🌌 Entangle

High-Performance, Low-Latency Remote Desktop Protocol.

Built with Rust (2026 Edition), Tauri v2, and QUIC. Designed for engineers and power users who need AnyDesk/TeamViewer performance with an open-source, secure-by-default architecture.

---

🎯 Core Objective

Build a lightweight, secure, low-latency alternative to commercial remote support tools. Entangle prioritizes Latency (RTT) and Frame Stability over perfect image quality, making it ideal for technical support and remote work.

🧱 Project Structure

A Rust workspace with strict separation of concerns:

• apps/desktop-client: Tauri v2 App (Rust + React/Vite/Shadcn).
• apps/signaling-server: Axum-based WebSocket Signaling & Discovery.
• crates/capture: Platform-native screen capture (SCKit for Mac, DXGI for Win).
• crates/encoder: Video encoding abstraction (H.264/H.265, Hardware accelerated).
• crates/net-transport: QUIC networking (Unreliable Datagrams for video, Reliable Streams for input).
• crates/input-injector: OS-level input simulation.
• crates/crypto-session: E2EE (X25519 + ChaCha20Poly1305).
• crates/shared-protocol: Shared packet definitions and enums.

🛠 Tech Stack

• Core: Rust 1.83+ (2026 Edition)
• Networking: QUIC (quinn), WebSockets (axum)
• Desktop: Tauri v2
• Frontend: React + Vite + TypeScript
• UI: Shadcn UI + Tailwind CSS
• Media: WebCodecs API (Frontend), ScreenCaptureKit/DXGI (Backend)

---

🚦 Current Status & Roadmap (Product-Ready Path)

✅ Completed

• High-level Workspace Architecture
• Shared Protocol Definitions
• Basic Signaling Server (WebSocket)
• Frontend Scaffolding (React + Shadcn UI)
• WebCodecs Video Decoder Hook
• Tauri Command Structure

🚧 In Progress

• Native Capture: Migrating from mock frames to ScreenCaptureKit (macOS) and DXGI (Windows).
• QUIC Transport: Replacing mock connection logic with quinn P2P datagrams.
• Hardware Encoding: Implementing VideoToolbox (macOS) and NVENC wrappers.

📅 Next Steps

• NAT Traversal: ICE/STUN integration for P2P connectivity.
• Dirty Rect Detection: Intelligent frame delta encoding to save bandwidth.
• Input Injection: Mapping frontend mouse/kb events to CGEvent (macOS) and SendInput (Windows).

---

🚀 Getting Started

Prerequisites

• Rust (1.83+ is recommended for 2026 Edition support)
• Node.js & pnpm
• macOS (Xcode installed) or Windows (Visual Studio C++ Build Tools)

Development

1. Start Signaling Server:

   cargo run -p signaling-server

2. Launch Desktop Client:

   pnpm install
   cargo tauri dev

---

⚠️ Architectural Constraints

1. Transport Layer: Use QUIC Datagrams for video (unreliable) and Streams for input/control (reliable).
2. Rendering: Do NOT decode on the CPU. Use the browser's WebCodecs API in the React frontend.
3. Capture: Always use Dirty Rect detection. Only encode pixels that actually changed.

---

📄 License

MIT © 2026 Parevo