skills.rest
gtbauke
gtbaukeCOMMUNITY
Explore all skills by this author →
GitHub

memory-systems

Designs agent memory architectures using vector stores, knowledge graphs, and temporal persistence frameworks.

Updated Feb 20, 2026
Search Tags:#knowledge graph#semantic search#persistence#agent memory#vector store#mem0#temporal reasoning
One-click install
npx skills add https://github.com/gtbauke/bj-utils --skill memory-systems-gtbauke
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: memory-systems
Source: https://github.com/gtbauke/bj-utils/tree/main/.agent/skills/memory-systems
Command: npx skills add https://github.com/gtbauke/bj-utils --skill memory-systems-gtbauke

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires numpy, and includes scripts (resource) and references (resource) components.

What problem does it solve? Agents lose all state when sessions end, forcing users to repeat context and preventing long-term knowledge accumulation. This Skill guides the design of layered memory architectures that persist facts, entities, and relationships across sessions with reliable retrieval. ## Core Features & Use Cases - Framework Selection: Compares production memory frameworks (Mem0, Zep/Graphiti, Letta, LangMem, Cognee) with benchmark data from LoCoMo, LongMemEval, DMR, and HotPotQA to match frameworks to retrieval patterns. - Layered Memory Design: Provides decision tables for working, short-term, long-term, entity, and temporal knowledge graph layers so you add complexity only when retrieval quality demands it. - Reference Implementations: Ships Python implementations of vector stores with metadata indexing, property graphs with entity registries, temporal knowledge graphs with validity intervals, and an integrated memory system. - Use Case: An engineer building a support agent that must remember user preferences and past issues across months of conversations uses this Skill to choose Zep for temporal queries, implement hybrid retrieval, and set up consolidation policies. ## Quick Start Ask the agent to help you choose and implement a memory framework for an agent that must persist user preferences and entity relationships across sessions.

Frequently Asked Questions about memory-systems

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I add long-term memory to an AI agent?▼

Start with file-system memory storing facts as structured JSON with timestamps, then escalate to a vector store like Mem0 when you need semantic search. Add a temporal knowledge graph such as Zep/Graphiti only when the agent needs relationship traversal or time-scoped queries.

Mem0 vs Zep vs Cognee: which memory framework should I choose?▼

Choose Mem0 for fast time-to-production with managed infrastructure, Zep/Graphiti for bi-temporal modeling and relationship reasoning, and Cognee for dense multi-layer semantic graphs with customizable ingestion pipelines. Letta fits agents needing deep self-introspection over their own memory.

What is a temporal knowledge graph for agent memory?▼

A temporal knowledge graph stores facts with valid_from and valid_until timestamps so agents can answer point-in-time queries like where a user lived in March 2024. It prevents stale facts from poisoning context by tracking when facts change.

Why does my agent retrieve outdated or conflicting memories?▼

This happens when memory lacks temporal validity tracking or consolidation. Prefer facts with the most recent valid_from timestamp, invalidate rather than discard old entries, and run consolidation when memory counts exceed thresholds or retrieval quality degrades.

How do I benchmark an agent memory system?▼

Evaluate against LoCoMo for long-conversation retention, LongMemEval for long-term memory, and HotPotQA for multi-hop reasoning. Treat results as signals for specific retrieval dimensions rather than absolute rankings, since no single benchmark is definitive.

When should I avoid complex memory architectures?▼

Avoid temporal knowledge graphs early on, since benchmark evidence shows simple approaches can win: Letta's filesystem agents scored 74% on LoCoMo versus Mem0's 68.5%. Add structure only when retrieval quality demonstrably degrades or multi-hop reasoning is required.

Related Skills

Explore More Software Engineering Skills→
rustamsrustams

context-fundamentals

Explain context engineering concepts for AI agent systems.

Community
Intermediate
LuizEduPPLuizEduPP

memory-systems

Design layered persistence architectures for agent memory frameworks.

Community
Advanced
MudraMartinMudraMartin

csharp-concurrency-patterns

Assess .NET concurrency needs and suggest appropriate patterns and tools.

Community
Advanced
aisa-groupaisa-group
90

rag-implementation

Build retrieval-augmented generation systems with embeddings, vector stores, and reranking.

Official
Intermediate