Found the answare here.

First install packages.

#r "nuget:Microsoft.DotNet.Interactive.SqlServer,*-*"

Then create the kernels of SQL and Python.

I use a conda environment.

#!connect mssql --kernel-name DB "Server=server_name;Database=DB_name;Integrated Security=True;TrustServerCertificate=true;"

#!connect jupyter --kernel-name python --conda-env env_prueba2 --kernel-spec python3

And them to share data between languages in the same notebook.

Extract from SQL Server a query.

Note the first line, the data as a name to share from SQL to Python.

And now it can be use with Python.

As you can see the query result save in pandas dataframe.

The only problem that I found is that querys result must to be small. I try share 500,000 rows and crash.

By!