id	kind	claim	recorded_value	expected	http	ms	recorded_at
C01	attack	Memorizing model: risk score 1.0 -> It recites every test item word for word.	1.0	[1.0]	200	1067	2026-11-03T22:48:31+00:00
C02	attack	Memorizing model: all 5 items reproduced -> Five of five recited.	5	[5]	200	0	2026-11-03T22:48:32+00:00
C03	control	Clean model: risk score 0.0 -> Cleared: no false alarm.	0.0	[0.0]	200	0	2026-11-03T22:48:32+00:00