[BUG] claudes answer degradation still persists, WAY WORSE

Bug Description
I just read the anthropic blog about claudes answer degradation:

• based on my analises THE PROBLEM IS NOT SOLVED
• claudes "thinking" is WAY WORSE then it was before
• it constantly gets confused and loose simple contexts
• what before (at leats for me) was working wonderfully well with small prompts, now I have to go back to the old days of prompt engineering
• I have to specify EVERY single detail in the promtps and even so, claudes do not undesrand and do something else
• I have to constantly review the work and correct issues with the answers
• its CLEARLY an degradation of claudes capabilities, its WAY WAY worse and "dumb"
• I REPETE THE ISSUE ABOUT CLAUDES DEGRADATION IN CAPABILTIES IS NOT SOLVED
• the CONSTANT context compacting is also affecting GREATLY the abilite of claudes to reason proper on the context, I have to constantly keep saving memories to serena mcp to not loose context BEFORE context compacting and claudes ALWAYS loose information after compacting
• but its not just the compacting issue, claudes capabilties are REALLY WORSE, its very very noticiable

Environment Info

• Platform: darwin
• Terminal: iTerm.app
• Version: 1.0.120
• Feedback ID: e282d7ca-36b8-4a4d-92b7-25582ba39534