I would prefer to have some kind of confirmation before something is actually done. For example Hey could show the actual command sequence to be executed and then ask if that is ok to do.
When would this really be needed? Hey isn't working with anything really sensitive right now.
For one thing it would be interesting to learn something about Linux. The command would show what would actually be needed. In my opinion Hey is probably only useful (used) by (would be / wannabe) programmers.
For the other thing: I would expect it to do more sensitive things in the future. And also: if it would create a directory where I would it not want to be, it would help to say simply no instead of having to remove it later ('Oh, thats the root where it was created; it should have been the Documents folder').
This could be good if you figure out how to support advanced arguments etc. Right now it seems to be a system to handle a large number of aliases for common commands.
I need some ideas for commands, if anyone had any suggestions that'd be appreciated.
Very interesting, thanks for the tip. This is my first time parsing natural language like this, it is a bit challenging.
How about supporting composite operations, which invoke more than one command, e.g. a shell pipeline, or an IFTTT-like join between web services? That would make this more than a shell translator.
I could do this using a keyword such as "then"
"Hey google "Hacker News" then find me a sushi bar"