Even with a normal standalone Python environment, you need for the user to install packages that your code will depend on. Your scripts can install the packages using either Python or JSL. See the scripting Index for documentation of JSL's Python Install Packages( ), and import jmputils's jpip()
You can wrap the import statement with a try: except: block to catch the error if an import fails, and in the except: block do the jmputils.jpip() install of the package. But remember at this point Python already tried, so in the except block after the install you will also need importlib and make a reload() on of the import module in question.
I believe the best way is to create a real Python package for your Python code, complete with a project.toml file that sets up all the dependencies so that you can then do a jpip('install', 'your_package') Host that on PyPi, your own repository or just a path in your filesystem to the package. See regular Python packaging guidelines on how to create Python packages. Then pip, via jpip, takes care of pulling in all of the dependencies.