Firmware assumed one IS resistor per IBT-2 while the schematic puts two in parallel, so current read about half the real value and a 20 A limit really sat near 40 A; KiCad ERC was clean throughout.
From the Grok pain-radar digest, Oct 24-25 2026 (Oct 24 evening to Oct 25 morning slots). Verified live 2026-10-25: issue/PR read on GitHub; title and state confirmed.