What is the difference between vector stores and knowledge graphs in agent memory?

Vector stores use semantic similarity for retrieval but lack relationship structure, whereas knowledge graphs explicitly preserve connections between entities, enabling more complex reasoning.

How does this skill help manage the Claude context window?

It provides strategies for 'just-in-time' memory loading, where only the most relevant snippets or subgraphs are injected into the context window, optimizing token usage and reducing latency.

When should I implement a temporal knowledge graph?

Use temporal knowledge graphs when your application needs to track how facts change over time, allowing for 'time-travel' queries and preventing outdated data from conflicting with current information.

What are the common layers in a sophisticated memory architecture?

A standard architecture includes Working Memory (immediate context), Short-Term Memory (session-persistent), Long-Term Memory (cross-session), and Entity Memory (structured facts about people/objects).

Memory Systems Design

Name: Memory Systems Design
Author: monmacllcapp

bymonmacllcapp

0•

Data Science & ML

Implements sophisticated memory architectures for AI agents to persist state, build knowledge graphs, and maintain long-term context.

This skill provides comprehensive guidance on designing and implementing multi-layered memory systems for AI agents, moving beyond simple context windows to include short-term session storage, long-term vector RAG, and advanced temporal knowledge graphs. It enables developers to build agents that learn from past interactions, maintain entity consistency across sessions, and reason over accumulated knowledge with high accuracy and reduced latency using benchmarked patterns like GraphRAG and Zep.

Key Features

01Layered architecture design (Working, Short-Term, Long-Term, and Permanent)

02Memory consolidation strategies to prevent unbounded growth and manage outdated facts

03Entity tracking to maintain identity and relationship consistency across sessions

04Temporal Knowledge Graph implementation for time-aware data retrieval

050 GitHub stars

06Performance-optimized retrieval patterns including semantic and entity-based search

Use Cases

01Creating time-sensitive enterprise agents that track changing entity states and historical data

02Developing complex research agents that construct structured knowledge bases from multiple sources

03Building persistent personal assistants that remember user preferences and history across months

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add monmacllcapp/skill-forks memory-systems

For use in Claude.ai and ChatGPT

Download Skill