Skip to content

marimo-team/marimo-studio

unversioned · b3babd97ff1bApache-2.0

Author and validate focused views from Marimo notebooks.

GitHub

Marimo Studio builds custom reports, apps, and presentations from one marimo notebook. Keep the analysis in Python, then shape each view with HTML, React, Svelte, or the browser libraries your work needs.

Experimental: Marimo Studio is changing rapidly.

Examples

One notebook supplies the data and calculations for three views, each with its own frontend and interaction model.

Notebook

marimo · Python · Polars

Load the Rio roster, derive age and medal counts, and inspect the data used by every view.

Marimo notebook loading and transforming the Rio athlete records

Publication report

Vanilla HTML

Summarize 11,538 athletes, 207 delegations, 28 sports, and 1,857 medalists.

Athlete report showing totals for athletes, delegations, sports, and medalists

Linked explorer

Svelte · Mosaic

Filter by sport or sex, search by name, and brush charts to update the roster, distributions, and totals together.

Athlete explorer with roster filters and a linked height and weight plot

Interactive briefing

Vanilla HTML · Shower · Three.js

Move through the roster, sports, medalists, and body profiles in a four-chapter presentation built from one point per athlete.

Athlete briefing showing the Olympic roster as an interactive point field

Explore the notebook and every live view.

Quickstart

Open a notebook in an environment that contains Studio:

uvx --with marimo-studio marimo edit analysis.py --sandbox

uvx is uv's temporary command runner. It creates an isolated Python environment for this invocation.

The thin Studio toolbar appears immediately. Add a cell such as mo.md("## Revenue is on target"), then click Add view. Save the notebook if prompted, choose HTML document, and click Create view. The generated view places displayable notebook cells inside a frontend document.

Studio connects three surfaces:

  • Notebook for Python and reactive computation
  • Source for the view project's HTML, styles, and browser code
  • Preview for the current artifact and notebook runtime

Notebook and Preview open side by side. The toolbar's source action opens the view editor beneath Preview. Marimo's agent sidebar stays available while you edit or focus the view.

Saving Source rebuilds Preview. A failed build reports the source problem and keeps the current artifact available.

Notebook, view source, and Preview in a custom layout

For terminal-first setup with an existing saved notebook, run:

uvx marimo-studio view create dashboard --target analysis.py

Build and run

notebook
  -> named view project
  -> validated artifact
  -> Python, Browser, or Prepared runtime
  -> rendered view

One notebook can publish several named views, each with its own source, build, and runtime. Choose Python for server-backed views, Browser for views that run with Pyodide in the browser, or Prepared for static publishing from precomputed results. Prepared exports keep the Python notebook source on the build machine.

Learn and operate

License

Marimo Studio is licensed under the Apache License 2.0.