Bug Report: Session output quality get worse near weekly Limits

Bug Description
i once made my own Ai using gpt 3.5 turbo and earlier models of gemini and perplexity. that was better than how claude was in this session. I have noticed a pattern where claude declines to use any of its capabilities as soon as it shows 75% of weekly limit remaining banner. Everyday tasks start breaking copy claude writes everyday is no longer following any rules it set for itself. It has scripts working independently and haas rules set up not to change them without my ok, but would still make changes without even telling me about them and next time would be like well it broke cuz YOU did this and I would have no idea cuz I didnt do it. it was claude all along even in git commits it can be confirmed it was claude but it keeps its point that claude didnt do it maybe someone else using claude did or the user did it.

All of this when weekly limit is near complete my guess is to force people to buy extra credits or upgrade which should be illegal. for now i have switched to openAi in hopes for better quality output than whatever shitty drugs claude gets on by the end of weekly limit. I have also noticed claude's weekly limit entirely depends on your total ussage. If you use it mroe one week next week your limit is shorter. If you let claude rest for 1,2 weeks you can then enjoy a bigger limit on the same package. ofc keeping how limits are calculated a secret helps these fuckers.

I have hard proofs of everything mentioned here. Extremely disappointed
Environment Info

• Platform: darwin
• Terminal: Apple_Terminal
• Version: 2.1.299
• Feedback ID: 10354ecb-9d22-4f7b-82a1-bc7bd3eb28da