The committers listed above are authorized under a signed CLA.

• ✅ login: RobinVdBroeck / name: Robin Van den Broeck (79e2adf, a856d6e)

Thanks @RobinVdBroeck for this one :) unfortunately from what I'm seeing, our exports.builder function is a pretty tight wrapper around yargs, which I'm familiar with but not an expert in. Right now, I don't know how to test custom values being passed into this builder, it seems to always read from process.argv, so we can't test it in a unit test without significant refactoring, and I agree that an integration test feels like overkill.

Curious if @mochajs/maintenance-crew has any insights here, otherwise this may sit for a bit while I work on other bugfixes and think about this one

Yeah, I think it's fine to omit testing for this one. The behavioral change is one line. Tests would be ideal but 🤷 it's old code locked into a design pattern we don't control.

I've pushed 2 new commits:

• One fixes the existing integration tests that depend on .only, since these now fail while being run in CI. I missed this in my original PR since I did not run tests with CI=true 😅
• I've added that --forbid-only is the default in CI environments to the error message. Let me know if this is too much, I was on the fence myself, and I'll drop the commit again.

Hi @RobinVdBroeck would you mind resolving the new conflicts? They should just be format fixes, npm run format:prettier should resolve most of them :)

The new message looks good to me, thank you