Getting started ​

You need Python 3.10 through 3.14 and uv. This page works with a marimo notebook named analysis.py, which the first command creates when it does not exist yet. It creates a view named dashboard, an HTML page, and places one notebook result in it.

Open the notebook with Studio ​

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

--sandbox installs the notebook's declared dependencies with uv. It manages packages and does not restrict what notebook code can access. --watch reloads the notebook when another editor or a coding agent changes the file.

Without a notebook yet, the command creates an empty analysis.py. Type import marimo as mo in the first cell, add a cell with mo.md("Hello from marimo"), run both, and save. A view needs at least one saved cell.

Create a view ​

Click Add view in the Studio toolbar. Name the view dashboard, choose HTML document, review the files it will create, and confirm. Save the notebook first if Studio asks, so the view has a folder beside it. Once a notebook has a view, the view menu offers New view for the next one.

Notebook and Preview open side by side. The new page already shows every notebook cell that displays output.

The same view can be created from a terminal instead:

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

Look at the files ​

Studio saved the view as a view project, a folder named after the notebook and the view:

text
analysis.py
studio/analysis/
  dashboard/
    view.toml     the provider that builds this view
    index.html    the page
    AGENTS.md     conventions for coding agents

The folder is an ordinary HTML project. Edit index.html in Studio's Source pane, which Edit view source opens beneath Preview, or in your own editor. Studio watches the folder and rebuilds Preview whenever a build input, such as index.html, is saved.

Commit studio/ with the notebook. Build output goes to __marimo__/studio/analysis/, which Git ignores.

Place a notebook result ​

Add a cell with this code and run it:

python
summary = mo.md("## Revenue is on target")
summary

Open the cell's actions menu, enter sales_summary in its Name field, and save. In the saved notebook file the cell reads:

python
@app.cell
def sales_summary(mo):
    summary = mo.md("## Revenue is on target")
    summary
    return (summary,)

A page names each notebook result it shows with a host, an element that Studio fills with that result. Place the cell in index.html with a <marimo-cell> host, inside the <main id="app-shell"> element that holds the page's content:

html
<marimo-cell name="sales_summary"></marimo-cell>

Save. Preview rebuilds and shows Revenue is on target. Change the message in the notebook and run the cell again. The view updates without a rebuild, because the page reads the cell's current output from the running notebook.

If a later edit breaks the build, Preview keeps the last page that built and Source shows the error.

Run the view ​

Stop the editor with Ctrl+C, then serve the notebook as an app:

console
uvx --with marimo-studio marimo run analysis.py --sandbox

The default view opens at /. A second view named report would open at /report/.

Next steps ​