I saw people mentioning the security fix wasn't adequate by using Claude, but what was the review process like? I find it extremely hard to believe nobody ran an adversarial agent review of the fixes before they got pushed out.
commit c4d62ae
c4d62ae
commit c4d62ae