A curated knowledge hub for AI Agent memory systems and RAG technologies, covering GraphRAG, LightRAG, KAG, Mem0, MemGPT with 50+ technical documents and open-source project indexes.
Project Positioning#
memory_agent_hub is a curated knowledge hub for AI Agent memory systems and RAG technologies, maintained by a community contributor, systematically covering 2024-2025 cutting-edge solutions.
Core Content#
Agent Memory Systems#
- Long/short-term memory architecture design
- Token-level Memory (discrete explicit storage)
- Parametric Memory (model weights, LoRA modules)
- Latent Memory (KV cache, embeddings)
- Factual/Experiential/Working Memory functional classification
RAG Technology Stack#
- GraphRAG (Microsoft): Local Search / Global Search strategies
- LightRAG (HKU): Lightweight graph RAG
- KAG (Ant Group): Knowledge Augmented Generation
- HippoRAG
Storage Solutions#
- Vector Databases: Milvus, pgvector, Chroma, HNSW indexing
- Graph Databases: Neo4j, pgRouting, pg_graph
- Search Engines: Elasticsearch, PGroonga, BM25/TF-IDF
Key Referenced Open Source Projects#
- Mem0: Hybrid storage memory framework
- MemGPT/Letta: Hierarchical memory architecture
- Zep (Graphiti): Temporal knowledge graph memory
- OpenViking: ByteDance's Agent context database
- Claude-Mem: Claude Code dedicated memory system
Documentation Structure#
memory_agent_hub/
├── 01_rag/KAG/ # KAG resources
├── 03_memory_agent/ # Memory Agent implementations
├── 04_openAI/ # OpenAI related
├── 05_prompt/ # Prompt engineering
├── zfig/ # Figure resources
├── 50+ Markdown technical docs
└── readme.md
Topics covered: Embedding model comparison, Rerank, Fine-tuning, NER entity recognition, community detection algorithms
Usage#
git clone https://github.com/1850298154/memory_agent_hub.git
cd memory_agent_hub
Recommended reading path: readme.md → 01.agent_memory_and_knowledge_base.md → 02.kg_rag_GraphRag_llm.md
Use Cases#
- AI Agent memory system design and solution selection
- RAG technology research and architecture design
- Vector/graph database selection comparison
- Technical research and frontier paper tracking
Notes#
- This project is a knowledge base/resource aggregation, not executable software
- License type not explicitly declared
- Maintainer identity: GitHub user 1850298154