Crate brainos_hippocampus 

Brain Hippocampus

Memory engine providing:

• Episodic memory (conversation storage with decay)
• Semantic memory (fact storage with vector embeddings)
• Procedural memory (learned workflows)
• Importance scoring (keyword-based, no LLM)
• Embedding pipeline (Ollama / OpenAI-compatible)
• Hybrid search (vector ANN + BM25 FTS5 + RRF fusion)
• Memory consolidation (sleep cycle)

Re-exports

pub use compactor::CompactConfig;

pub use compactor::CompactStats;

pub use compactor::Compactor;

pub use compactor::DefaultCompactor;

pub use consolidation::ConsolidationConfig;

pub use consolidation::ConsolidationReport;

pub use consolidation::Consolidator;

pub use consolidation::PromotionCandidate;

pub use dual_memory::DualMemoryError;

pub use dual_memory::DualMemoryReader;

pub use dual_memory::GraphCandidate;

pub use dual_memory::GraphCandidates;

pub use dual_memory::MemoryEntry;

pub use embedding::Embedder;

pub use embedding::EmbeddingError;

pub use embedding::EmbeddingProvider;

pub use episodic::Episode;

pub use episodic::EpisodicStore;

pub use episodic::Session;

pub use graph::Edge;

pub use graph::EdgeKind;

pub use graph::EpisodicGraph;

pub use graph::GraphError;

pub use graph::GraphHit;

pub use graph::Node;

pub use graph::NodeKind;

pub use graph::SqliteGraph;

pub use importance::ImportanceScorer;

pub use importance::ImportanceSignals;

pub use search::Memory;

pub use search::MemorySource;

pub use search::RecallConfig;

pub use search::RecallEngine;

pub use semantic::Fact;

pub use semantic::NamespaceStats;

pub use semantic::SemanticResult;

pub use semantic::SemanticStore;

Modules

compactor

Graph compactor — applies half-life decay to every node’s weight and prunes nodes whose decayed weight falls below the cutoff.

consolidation

Memory consolidation pipeline — “sleep-like” memory optimization.

dual_memory

Dual memory model — reconciliation layer between the legacy episodes table and the newer episodic graph.

embedding

Embedding pipeline — Ollama and OpenAI-compatible backends.

episodic

Episodic memory — SQLite-backed conversation store.

graph

Episodic graph store — typed nodes + typed edges backed by the nodes / edges tables.

importance

Importance scoring — keyword-based relevance tagging.

search

Recall engine — hybrid search with RRF fusion.

semantic

Semantic memory — RuVector-backed vector memory.