Hey @Tad_Ye - Server JS and Python code are evaluated in a sandbox, so there might be some limitations on the CLI aspects and downloading files locally. However, you can generally import and use custom libraries for either (still beta and under feature flags we can enable for your org currently).
For example, the requests Python library allows getting the data from an endpoint, and you can then use that back in the workflow to do whatever you need to do.
For S3 specifically, you could use an AWS SDK they release, or probably easier to just create an S3 resource to handle that step if you don't need to do something pretty custom.