Shiny for Python

Shiny for Python is an open-source framework from Posit for building interactive web applications and dashboards in pure Python. It is released under the MIT license and installs from PyPI or conda-forge. Apps are created from templates with the shiny create command and run locally with shiny run.

What It Is

Shiny for Python lets data scientists and developers build web apps without writing web-development code. Its reactive execution engine tracks dependencies between inputs and outputs, so outputs re-render only when their upstream components change, with no manual callbacks or state variables. It is built on Starlette and asyncio, and supports CSS and JavaScript customization. It works with Python packages such as pandas, plotly, seaborn and ipyleaflet. The project offers two authoring styles: Shiny Express and Shiny Core.

Workflow and Setup

Install with uv pip install shiny, then run uvx shiny create --template dashboard-tips to scaffold an app from a template. Apps are run with shiny run app.py --reload --launch-browser or through the Shiny extension. The site provides components, layouts, templates, a gallery and a browser-based playground (Shinylive). A testing API is also documented.

AI Apps and Agent Skills

The site highlights a natural-language data querying template (querychat) and a Shiny for AI section. The package also bundles Agent Skills, reference material that helps coding agents such as Claude Code and Cursor build, style, test and debug Shiny apps. They can be installed with uvx library-skills, and shiny skills list shows what is bundled.

Deployment

Apps can be hosted on your own servers or on Connect Cloud, shinyapps.io, Hugging Face, or run in the browser with Shinylive.