PyNote

PyNote needs no install, account or server. Python runs on your own machine, inside the browser, so your code and data stay with you. I prototyped it as Streamlit apps in 2025, then built it as a standalone browser app. The notebook above is the real app, open at the tutorial for its UI components. Python starts the first time it loads.

Python off the main thread

Python runs in Pyodide, CPython compiled to WebAssembly, inside a Web Worker. The editor and the interpreter only talk through messages, so a long-running cell never freezes typing or scrolling.

UI controls written in Python

The pynote_ui module lets Python code render sliders, toggles, inputs, forms and layouts. Components are live objects. When Python sets a property, the control updates in place without re-running the cell. A small message protocol between the worker and the SolidJS interface keeps controls, charts and Python state in sync.

from pynote_ui import Slider, Text, Group

demo_slider = Slider(min=0, max=100, value=50, label="Adjust Me")
output_text = Text(content="Value: 50")

def on_change(data):
    output_text.content = f"Value: {int(data['value'])}"

demo_slider.on_update(on_change)
Group([demo_slider, output_text])

Reactive execution

In reactive mode, PyNote tracks which cells read which variables and re-runs the cells that depend on a change, as marimo does. Sequential, Hybrid and Concurrent modes give you manual control.

Nothing lost on refresh

Notebooks and settings survive refreshes and closed tabs. Undo and redo cover structural changes like adding, moving and deleting cells. Files live in an in-browser file system that Python can read, and a side panel manages them.

Tests

Vitest covers the execution queue, the dependency graph and component lifecycles. Playwright runs the app end to end: executing Python, editing, the files panel, plotting with matplotlib, undo and redo, and restoring a session after a reload. CI runs the type checks and publishes coverage reports.