Module multi_memory 

Multi-Memory System

Hierarchical memory system with short-term, long-term, and entity memory.

Inspired by CrewAI’s memory architecture.

Memory Types

• Short-term Memory: Recent conversation context, auto-expires
• Long-term Memory: Persistent facts and learnings, vector-searchable
• Entity Memory: Knowledge about specific entities (people, places, concepts)

Example

use cortex::multi_memory::{MultiMemory, MemoryConfig};

let memory = MultiMemory::new(MemoryConfig::default());

// Store in short-term (recent context)
memory.short_term.add("User asked about Rust").await?;

// Store in long-term (persistent knowledge)
memory.long_term.store("Rust is a systems programming language", embedding).await?;

// Store entity information
memory.entity.update("Rust", "category", "programming_language").await?;
memory.entity.update("Rust", "creator", "Mozilla").await?;

// Query across all memory types
let context = memory.recall("Tell me about Rust", query_embedding).await?;

Structs

Entity

An entity with attributes

EntityMemory

Entity memory for tracking knowledge about specific entities

LongTermEntry

Entry in long-term memory with embedding

LongTermMemory

Long-term memory with semantic search

MemoryConfig

Configuration for the multi-memory system

MemoryItem

A memory item with metadata

MemoryStats

Memory statistics

MultiMemory

Unified multi-memory system combining all memory types

RecallResult

Result of a memory recall

ShortTermMemory

Short-term memory for recent conversation context

Enums

MemorySource

Source of a memory