FoundationModelsChat

A modern, local AI chat application built with SwiftUI, Swift Concurrency, and The Composable Architecture (TCA), leveraging Apple Foundation Models for efficient local LLM interactions on macOS and iOS.

🚀 Features

✅ Local LLM Chat using Apple Foundation Models
✅ Streaming responses with smooth TCA-based state management
✅ Adaptive, Apple Intelligence-inspired UI with dynamic gradients and glass effects
✅ Structured local database with sharing-grdb for chat history
✅ Cross-platform macOS and iOS support with efficient window management
✅ Minimal, clear architecture optimized for extension and experimentation

🧩 Architecture

The project utilizes The Composable Architecture (TCA) for clear, testable, and modular state management. The following diagram illustrates the current reducer structure:

%%{ init : { "theme" : "default", "flowchart" : { "curve" : "monotoneY" }}}%%
graph LR
    AppReducer ---> AppDelegateReducer
    AppReducer ---> OnboardingReducer
    HomeReducer ---> DialogsListReducer
    HomeReducer -- optional --> MessageListReducer
    MessageListReducer ---> MessageListInputReducer

    AppDelegateReducer(AppDelegateReducer: 1)
    DialogsListReducer(DialogsListReducer: 1)
    MessageListInputReducer(MessageListInputReducer: 1)
    MessageListReducer(MessageListReducer: 1)
    OnboardingReducer(OnboardingReducer: 1)