Loading the drill…
The scenarios live in hanko/drill.py and run in the test suite on every change (tests/test_drill.py), so a defence that stops holding fails the build before it reaches this page.
Each scenario injects one real failure into Hanko's real code: not a mock of Hanko, a mock of the world around it. Then it computes whether the promise held, so a scenario can come back BROKE. One did, the first time it ran: a restarted sweep recorded the same decision twice. The fix is D-18 in the decision log. Everything runs offline against recorded responses, so pressing a button here costs nothing and calls no one.
The scenarios live in hanko/drill.py and run in the test suite on every change (tests/test_drill.py), so a defence that stops holding fails the build before it reaches this page.