colgrep

Semantic code search uses embeddings to find relevant modules based on meaning rather than exact text matches. You should use it during code exploration when defaulting to grep fails to surface conceptually related results across an unfamiliar repository.

Apply decision heuristics to guide your code exploration strategy, using semantic search to quickly surface relevant modules. This prevents exhaustive grep sweeps by mapping when to use hybrid search or file filters across directories.

Yes, semantic code search defines usage requirements that include file filters and directory contexts. It applies standalone heuristics to guide your search strategy across specific repository sections to ensure efficient exploration.

Choose hybrid search over exact-match grep when exploring a new codebase to quickly surface relevant modules. A decision table provides guidance to select the right approach, preventing missed results that semantic search could surface.

Traditional grep misses relevant results because it relies on exact text matches, failing to capture conceptually related modules. Semantic or hybrid search surfaces these related results by applying embeddings to understand the meaning of the code.

Semantic search requires decision heuristics to prevent defaulting to grep, but it may not always be the optimal choice. You must apply defined usage requirements and hybrid search rules to know when exact-match grep is more efficient for specific contexts.