A computer-implemented method for evaluating software code includes measuring a first coverage of a test applied to the software code and then making a modification in a first section of the software code. A second coverage of the test applied to the software code is measured after making the modification. A difference is identified between the first coverage and the second coverage in a second section of the software code, which is separate from the first section and was not modified since the first coverage was measured, and an indication of the difference is output.