Skip to content

The Nexis ecosystem

Nexis is more than one app — it’s a developer ecosystem. The terminal is the flagship, but around it has grown a family of tools that share a stack (Tauri 2 + Rust + React 19), a design language, and a philosophy: local-first, lightweight, open source, and zero telemetry.

Project What it is Stack
Nexis The AI-native terminal & developer environment (ADE) — the flagship. Tauri 2 · Rust · React 19
nexis-ml Python ML engine — train small models on small data, metrics streamed live into Nexis. Python · PyTorch
nexis-ml-rs Python-free, single-binary ML engine — same protocol and run store as nexis-ml, GPU via wgpu, ONNX export. Rust · burn
Nexis Benchmark Desktop app for benchmarking local AI models — ONNX/GGUF across inference backends. Tauri 2 · Rust · React 19
Dev Dashboard Git status across all your local repos in one glance, scanned natively in ~20 ms. Tauri 2 · Rust (git2 + rayon) · React 19
nexisdev.org The marketing site. Next.js 16 · shadcn/ui · Tailwind v4
This wiki The documentation you’re reading, at wiki.nexisdev.org. Astro · Starlight
  • Nexis is the hub. Its ML Lab panel spawns an engine — nexis-ml or nexis-ml-rs — to do real training work, and renders the streamed metrics live.
  • The two ML engines are interchangeable: both speak the same NDJSON protocol and write the same run store, so a run produced by one can be read by the other.
  • Nexis Benchmark closes the loop on local AI: it measures the models you’d run in Nexis (via LM Studio, Ollama, or ONNX), and it drives nexis-ml-rs as one of its real benchmark backends.
  • Dev Dashboard answers “which of my repos needs attention?” before you even open a terminal — and can drop you into one (your $TERMINAL) at the selected repo.
  • Everything visual descends from the Nexis design system: borderless custom window chrome, OKLCH light/dark themes, and the same typography and cursor set. Dev Dashboard is built directly from the design language extracted from Nexis.

Every project in the ecosystem follows the same rules — see Philosophy for the long version:

  • Local-first. Your code, your models, your data stay on your machine. AI runs on your own API keys or fully offline.
  • Lightweight. Tauri and the system webview instead of Electron; the Nexis terminal itself is under 10 MB.
  • Zero telemetry. None of the apps phone home.
  • Open source. All repos are public under github.com/rwetz, Apache-2.0 licensed.