9 real Playwright recordings (8 benchmark + your live run) · scored by wm-mcp v0.8.0 with calibrated anomaly probabilities · defects fire at ~6s
broken_img
6.0s broken image
wm verdict: 4.0–5.88s · z=1.31 · peak P(anomaly)=0.199
0s → end · bar height = P(anomaly) per 2s window
button_color
6.0s button color change
wm verdict: 4.0–5.88s · z=1.59 · peak P(anomaly)=0.204
0s → end · bar height = P(anomaly) per 2s window
layout_shift
6.0s layout shift
wm verdict: 5.0–6.88s · z=2.24 · peak P(anomaly)=0.943
0s → end · bar height = P(anomaly) per 2s window
live_modal
LIVE run you watched — 6.0s phantom modal
wm verdict: 6.0–7.88s · z=1.96 · peak P(anomaly)=0.979
0s → end · bar height = P(anomaly) per 2s window
modal
6.0s phantom modal
wm verdict: 5.0–6.88s · z=2.36 · peak P(anomaly)=0.939
0s → end · bar height = P(anomaly) per 2s window
none
clean control — no defect
wm verdict: 4.0–5.88s · z=2.05 · peak P(anomaly)=0.302
0s → end · bar height = P(anomaly) per 2s window
stuck_spinner
5.0s stuck spinner
wm verdict: 4.0–5.88s · z=1.12 · peak P(anomaly)=0.184
0s → end · bar height = P(anomaly) per 2s window
theme_flash
6.0s theme flash (250ms)
wm verdict: 5.0–6.88s · z=1.96 · peak P(anomaly)=0.917
0s → end · bar height = P(anomaly) per 2s window
toast
6.0s toast (small corner)
wm verdict: 0.0–1.88s · z=1.45 · peak P(anomaly)=0.214
0s → end · bar height = P(anomaly) per 2s window