Tether

A ready-to-use template for building AI/ML desktop applications.

Stack: React + Python (FastAPI) + Tauri

Core Principles

1. Frontend devs focus on React - No Rust or Python knowledge required
2. ML devs focus on Python - No Rust or frontend knowledge required
3. Rust is invisible - Works correctly in the background
4. Easy distribution - One-click installers for end users

Quick Start

npx create-tether-app my-app
cd my-app
pnpm dev

Run npx create-tether-app --help for options (LLM backend, Tailwind, etc.).

Project Structure

my-app/
├── frontend/          # React 18 + TypeScript + Vite
├── backend/           # Python 3.11+ + FastAPI + uvicorn
└── src-tauri/         # Tauri 2.x (Rust shell)

Technology Stack

Layer	Technology	Purpose
Desktop	Tauri 2.x	Lightweight, secure shell
Frontend	React 18 + Vite	Fast, modern UI development
Backend	FastAPI	Async Python API server
ML	Ollama / llama-cpp-python	Local LLM inference
Package Manager	uv	Fast Python dependency management

Documentation

• Getting Started - First steps
• Frontend Guide - React/TypeScript development
• Backend Guide - Python/FastAPI development
• Deployment - Building installers
• Development - Build optimization tips

What's Included

• Chat UI component with message history
• LLM service abstraction (Ollama, OpenAI, local models)
• Health check and model switching endpoints
• Image/vision support for multimodal models
• Thinking mode for reasoning models
• Automatic port allocation and process management

License

MIT License - see LICENSE for details.