Would this introduce a syntax error if the call contains no arguments? Are you interested in writing a more robust add_argument utility, similar to the remove_argument that exists today?

Yes it would. open() and similar calls all have a required first argument AFAIK but it's still a pretty dirty and unsafe fix.

I'd be interested in adding that utility for sure! Will set this to draft and use the autofix as a test case for the utility if there is no urgency behind it.

Awesome, thanks @qdegraaf!

Implemented a coarse first version which only handles keyword arguments, but does check for trailing commas and work with empty calls. It explicitly forbids positional arguments but could still maybe break when adding a first argument to a class instantiation.

I could either use the lexer to check for parentheses and panic if they aren't there and leave the implementation of that to a future PR or implement it in this one. What's your preference? If the latter is preferred, if you happen to know of a rule where something like that happens on which I can test the util in absence of unit test infrastructure let me know.

I think it's fine to leave this as function-only for now, so we can assume the parentheses always exist.