Memory System Documentation¶
Home > Memory System
Ahoy, matey! Welcome to the memory system documentation - where agents remember what they learned and share knowledge across sessions.
Quick Navigation¶
New to memory systems? Start here:
- 5-Type Memory Guide - NEW! Understand the 5 psychological memory types
- Quick Reference - NEW! One-page cheat sheet for memory system
- Agent Memory Integration - How agents share and persist knowledge
- Agent Memory Quickstart - Get started in 5 minutes
Looking for something specific?
- 5-Type Memory System - NEW! Episodic, Semantic, Procedural, Prospective, Working
- Terminal Visualization - NEW! View memory graph in terminal
- Kùzu Backend - NEW! Graph database schema
- Neo4j Memory System - Graph-based memory
- Testing & Validation - Memory system tests
- External Knowledge - Import external data
5-Type Memory System¶
NEW! Psychological memory model with 5 types, pluggable backends, and automatic hooks.
Core Documentation¶
- 5-Type Memory Guide - Complete user guide for all 5 memory types
- Developer Reference - Architecture, API reference, integration patterns
- Quick Reference - One-page cheat sheet and decision tree
Architecture & Schema¶
- Kùzu Memory Schema - Graph database schema with 5 node types
- Backend Architecture - Pluggable backend design (SQLite, Kùzu, Neo4j)
- Evaluation Framework - Compare backend quality and performance
Features¶
Memory Types:
- Episodic: Session-specific events and conversations
- Semantic: Cross-session knowledge and patterns
- Prospective: Future intentions and reminders
- Procedural: How-to workflows and procedures
- Working: Active task state (auto-cleared on completion)
Backends:
- Kùzu (default): Embedded graph database, zero infrastructure
- SQLite: Relational storage, fast and simple
- Neo4j: External graph database (optional)
Quality Gates:
- Multi-agent review (3 agents: analyzer, patterns, knowledge-archaeologist)
- Trivial content filtering
- Selective retrieval with token budgets
Terminal Visualization¶
NEW! View memory graph directly in your terminal with beautiful Rich-powered trees.
- Memory Tree Visualization Guide - How to use
amplihack memory tree - Command:
amplihack memory tree [--session SESSION] [--type TYPE] - Color-coded by memory type with emoji indicators
- Star ratings for importance scores
- Filter by session, type, or agent
Kùzu Backend¶
NEW! Embedded graph database as default memory backend.
- Kùzu Memory Schema - Complete schema documentation
- 5 separate node types (EpisodicMemory, SemanticMemory, etc.)
- 11 semantic relationships (CONTAINS_EPISODIC, CONTRIBUTES_TO_SEMANTIC, etc.)
- Sessions as first-class entities
- Zero Docker overhead (embedded database)
Neo4j Memory System¶
Advanced graph-based memory for complex knowledge representation:
- Neo4j Implementation Summary - Current status
- Neo4j Phases 1-6 Complete - Implementation milestones
- Neo4j Validation Checklist - Ensure proper setup
- Neo4j Quick Reference - Fast answers
Research & Deep Dives¶
For comprehensive research on the Neo4j memory system:
- Neo4j Memory Research - Complete research archive
- Executive Summary
- Technical Research
- Design Patterns
- Integration Guides
- External Knowledge
Testing & Validation¶
Memory system performance and validation:
- A/B Test Summary - Performance comparisons
- A/B Test Quick Reference - Test results at a glance
- Effectiveness Test Design - How we measure success
- Final Cleanup Report - Memory system cleanup
Code Quality¶
- Code Review PR 1077 - Example: Memory system review
- Zero-BS Audit - Quality audit results
External Knowledge¶
Import and integrate external data sources:
- External Knowledge Integration - Import external data sources
- Blarify Integration - Connect with Blarify knowledge base
- Blarify Quickstart - Get started with Blarify
- Blarify Architecture - Understanding the Blarify integration
Memory Patterns¶
Advanced patterns for memory collaboration:
- Agent Type Memory Sharing Patterns - Patterns for memory collaboration
- Neo4j Phase 4 Implementation - Latest features
Related Documentation¶
- Agents Overview - How agents use memory
- Document-Driven Development - Memory in DDD workflow
- Goal Agent Generator - Create custom goal-seeking agents with memory
Need help? Check the Troubleshooting Guide or Discoveries for common issues.