You can index every repo in Cursor. It still can't tell you what breaks.

Short summary

Cursor's semantic search indexing can be tuned to make queries fast on large monorepos through .cursorignore files and package-level scoping, reducing indexing time from hours to minutes. However, no amount of tuning overcomes a fundamental limitation: semantic search finds text similarity, not declared dependencies, so Cursor cannot reliably detect which remote repositories break when you change a shared module. The tuning itself is worth doing, but developers need additional tooling for cross-repo dependency impact analysis.

• •Cursor uses vector embeddings for semantic code search; tuning file scope dramatically improves performance
• •Practical strategies: use .cursorignore, open package directories as workspace root, create shell aliases per package
• •Hard architectural limit: semantic search can't resolve cross-repository dependencies, so it won't tell you what breaks in other repos

Generated with AI, which can make mistakes.