Thanks for the suggestion! Something you may wish to consider is implementing alternative tests that perform the opening of the generated file after executing the logic of the original test. You can mark these tests with a special category that you program NCrunch to ignore via a custom engine mode. In this way, you have two tests - one that executes the additional logic manually, and one that doesn't execute this logic and can run automatically. In this way, the execute path of the tests remain consistent, so you won't see strange issues with code coverage shifting around depending upon the manner in which the test is executed. Do you think this would work for you?

Understood - thanks for sharing the suggestion. I'll look into it and will see what I can do :)