Expand description
Agent memory system — Memory trait, in-memory and PostgreSQL stores, BM25 and vector recall, Ebbinghaus decay, reflection, and consolidation.
Modules§
- bm25
- BM25 keyword scoring for memory recall.
- consolidation
- Memory consolidation pipeline — clusters and merges episodic memories into semantic summaries.
- embedding
- Embedding providers for semantic memory retrieval.
- hybrid
- Hybrid retrieval — cosine similarity + BM25 fused via Reciprocal Rank Fusion.
- in_
memory - In-memory
Memoryimplementation backed by a sortedVec. - namespaced
- Namespaced memory wrapper that scopes all operations to a tenant or agent prefix.
- pruning
- Memory entry pruning — removes weak or stale entries to keep the store compact.
- reflection
- Reflection tracker — triggers agent self-reflection when cumulative memory importance exceeds a threshold.
- scoring
- Composite memory recall scoring — recency, importance, relevance, and Ebbinghaus strength.
- shared_
tools - Shared-memory tool definitions for inter-agent institutional memory access.
- tools
- Per-agent memory tool definitions (store, recall, update, forget, consolidate).
Structs§
- Memory
Entry - A single memory entry stored by an agent.
- Memory
Query - Query parameters for recalling memories.
Enums§
- Confidentiality
- Access classification for memory entries.
- Memory
Type - Classification of a memory entry’s origin and purpose.
Traits§
- Memory
- Trait for persistent memory stores.