Simple question: Why prefering the use of .`catch(expect)` instead of `expect().toThrow/toThrowError()` ?

I'm being picky here, but can't you just do `updateById(user.id, { password });`, it's only one character more than `userId` but one line shorter too :man_shrugging:

Why can't we just use/re-format the yup error? This way it would allows us to only keep one validator up-to-date as a source of thruth.

"User's password successfully reset" ?

Can't we assume that if the process isn't confirmed, then the cli command has failed? (thus `process.exit(1)`)?

Just a question here: Why did you make the choice to error/exit the process if one of the information is missing instead of just asking it to complete the command?

According to the last comment, it might be interesting to create a folder structure to match the scopes of the commands (eg: admin/user/reset-password)

Suggestion: Not sure about the name though. We don't mention the term `user` anywhere. `admin` can refer to many things, in the future `admin:reset-password` could be something else, no? What about `admin:user:reset-password` or `admin:user:reset` to indicates the different nested scopes?

Missing uppercase `O` for `output` (consistency with other options)