useCallKit: true, // ✅ Native CallKit handles UI
  ),
  socketAdapter: socketAdapter,
);

What happens:

  • ✅ Native full-screen incoming call UI on iOS/Android
  • ✅ Works on locked screen
  • ✅ System ringtone plays
  • ✅ User accepts/declines via native UI
  • ✅ Your callbacks (onIncomingCall, onCallAccepted, etc.) still fire for navigation

Option 2: Custom App UI

When useCallKit = false, your app handles all call UI using callbacks.

await RtcCommunicationService.initialize(
  config: CommunicationConfig(
    zegoConfig: ZegoConfig(
      appId: YOUR_ZEGO_APP_ID,
      appSign: 'YOUR_ZEGO_APP_SIGN',
    ),
    socketConfig: SocketConfig(url: 'https://your-server.com'),
    callKitConfig: CallKitConfig(  // Optional: still provide config for reference
      appName: 'My App',
    ),
    useCallKit: false, // ❌ CallKit disabled - app handles UI
  ),
  socketAdapter: socketAdapter,
);

What happens:

  • ❌ No native CallKit UI
  • ✅ onIncomingCall callback fires
  • ✅ You show your custom incoming call screen
  • ✅ User accepts/declines via your UI
  • ✅ You call rtcService.acceptCall() or rtcService.rejectCall()

Implementation Example

With Custom UI (useCallKit = false)

class MyApp extends StatefulWidget {
  @override
  _MyAppState createState() => _MyAppState();
}

class _MyAppState extends State<MyApp> {
  late RtcCommunicationService rtcService;

  @override
  void initState() {
    super.initState();
    _initializeRTC();
  }

  Future<void> _initializeRTC() async {
    await RtcCommunicationService.initialize(
      config: CommunicationConfig(
        zegoConfig: ZegoConfig(
          appId: YOUR_APP_ID,
          appSign: 'YOUR_APP_SIGN',
        ),
        socketConfig: SocketConfig(url: 'https://your-server.com'),
        useCallKit: false, // Custom UI mode
      ),
      socketAdapter: socketAdapter,
    );

    rtcService = RtcCommunicationService.instance;

    // Listen to call events
    rtcService.callEvents.listen((event) {
      switch (event.type) {
        case CallEventType.incomingCall:
          // Show YOUR custom incoming call screen
          Navigator.push(
            context,
            MaterialPageRoute(
              builder: (_) => CustomIncomingCallScreen(
                callData: event.callData,
                onAccept: () async {
                  await rtcService.acceptCall();
                  Navigator.pushReplacement(
                    context,
                    MaterialPageRoute(builder: (_) => CallScreen()),
                  );
                },
                onReject: () async {
                  await rtcService.rejectCall();
                  Navigator.pop(context);
                },
              ),
            ),
          );
          break;

        case CallEventType.callEnded:
          Navigator.popUntil(context, (route) => route.isFirst);
          break;

        // Handle other events...
      }
    });
  }
}

// Your custom incoming call screen
class CustomIncomingCallScreen extends StatelessWidget {
  final CallData callData;
  final VoidCallback onAccept;
  final VoidCallback onReject;

  const CustomIncomingCallScreen({
    required this.callData,
    required this.onAccept,
    required this.onReject,
  });

  @override
  Widget build(BuildContext context) {
    return Scaffold(
      backgroundColor: Colors.black87,
      body: Center(
        child: Column(
          mainAxisAlignment: MainAxisAlignment.center,
          children: [
            CircleAvatar(
              radius: 60,
              backgroundImage: NetworkImage(callData.caller.avatarUrl ?? ''),
            ),
            SizedBox(height: 24),
            Text(
              '${callData.caller.userName} is calling...',
              style: TextStyle(color: Colors.white, fontSize: 24),
            ),
            SizedBox(height: 48),
            Row(
              mainAxisAlignment: MainAxisAlignment.spaceEvenly,
              children: [
                // Reject button
                IconButton(
                  icon: Icon(Icons.call_end, size: 40),
                  color: Colors.red,
                  onPressed: onReject,
                ),
                // Accept button
                IconButton(
                  icon: Icon(Icons.call, size: 40),
                  color: Colors.green,
                  onPressed: onAccept,
                ),
              ],
            ),
          ],
        ),
      ),
    );
  }
}

Comparison

Feature useCallKit = true useCallKit = false
Native full-screen UI ✅ Yes ❌ No
Works on locked screen ✅ Yes ❌ No (requires push notification)
System ringtone ✅ Yes ❌ No (you handle audio)
Custom UI design ❌ Limited ✅ Full control
iOS CallKit integration ✅ Yes ❌ No
Android ConnectionService ✅ Yes ❌ No
Implementation complexity ⭐ Easy ⭐⭐⭐ More work

Use Cases

When to use useCallKit = true (Native UI)

  • ✅ You want WhatsApp/Telegram-like native call experience
  • ✅ Calls must work on locked screens
  • ✅ You want system-level call integration
  • ✅ Less implementation work

When to use useCallKit = false (Custom UI)

  • ✅ You need fully custom call UI design
  • ✅ You want to show additional info on incoming call screen
  • ✅ Your app has unique UX requirements
  • ✅ You're okay with more implementation work
  • ✅ Testing/debugging without native UI complexity

Migration

If you have existing code using CallKit, nothing changes by default:

// Old code - still works the same
await RtcCommunicationService.initialize(
  config: CommunicationConfig(
    callKitConfig: CallKitConfig(appName: 'My App'),
    // useCallKit defaults to true
  ),
);

To switch to custom UI:

// New code - custom UI mode
await RtcCommunicationService.initialize(
  config: CommunicationConfig(
    callKitConfig: CallKitConfig(appName: 'My App'),
    useCallKit: false, // 🔄 Add this line
  ),
);

// Then implement your custom incoming call UI
rtcService.callEvents.listen((event) {
  if (event.type == CallEventType.incomingCall) {
    showYourCustomIncomingCallScreen();
  }
});

Important Notes

  1. Callbacks always fire: Regardless of useCallKit setting, all callbacks (onIncomingCall, onCallAccepted, onCallEnded, etc.) always fire. This allows you to navigate to the appropriate screens.

  2. Background calls: When useCallKit = false, incoming calls won't work on locked screens without additional push notification setup.

  3. Testing: Set useCallKit = false during development to test without native UI complexity.

  4. Platform support:

    • iOS: Native CallKit fully supported
    • Android: Native ConnectionService fully supported
    • Both work with useCallKit = false

Questions?

If you have questions, check the main README.md or open an issue.

CallKit flag history

Overview

Added useCallKit boolean flag to make native CallKit UI optional. When disabled, the app handles call UI using callbacks.

Changes Made

1. RtcCallController (lib/src/controllers/rtc_call_controller.dart)

Added Fields

final bool useCallKit; // Flag to enable/disable CallKit (line 47)

Constructor Changes

RtcCallController({
  // ... existing parameters
  this.useCallKit = true, // NEW: Default to true (backward compatible)
  // ...
})

Initialization Logic

  • Only initializes CallKit when BOTH conditions are true:
    • nativeCallKitConfig != null
    • useCallKit == true

Updated Methods

All methods that interact with CallKit now check the flag:

  • init() - Only initializes CallKit if useCallKit == true
  • receiveIncomingCall() - Only shows CallKit UI if useCallKit == true
    • Important: onIncomingCall callback ALWAYS fires regardless of flag
  • rejectCall() - Only ends CallKit if useCallKit == true
  • endCall() - Only ends CallKit if useCallKit == true
  • markCallConnected() - Only marks CallKit connected if useCallKit == true
  • _handleCallAnswerEvent() - Only marks CallKit connected if useCallKit == true
  • _handleCallRejectEvent() - Only ends CallKit if useCallKit == true
  • _handleCallCancelEvent() - Only ends CallKit if useCallKit == true
  • dispose() - Only disposes CallKit if useCallKit == true

2. CommunicationConfig (lib/src/api/models/communication_config.dart)

Added Field

/// Enable native CallKit UI
/// When true: Native CallKit handles incoming call UI
/// When false: App handles UI using callbacks (onIncomingCall, etc.)
/// Default: true (use CallKit if config is provided)
final bool useCallKit;

Constructor Changes

const CommunicationConfig({
  // ... existing parameters
  this.useCallKit = true, // NEW: Default to true
  // ...
});

3. RtcCommunicationService (lib/src/api/rtc_communication_service.dart)

Updated Initialization

_rtcController = RtcCallController(
  // ... existing parameters
  useCallKit: _config.useCallKit, // NEW: Pass through from config
  // ...
);

4. Bug Fixes

Fixed Syntax Error (line 365)

// Before
} else{

// After
} else {

5. Documentation

Documented in the CallKit usage section of this README.

  • Comprehensive guide on using the new flag
  • Code examples for both modes
  • Comparison table
  • Migration guide
  • Use case recommendations

Backward Compatibility

✅ 100% Backward Compatible

Default value is true, so existing code continues to work exactly as before:

// Existing code - no changes needed
await RtcCommunicationService.initialize(
  config: CommunicationConfig(
    callKitConfig: CallKitConfig(appName: 'My App'),
    // useCallKit defaults to true
  ),
);

Usage Examples

Native CallKit (Default)

CommunicationConfig(
  callKitConfig: CallKitConfig(appName: 'My App'),
  useCallKit: true, // or omit - defaults to true
)

Custom App UI

CommunicationConfig(
  callKitConfig: CallKitConfig(appName: 'My App'),
  useCallKit: false, // CallKit disabled
)

// Handle incoming calls in your app
rtcService.callEvents.listen((event) {
  if (event.type == CallEventType.incomingCall) {
    showMyCustomIncomingCallScreen();
  }
});

Benefits

  1. Flexibility: Developers can choose native or custom UI
  2. Testing: Easier to test without native UI complexity
  3. Custom UX: Full control over call UI design
  4. Backward Compatible: No breaking changes
  5. Clean Architecture: Flag is passed through all layers

Testing Checklist

  • x CallKit works when useCallKit = true
  • x CallKit is bypassed when useCallKit = false
  • x Callbacks fire in both modes
  • x No null pointer errors
  • x Backward compatibility maintained
  • Integration tests with real ZEGO calls
  • Test on iOS and Android

Next Steps

  1. Test with real calls on iOS/Android
  2. Update main README.md with useCallKit documentation
  3. Add unit tests for flag behavior
  4. Consider adding to example app

Files Modified

  1. lib/src/controllers/rtc_call_controller.dart
  2. lib/src/api/models/communication_config.dart
  3. lib/src/api/rtc_communication_service.dart
  4. The CallKit usage section of this README
  5. CHANGELOG_CALLKIT_FLAG.md (this file)

Libraries

call_kit_for_zego
RTC Call Kit - A comprehensive Flutter package for real-time audio/video calling