semantic-search

Semantic search finds code by what it does using embeddings and natural language queries. Create a local index with odino, then query with conceptual phrases like 'authentication flow' to get ranked results with file paths and content previews.

You need odino CLI installed and a configured embedding model like BGE to build a local index. Run `odino index` to create the index, then verify with `odino status` before querying your code and documentation.

Yes, semantic search indexes code, docs, and notes together in a local project. Query conceptually across all three types and get results ranked by relevance with file locations and content samples for immediate navigation.

Semantic search understands intent and meaning rather than matching exact strings. It handles conceptual queries like 'where is the auth logic' and returns relevant files even when terminology differs, vs. grep which requires exact text matches.

No, semantic search operates entirely locally using odino and a local embedding model. Your codebase index and queries stay on your machine without external API calls or cloud dependencies.

Semantic search traverses to your project root to find .odino automatically. Moving the project may require reindexing, but the tool handles discovery of the index location, so you can rebuild quickly with `odino index` as needed.