Python library wrapping native OS webviews via PyTauri to build UIs that run in Jupyter, FastAPI, or as compiled desktop apps from one codebase.
Key Takeaways
Uses lightweight native OS webviews instead of bundling Chromium, directly targeting the Electron bloat problem for Python desktop apps.
Single codebase deploys three ways: Jupyter anywidget, browser via FastAPI WebSockets, or standalone desktop executable via PyTauri.
UI is declarative: Pydantic models placed in 7 layout positions cover toolbars, modals, toasts, and inputs.
Two-way Python-to-JavaScript event bridge ships with out-of-the-box support for AgGrid, Plotly, and TradingView.
Includes OAuth2, Redis RBAC state management, hot reloading, and 60+ CSS variables for theming.
Hacker News Comment Review
The “rendering engine” branding is misleading: commenters noted it is more accurately a web-based UI toolkit that deploys to desktop via Tauri, not a graphics/render engine.
The slim install hides a 30.8MB vendored binary (ext_mod.cpython-310-darwin.so) pulled from the pytauri project, which is the actual heavy lifter under the hood.
Notable Comments
@simonw: Identified the 30.8MB vendored PyTauri binary inside the wheel, sourced from github.com/pytauri/pytauri.
@simonw: Posted a uv run --with pywry python one-liner to try it without installing.
