I'm trying to figure out a good way so that snapshots can be built using a cron schedule and non-snapshots will be ignored.

Since this is intended to be used with a custom image. Is this necessary to be in upstream? It complicates the repository.

It's also going to complicate the release process. Currently, releases are built automatically when a tag is pushed (the tags are manually created and pushed by me). The releases are created as draft releases, meaning they will need to be manually updated and published to be available. Release notes are manually added. If you want to have this full automated, the pipeline need to automatically create and push a tag (as far as I know, all GitHub releases need to have tag).

Then as you mention, how does one handle non-snapshots? I've noticed now that it's possible to identify how the pipeline was started, see [1]. Either use that or create a separate workflow (that is, a new YAML file).

If you do this downstream instead, I can help you out to get the pipeline going.

[1] https://docs.github.com/en/actions/using-workflows/events-that-trigger-workflows#schedule

Good point. I am able to build it in my fork and can maintain a copy there, that sounds like a better plan.