Is it me or is Claude memory causing fixation?

I was working on some feature for tinyagents.dev and i kept updating the idea while working on it.

I wanted to build few features and then I changed my plan but in every new session it would totally misunderstand the context and i had to finally pause and ask it reindex and update its memory entirely.

This is especially worse when you have say openai and claude code working in tandem.