I have been doing LLVM-related development work in the past versions of Cursor. As we all know, the code base of LLVM is very large. Codebase Indexing works very well, and I can easily use Ask or Agent to complete many tasks. But after upgrading to 0.48, Cursor directly rejected the Indexing request, and still reported too many files after adding some entries to cursorignore. This hindered and delayed the development process, and I may have to roll back to the previous version to continue working.
Is there any way to change this situation?