What are the Puppet/Chef best practices when deploying new versions of an application?

Question

Chef

The question is rather about the philosophy of these 2 configuration management systems and their integration into a pipeline.
For example, I have a CI / CD process. Each release is numbered. I collect them, for example, in jenkins and upload the artifacts to the ftp server. There is a desire to be able to both deploy the latest snapshot and a specific release.
What is the correct way (as intended by the developers) to pass the numbers of these releases to Chef / Puppet so that they take them from the repository? Chef has roles. puppet has r10k/hiera. Moreover, in the case of puppet enterprise, the environment configurations are stored in git.
In the case of puppet, the last time I used AWS CodeDeploy (well, or the same can be done with jenkins via ssh), I put a file with the release number in /etc/facter, after which I pulled the puppet agent.
In the case of chef, you can do the same as for pappet and you can update the role after the build. Not that it was difficult, but it looks a bit crutch in my opinion.