Skip to main content
The ChatHistoryProvider interface connects any LLM to AgentChatView: 
abstract class ChatHistoryProvider implements Listenable {
  Stream<String> transcribeAudio(XFile audioFile);
  Stream<String> sendMessageStream(String prompt, {Iterable<Part> attachments});
  Iterable<ChatMessage> get history;
  set history(Iterable<ChatMessage> history);
}
Built-in providers:
  • DartanticProvider: Wraps dartantic_ai for multi-provider support
  • EchoProvider: Minimal example, useful for testing

​
Implementation Guide

To build your own provider:
  1. Configuration: Allow model creation as a parameter
  2. History: Manage history, notify listeners, support serialization
  3. Messages: Map ChatMessage and Part types to your LLM format
  4. LLM calls: Implement sendMessageStream and transcribeAudio
Example structure: 
class MyProvider extends ChatHistoryProvider with ChangeNotifier {
  MyProvider({
    required MyLlmModel model,
    Iterable<ChatMessage>? history,
  }) : _model = model,
       _history = history?.toList() ?? [];

  final MyLlmModel _model;
  final List<ChatMessage> _history;

  @override
  Stream<String> sendMessageStream(
    String prompt, {
    Iterable<Part> attachments = const [],
  }) async* {
    final userMessage = ChatMessage.user(prompt);
    _history.add(userMessage);

    // Call your LLM here and stream the response
    final response = await _model.generate(prompt);

    _history.add(ChatMessage.model(response));
    yield response;
    notifyListeners();
  }

  @override
  Iterable<ChatMessage> get history => _history;

  @override
  set history(Iterable<ChatMessage> history) {
    _history.clear();
    _history.addAll(history);
    notifyListeners();
  }

  @override
  Stream<String> transcribeAudio(XFile audioFile) async* {
    // Implement audio transcription
    yield 'Transcribed text';
  }
}
See the EchoProvider implementation for a complete example.