Presumably to crank out multiple features at the same time. After spending some time writing project rules for Cursor I've gotten it to reliably implement end-to-end CRUD operations from a simple description of the fields and functionality. It's pretty neat and surprisingly accurate but it does take time to generate on the order of ~1k LOC so I understand the desire for parallelization. If you have a well-factored codebase with loose coupling, good abstractions, etc. this should be pretty doable without them stepping on each other.
Presumably to crank out multiple features at the same time. After spending some time writing project rules for Cursor I've gotten it to reliably implement end-to-end CRUD operations from a simple description of the fields and functionality. It's pretty neat and surprisingly accurate but it does take time to generate on the order of ~1k LOC so I understand the desire for parallelization. If you have a well-factored codebase with loose coupling, good abstractions, etc. this should be pretty doable without them stepping on each other.