I haven't yet found a "boring, easy" problem that an AI could generate more easily than Vim.

Vim can write basic unit tests?