// the find
stitchfix/pyxley
Python helpers for building dashboards using Flask and React
Pyxley is a Stitch Fix internal tool they open-sourced for building Flask-backed dashboards using React, targeting data scientists who want interactive charts without writing JavaScript. It wraps charting libraries (MetricsGraphics, NVD3, Plotly, DataTables) in Python classes and generates the React mount code via Jinja templates.
The pandas-first design is the right call for the audience — data scientists live in DataFrames and the library handles the JSON serialization differences between charting libraries. Supporting multiple chart backends (MetricsGraphics, NVD3, Plotly, DataTables) from a single Python API is genuinely useful. The Flask integration is clean: UILayout handles route registration so you don't wire up endpoints by hand. Examples for each chart type are included and runnable.
This is abandoned internal tooling from around 2016. It uses React.render() (the pre-0.14 API that was deprecated a decade ago) — any attempt to run this today means pinning ancient React versions or rewriting the JSX layer. No tests: the README says 'Coming Soon' and tests/README.md exists but the suite is essentially empty. The Jinja-templating approach to generate React mount code is a clever hack that became a dead end — modern React (hooks, concurrent mode, server components) simply doesn't work this way. There's a companion repo (pyxleyJS) that also appears unmaintained. Adopt this and you own it.