Cursor release a dynamic context discovery for all models which reduced total tokens by 46.9%

→ View original post on X — @testingcatalog