Expand description
claw-vector is the semantic memory engine for ClawDB.
It provides HNSW-backed approximate nearest-neighbour search, SQLite persistence, memory-mapped vector files, and a gRPC client to a Python embedding microservice.
Re-exports§
pub use collections::CollectionManager;pub use config::VectorConfig;pub use config::VectorConfigBuilder;pub use engine::VectorEngine;pub use error::VectorError;pub use error::VectorResult;pub use types::Collection;pub use types::CollectionStats;pub use types::DistanceMetric;pub use types::EngineStats;pub use types::HybridQuery;pub use types::IndexType;pub use types::MetadataFilter;pub use types::RerankerConfig;pub use types::SearchMetrics;pub use types::SearchQuery;pub use types::SearchResponse;pub use types::SearchResult;pub use types::VectorRecord;
Modules§
- collections
- High-level collection management operations.
- config
- Engine configuration and builder.
- embeddings
- gRPC client to the Python embedding microservice and related types.
- engine
- Top-level VectorEngine lifecycle coordinator.
- error
- Unified error type and result alias.
- grpc
- gRPC server and generated proto bindings.
- index
- HNSW, flat, and selector index implementations.
- search
- ANN search, hybrid search, reranking, and metadata filters.
- store
- SQLite and mmap vector storage backends.
- types
- Core domain types (VectorRecord, Collection, SearchResult, etc.).