Test Coverage Code Review Framework

Test coverage reviews fail when they evaluate coverage percentages rather than coverage completeness. A function with 95% line coverage can still have no test for its primary error condition if the error condition is triggered by an input that happens to execute through a different code path. Coverage metrics measure which lines ran — not whether the right assertions were made. A test that runs every line but only asserts that no exception was thrown covers nothing useful.

The Test Coverage Review Framework identifies missing test cases by working from the specification backward — what should be tested, then whether it is — rather than from coverage metrics forward.