Test coverage
also: code coverage, coverage report
A measure of how much of your code actually runs while your tests run, usually shown as a percentage.
A metric from tools like Istanbul or c8 showing which lines, branches or functions executed during the test suite. It reveals untested code, but a line running during a test is not the same as its result being checked.
A security guard's patrol log shows which rooms they walked through. It doesn't show whether they actually checked the locks in each room.
Agents sometimes chase a coverage number. A test can run code without checking anything meaningful, and it still counts. Use coverage to find untested areas, not as proof things work.
Coverage for src/billing is now 88%. The untested parts are the refund error paths, which need a simulated payment provider failure to test.
Run the coverage report and tell me which important files have little or no test coverage. Don't write tests just to raise the number.
Believing high coverage means the code is correct. Coverage measures which code ran during tests, not whether the tests checked the right results.