[Bug] Degraded model output quality: Claude makes incorrect statements requiring frequent corrections since v4.7

Bug Description
❯ It seems like over the last several sessions you constantly make statements that you end up walking
back when challenged. This is starting to become a very real problem. It started after 4.7 was
introduced and has consistently been degrading over time. Is there a significant decrease in model
output quality or is this something I am just imagining? It is starting to feel like Claude Code has
gone from a Sr/Staff Engineer who could autonomously solve problems and effectively execute to an
entry level engineer who consistently has to be redirected, corrected, and has to walk back their
statements. It is concerning.

Environment Info

• Platform: darwin
• Terminal: iTerm.app
• Version: 2.1.143
• Feedback ID: e863b1bc-9131-4e5e-b0de-23cb117aab16

Errors

[]