CONFIRMATION: REAL SERVER, SAME MEASURE PROTOCOL
Three consecutive batches of nine MEASURE rounds on the large parent, on each commit.
Per batch: 54 starts, 27 quick results, 27 status reads, 27 long-task cancellations.
Median is median of batch medians; spread is max minus min batch median; all samples retained.
Stable improvement: gain > larger of base/head batch-median spreads.
The initial nine-round pass is retained separately; confirmation was specified before base results were available.
No isolated database-copy or heap-allocation claim.

Task start (whole operation): 168.554 [159.055-182.417] -> 117.837 [115.459-124.082] ms; gain 50.716; max spread 23.362; stable=True
base batch medians: 159.055, 182.417, 168.554
head batch medians: 124.082, 115.459, 117.837
Task finish (whole operation): 804.859 [792.376-866.194] -> 154.534 [143.620-157.859] ms; gain 650.324; max spread 73.818; stable=True
base batch medians: 792.376, 866.194, 804.859
head batch medians: 154.534, 157.859, 143.620
Parent status-sync (exclusive operation): 83.632 [73.259-84.075] -> 34.381 [31.553-34.434] ms; gain 49.250; max spread 10.816; stable=True
base batch medians: 73.259, 84.075, 83.632
head batch medians: 34.381, 31.553, 34.434
Task cancel (whole operation): 1115.515 [1047.839-1193.961] -> 328.955 [311.617-352.317] ms; gain 786.561; max spread 146.122; stable=True
base batch medians: 1047.839, 1193.961, 1115.515
head batch medians: 352.317, 328.955, 311.617
Startup parent provider store reads: 66.555 [58.264-67.780] -> 23.206 [19.291-23.662] ms; gain 43.349; max spread 9.516; stable=True
base batch medians: 58.264, 67.780, 66.555
head batch medians: 23.206, 19.291, 23.662
Completion planning store reads: 778.896 [752.273-824.375] -> 18.073 [17.880-22.148] ms; gain 760.823; max spread 72.102; stable=True
base batch medians: 752.273, 824.375, 778.896
head batch medians: 17.880, 22.148, 18.073
Parent status-sync store reads: 66.666 [60.224-66.676] -> 6.259 [6.165-8.952] ms; gain 60.408; max spread 6.452; stable=True
base batch medians: 60.224, 66.666, 66.676
head batch medians: 6.259, 6.165, 8.952
Cancellation parent task store read: 16.913 [16.533-17.816] -> 0.317 [0.316-0.323] ms; gain 16.596; max spread 1.283; stable=True
base batch medians: 16.533, 16.913, 17.816
head batch medians: 0.316, 0.323, 0.317
Parent run interrupt-check store call: 10.283 [9.816-11.183] -> 0.515 [0.400-0.644] ms; gain 9.768; max spread 1.367; stable=True
base batch medians: 9.816, 11.183, 10.283
head batch medians: 0.644, 0.400, 0.515
base counts: {"large": {"runs": 373, "nodes": 41662, "subagents": 916, "turn_items": 42036}, "tiny": {"runs": 19, "nodes": 92, "subagents": 54, "turn_items": 112}}
base behavior: {"measureDone": 27, "errors": [], "tools": {"delegate_task": 162, "task_status": 81, "task_cancel": 81}, "persistedTaskOutcomes": [{"prompt": "PERF:QUICK", "status": "completed", "result": "QUICK_OK", "n": 835}, {"prompt": "PERF:LONG", "status": "interrupted", "result": "LONG_READY", "n": 135}], "liveRunsAfterMeasurement": 0}
head counts: {"large": {"runs": 374, "nodes": 41664, "subagents": 916, "turn_items": 42039}, "tiny": {"runs": 19, "nodes": 92, "subagents": 54, "turn_items": 112}}
head behavior: {"measureDone": 27, "errors": [], "tools": {"delegate_task": 162, "task_status": 81, "task_cancel": 81}, "persistedTaskOutcomes": [{"prompt": "PERF:QUICK", "status": "completed", "result": "QUICK_OK", "n": 835}, {"prompt": "PERF:LONG", "status": "interrupted", "result": "LONG_READY", "n": 135}], "liveRunsAfterMeasurement": 0}
