Archival Counterpart Censuses for Never-alerted IceCube Events and the Ultra-high-energy Neutrino KM3-230213A

Abstract Two archival cases lack a transient-focused counterpart census: IceCube events that pass alert-grade cuts in final offline reconstruction but never generated a real-time notice, and the ultra-high-energy event KM3-230213A, whose published counterpart work characterizes catalogued blazars, not archival transients. This work applies one census to both. In the current IceCube public track release (IceTracks-DR2; 2008–2022, 1.64 million tracks), 290 events pass these cuts; an audit against the complete AMON notice record and the IceCat-1 alert catalog isolates six (50–115 TeV, 2019–2021) with no public alert or per-event catalog entry. Each error region was vetted against transient catalogs, ASAS-SN light curves, spectroscopic surveys, Gaia DR3 astrometry, X-ray archives, and Fermi-LAT flare states. Four of the six localization regions yield no viable archival counterpart; one contains only a chance-level LAT-steady BL Lac. The sixth contains AT 2021iis, a 420 mJy radio quasar at z = 1.6827 whose sustained optical brightening began 110 days before the 50 TeV event and was rising at its epoch, followed by mid-infrared brightening consistent with a dust echo; the window transient count is unremarkable ( p = 0.11). Applied to KM3-230213A (∼220 PeV), the census returns nulls, bounds the published candidates, and adds an ATLAS-detected, ASAS-SN-confirmed optical flare of PKS 0605−08 (1.9 Jy, z = 0.87, inside the 99% localization) peaking 35 days before the neutrino. Flare-rate statistics leave both coincidences modest (≈50% chance of a flare within ±40 days of any epoch), and a triggered search on its 2018 flare shows no IceCube excess. The products are the event catalog, the census methodology with documented public-interface failure modes, and two falsifiable candidates.