What is the rationale to make this a separate command from pull? I feel as though it does essentially the same thing, but --extract may be clearer to a caller.

I'm concerned of the checkout name, because of the association with git.

I'd propose a UX like:

$ coder templates pull <name>
Warning: Binary output can mess up your terminal. Use "--extract -" to tell
Warning: coder to output it to your terminal anyway, or consider "--extract
Warning: <FOLDER>" to extract to a directory.

This output is taken verbatim from curl.

Well, one reason to make it a separate command is because I think the more common thing will be to download into an unpacked tree, so I don't want people to always have to type --extract. I'm open to different names, but I chose checkout precisely because of the similarity with version control systems.

checkout gives the appearance that it's checking out a ref if you're familiar with Git. I don't think it'd be synonymous with cloning a repository, which this does.

I think extract is better, but having two commands I think is excessively verbose for what this does. It's not a big deal to type --extract, and I'm fine to swap the default behavior to extract to a directory too.

Okay. I'll swap the behaviors and add a --archive flag.