A mobile-first AI conversation app where multiple configured assistants can answer in the same room.

多 AI 群聊Tauri Android本地配置OpenAI 兼容接口移动端优先跨平台桌面

Hyacinth is a cross-platform AI chat application built with Rust, Tauri 2, React, and Vite. It prioritizes Android while retaining desktop builds, and lets each AI keep independent identity, endpoint, model, prompt, key, and temperature settings.

ShellTauri 2

BackendRust

FrontendReact

TargetAndroid

Repository

/home/lry/Projects/RustRepo/Hyacinth

Technical Stack

RustTauri 2ReactViteTypeScriptAndroid SDKOpenAI-compatible APIslocalStorageNative app config

01 / Multi-AI room

Each assistant has its own configuration, then selected assistants reply together.

Every AI can define name, avatar text, API key, base URL, model, system prompt, and temperature. Users can select multiple assistants for the same message, creating a group-chat workflow over OpenAI-compatible chat/completions APIs.

02 / Local persistence

Settings and conversations stay local to the device.

The Tauri backend stores assistant settings in the native application configuration directory, while chat history is kept in browser localStorage. This keeps personal API configuration and everyday conversation state close to the user.

03 / Cross-platform build path

Android is the priority, but the same codebase keeps a desktop route open.

The project includes npm build scripts for Tauri desktop and Android workflows, with Android Studio SDK, NDK, build tools, command-line tools, Rust Android targets, and Linux WebKitGTK dependencies documented in the README.

What I Built

Connects to OpenAI-compatible chat/completions endpoints rather than a single fixed provider.

Supports independent model, prompt, endpoint, API key, avatar text, and temperature per assistant.

Allows multiple selected AIs to respond in one conversation, turning single-user chat into a group workflow.

Uses an Android-first responsive layout while keeping a desktop AI management panel.

Documents Android SDK, NDK, Rust target, and Linux desktop build prerequisites.