[Feature Request] Claude model should respect CLAUDE.md configuration file directives

Bug Description
what is the point of a CLAUDE.md if the model does not respect it? i think the models should be trained to listen to the user and CLAUDE.md. i understand Anthropic needs safeguards but if the model isn't being trained to respect or try to understand the user, it's not really useful or trust worthy.

Environment Info

• Platform: linux
• Terminal: xterm-256color
• Version: 2.1.266
• Feedback ID: d510e72c-6aab-497e-884c-b53042c5fb3a

Errors

[]