codebase-rag

You can perform semantic search by running a query against a persistent, per-machine index to locate code, documentation, and patterns. It returns ranked results with file paths and relevance scores to guide your exploration.

Codebase indexing works by auto-detecting the project root, building a per-machine cache, and using a filesystem watcher to keep results up-to-date. This ensures your search index reflects the latest file modifications automatically.

Yes, you can analyze refactoring impact by querying what depends on a file to see callers and callees. This discovers related modules and helps coordinate changes during maintenance and refactoring tasks.

Yes, you need a project root for semantic code search because the indexing server auto-detects it to build the per-machine cache. The project root establishes the boundary for file discovery and indexing.

The best way to discover related modules during onboarding is using semantic search to locate code and patterns across the project. It helps new developers rapidly understand design constraints and navigate the codebase.

If semantic search is not returning expected results, the index might be limited by per-machine cache boundaries or filesystem watcher delays. Ensure your project root is correctly detected so all relevant files are indexed and watched.