AgentDB Vector Search

Semantic search retrieves documents by meaning rather than keywords. This Skill indexes vector representations of documents and queries them using similarity metrics, enabling fast, scalable retrieval across large knowledge bases without exact keyword matching.

RAG systems require vector embeddings, a retrieval index, and integration with a language model. This Skill provides the indexing and querying layer for semantic retrieval; you supply an embedding model and Node.js 18+ environment with AgentDB v1.0.7+ to complete the pipeline.

Yes. This Skill supports HNSW (Hierarchical Navigable Small World) indexing for high-performance approximate nearest-neighbor search, along with quantization options, enabling efficient queries on large vector collections without exhaustive comparison.

You need an embedding model to generate dense vector representations from text. This Skill works with OpenAI embeddings or alternative models; the vectors themselves are indexed and queried, so any model producing compatible dense vectors is compatible.

Quantization reduces vector dimensionality or precision, lowering memory footprint and query latency while maintaining retrieval quality. This Skill supports quantization options to balance speed and accuracy for domain-specific retrieval at scale.

Keyword search matches exact terms; semantic search matches meaning by comparing vector representations. Semantic search captures synonyms, context, and intent, making it more effective for document retrieval across knowledge bases and RAG applications.