Awesome Graph-based Agent Memory

This repository provides a comprehensive collection of research papers, benchmarks, and open-source projects on Graph-based Agent Memory. It includes contents from our survey paper 📖"Graph-based Agent Memory: Taxonomy, Techniques, and Applications" and will be continuously updated.

🤗 You are very welcome to contribute to this repository by launching an issue or a pull request. If you find any missing resources or come across interesting new research works, please don’t hesitate to open an issue or submit a PR!

📫 Contact us via emails: chang.yang@connect.polyu.hk, qinggang.zhang@polyu.edu.hk

---

📜 Catalog

Awesome Graph-based Agent Memory

• 🔥 News
• 📖 Overview
• 📚 Related Survey
• 🪴 Taxonomy
  • Memory Extraction
  • Memory Storage
  • Memory Retrieval
  • Memory Evolution
• 🏆 Benchmark
• 📦 Projects
• 📃 Citation

---

🔥 News

• [2026-02-03] 🔥🔥 Our survey on Graph-based Agent Memory is released.

📚 Related Survey Papers

• (TOIS'25) A Survey on the Memory Mechanism of Large Language Model-based Agents [Paper]
• (TMLR'25) The AI Hippocampus: How Far are We From Human Memory? [Paper]
• (arXiv'25) Memory in the Age of AI Agents [Paper]
• (arXiv'25) Memory in LLM-based Multi-agent Systems: Mechanisms, Challenges, and Collective Intelligence [Paper]
• (arXiv'25) AI Meets Brain: Memory Systems from Cognitive Neuroscience to Autonomous Agents [Paper]
• (arXiv'25) From Human Memory to AI Memory: A Survey on Memory Mechanisms in the Era of LLMs [Paper]

🪴 Taxonomy

Memory Extraction

Textual Data

• (arXiv‘25) Can an LLM Induce a Graph? Investigating Memory Drift and Context Length [Paper]
• (arXiv‘24) On the structural memory of llm agents [Paper]
• (arXiv‘25) Personaagent with graphrag: Community-aware knowledge graphs for personalized llm [Paper]
• (arXiv‘25) Scaling graph chain-of-thought reasoning: A multi-agent framework with efficient llm serving [Paper]
• (ACL‘25) HiAgent: Hierarchical Working Memory Management for Solving Long-Horizon Agent Tasks with Large Language Model [Paper]

Sequential Data

• (NeurIPS‘23) Reflexion: Language agents with verbal reinforcement learning [Paper]
• (arXiv‘25) Mem-α: Learning Memory Construction via Reinforcement Learning [Paper]

Multimodal Data

• (arXiv‘25) MemoryVLA: Perceptual-Cognitive Memory in Vision-Language-Action Models for Robotic Manipulation [Paper]
• (arXiv‘23) Multi-Temporal Lip-Audio Memory for Visual Speech Recognition [Paper]
• (NeurIPS‘24) Optimus-1: Hybrid multimodal memory empowered agents excel in long-horizon tasks[Paper]

Memory Storage

Knowledge Graph Structure

• (TMLR'25) MemLLM: Finetuning LLMs to Use An Explicit Read-Write Memory [Paper]
• (arXiv'25) AriGraph: Learning Knowledge Graph World Models with Episodic Memory for LLM Agents [Paper]
• (arXiv'25) Mem0: Building Production-Ready AI Agents with Scalable Long-Term Memory [Paper]

Hierarchical Memory Structure

• (arXiv'26) Beyond RAG for Agent Memory: Retrieval by Decoupling and Aggregation [Paper]
• (arXiv'25) ENGRAM: Effective, Lightweight Memory Orchestration for Conversational Agents [Paper]
• (arXiv'25) SGMem: Sentence Graph Memory for Long-Term Conversational Agents [Paper]
• (arXiv'25) G-Memory: Tracing Hierarchical Memory for Multi-Agent Systems [Paper]
• (arXiv'25) LLM-Powered Decentralized Generative Agents with Adaptive Hierarchical Knowledge Graph for Cooperative Planning [Paper]
• (EMNLP'24) Crafting Personalized Agents through Retrieval-Augmented Generation on Editable Memory Graphs [Paper]

Temporal Graph Structure

• (arXiv'25) Zep: A Temporal Knowledge Graph Architecture for Agent Memory [Paper]
• (arXiv'25) TReMu: Towards Neuro-Symbolic Temporal Reasoning for LLM-Agents with Memory in Multi-Session Dialogues [Paper]
• (arXiv'25) MemoTime: Memory-Augmented Temporal Knowledge Graph Enhanced Large Language Model Reasoning [Paper]

Hypergraph Structure

• (arXiv'25) HyperGraphRAG: Retrieval-Augmented Generation via Hypergraph-Structured Knowledge Representation [Paper]
• (arXiv'25) HyperG: Hypergraph-Enhanced LLMs for Structured Knowledge [Paper]

Hybrid Graph Architectures

• (arXiv'26) MAGMA: A Multi-Graph based Agentic Memory Architecture for AI Agents [Paper]
• (arXiv'24) Optimus-1: Hybrid Multimodal Memory Empowered Agents Excel in Long-Horizon Tasks [Paper]
• (arXiv'24) KG-Agent: An Efficient Autonomous Agent Framework for Complex Reasoning over Knowledge Graph [Paper]

Memory Retrieval

Retrieval Techniques

Similarity-based

• (CVPR'25) Seeing, Listening, Remembering, and Reasoning: A Multimodal Agent with Long-Term Memory [Paper]
• (aXiv'25) Zep: a temporal knowledge graph architecture for agent memory. [paper]
• (aXiv'25) Mem0: Building production-ready ai agents with scalable long-term memory. [paper]
• (arXiv'25) G-Memory: Tracing Hierarchical Memory for Multi-Agent Systems [Paper]

Rule-based

• (arXiv'25) MemInsight: Autonomous Memory Augmentation for LLM Agents [Paper]
• (arXiv'25) Neural Graph Memory: A Structured Approach to Long-Term Memory in Multimodal Agents [Paper]
• (IEEE Trans. Big Data) FinMem: A Performance-Enhanced LLM Trading Agent with Layered Memory and Character Design [Paper]
• (CHI'24) "My agent understands me better": Integrating Dynamic Human-like Memory Recall and Consolidation in LLM-Based Agents [Paper]
• (ACL'25 Findings) TReMu: Towards Neuro-Symbolic Temporal Reasoning for LLM-Agents with Memory in Multi-Session Dialogues [Paper]

Graph-based

• (aXiv'25) Mem0: Building production-ready ai agents with scalable long-term memory. [paper]
• (aXiv'25) Zep: a temporal knowledge graph architecture for agent memory. [paper]
• (arXiv'25) G-Memory: Tracing Hierarchical Memory for Multi-Agent Systems [Paper]
• (arXiv'25) SGMem: Sentence Graph Memory for Long-Term Conversational Agents [Paper]
• (arXiv'25) Hierarchical Memory for High-Efficiency Long-Term Reasoning in LLM Agents [Paper]
• (arXiv'25) LiCoMemory: Lightweight and Cognitive Agentic Memory for Efficient Long-Term Reasoning [Paper]

Temporal-based

• (arXiv'25) AssoMem: Scalable Memory QA with Multi-Signal Associative Retrieval [Paper]
• (arXiv'25) Mnemosyne: An Unsupervised, Human-Inspired Long-Term Memory Architecture for Edge-Based LLMs [Paper]
• (aXiv'25) Zep: a temporal knowledge graph architecture for agent memory. [paper]
• (arXiv'25) LiCoMemory: Lightweight and Cognitive Agentic Memory for Efficient Long-Term Reasoning [Paper]
• (arXiv'25) MemoTime: Memory-Augmented Temporal Knowledge Graph Enhanced Large Language Model Reasoning [Paper]

RL-based

• (arXiv'26) MAGMA: A Multi-Graph based Agentic Memory Architecture for AI Agents [Paper]
• (arXiv'25) Memento: Fine-tuning LLM Agents without Fine-tuning LLMs [Paper]
• (arXiv‘25) Mem-α: Learning Memory Construction via Reinforcement Learning [Paper]
• (arXiv'25) From Experience to Strategy: Empowering LLM Agents with Trainable Graph Memory [Paper]
• (arXiv'25) Memory-augmented Query Reconstruction for LLM-based Knowledge Graph Reasoning [Paper]
• (aXiv'25) Memory-r1: Enhancing large language model agents to manage and utilize memories via reinforcement learning. [paper]
• (ACL'25) In Prospect and Retrospect: Reflective Memory Management for Long-term Personalized Dialogue Agents [Paper]
• (EMNLP'24) Crafting Personalized Agents through Retrieval-Augmented Generation on Editable Memory Graphs [Paper]

Agent-based

• [Arxiv'24] Leveraging knowledge graph-based human-like memory systems to solve partially observable markov decision processes. [paper]
• [Arxiv'24] Arigraph: Learning knowledge graph world models with episodic memory for llm agents. [paper]
• [NeurIPS'24] Optimus-1: Hybrid multimodal memory empowered agents excel in long-horizon tasks. [paper]
• [Arxiv'25] Omni Memory System for Personalized, Long Horizon, Self-Evolving Agents. [paper]
• [ICML'25] Cradle: Empowering Foundation Agents towards General Computer Control. [paper]
• [Arxiv'25] Llm-powered decentralized generative agents with adaptive hierarchical knowledge graph for cooperative planning. [paper]
• [Arxiv'25] MemGPT: Towards LLMs as Operating Systems. [paper]
• [Arxiv'25] Cognitive Architectures for Language Agents. [paper]
• [ACL'25] Hiagent: Hierarchical working memory management for solving long-horizon agent tasks with large language model. [paper]
• [Arxiv'25] GraphCogent: Mitigating LLMs' Working Memory Constraints via Multi-Agent Collaboration in Complex Graph Understanding. [paper]
• (Arxiv'25) Memory-r1: Enhancing large language model agents to manage and utilize memories via reinforcement learning. [paper]
• (Arxiv'25) Enhancing Multi-Agent Systems via Reinforcement Learning with LLM-based Planner and Graph-based Policy. [paper]
• (Arxiv'25) Collaborative Memory: Multi-User Memory Sharing in LLM Agents with Dynamic Access Control. [paper]
• [Arxiv'26] SwiftMem: Fast Agentic Memory via Query-aware Indexing [paper]

Retrieval Strategy

Multi-round Retrieval

• [arXiv'25] General Agentic Memory Via Deep Research. [paper]
• [ACL'25 Findings] Bridging intuitive associations and deliberate recall: Empowering llm personal assistant with graph-structured long-term memory. [paper]
• [arXiv'25] Rcr-router: Efficient role-aware context routing for multi-agent llm systems with structured memory. [paper]
• [arXiv'25] MemSearcher: Training LLMs to Reason, Search and Manage Memory via End-to-End Reinforcement Learning. [paper]
• [arXiv'25] MemoTime: Memory-Augmented Temporal Knowledge Graph Enhanced Large Language Model Reasoning. [paper]
• [arXiv'25] On the structural memory of llm agents. [paper]
• [arXiv'25] LEGOMem: Modular Procedural Memory for Multi-agent LLM Systems for Workflow Automation. [paper]
• [arXiv'25] Ghost in the minecraft: Generally capable agents for open-world environments via large language models with text-based knowledge and memory. [paper]

Post-Retrieval

• [ACL'25 Findings] Simgrag: Leveraging similar subgraphs for knowledge graphs driven retrieval-augmented generation. [paper]
• [arXiv'25] Mirix: Multi-agent memory system for llm-based agents. [paper]
• [arXiv'25] MemGen: Weaving Generative Latent Memory for Self-Evolving Agents. [paper]

Multi-Source Retrieval

• [arXiv'25] MemSearcher: Training LLMs to Reason, Search and Manage Memory via End-to-End Reinforcement Learning. [paper]
• [arXiv'25] ** MEM1: Learning to Synergize Memory and Reasoning for Efficient Long-Horizon Agents**. [paper]

Memory Evolution

Internal Self-Evolving

• (arXiv'26) MAGMA: A Multi-Graph based Agentic Memory Architecture for AI Agents [Paper]
• (aXiv'25) Zep: a temporal knowledge graph architecture for agent memory. [paper]
• (aXiv'25) Nemori: Self-organizing agent memory inspired by cognitive science. [paper]
• (aXiv'25) Mem-α: Learning Memory Construction via Reinforcement Learning. [paper]
• (aXiv'24) From local to global: A graph rag approach to query-focused summarization. [paper]
• (aXiv'23) RecallM: An Adaptable Memory Mechanism with Temporal Understanding for Large Language Models. [paper]
• (aXiv'25) Agent kb: Leveraging cross-domain experience for agentic problem solving. [paper]
• (aXiv'25) Mem0: Building production-ready ai agents with scalable long-term memory. [paper]
• (aXiv'25) Flex: Continuous agent evolution via forward learning from experience. [paper]
• (NeurIPS'23) Reflexion: language agents with verbal reinforcement learning. [paper]
• (ICLR'24) Think-on-Graph: Deep and Responsible Reasoning of Large Language Model on Knowledge Graph. [paper]
• (TKDE'26) Reliable Reasoning Path: Distilling Effective Guidance for LLM Reasoning With Knowledge Graphs. [paper]
• (ICLR'24) Reasoning on Graphs: Faithful and Interpretable Large Language Model Reasoning. [paper]
• (AAAI'24) Memorybank: Enhancing large language models with long-term memory. [paper]
• (EMNLP'25) Memory OS of AI Agent. [paper]
• (aXiv'23) MemGPT: Towards LLMs as Operating Systems. [paper]
• (ICML'25) From RAG to Memory: Non-Parametric Continual Learning for Large Language Models. [paper]

External Self-Exploration

• (AAAI'24) Expel: Llm agents are experiential learners. [paper]
• (ICRA'24) MATRIX: multi-agent trajectory generation with diverse contexts. [paper]
• (aXiv'25) Memory-r1: Enhancing large language model agents to manage and utilize memories via reinforcement learning. [paper]
• (aXiv'25) Inside-out: Hidden factual knowledge in llms. [paper]
• (aXiv'25) Memevolve: Meta-evolution of agent memory systems. [paper]
• (aXiv'26) Beyond static summarization: Proactive memory extraction for llm agents. [paper]
• (aXiv'25) Agentevolver: Towards efficient self-evolving agent system. [paper]
• (TechReport_Moonshot'25) Kimi K2.5. [paper]

🏆 Benchmarks

💻 Open-source Project

• LangMem: a software development kit (SDK) from LangChain designed to give AI agents long-term memory
• LightMem: Lightweight and efficient memory-augmented generation
• O-Mem: Omni memory system for personalized, long horizon, self-evolving agents
• OpenMemory: A cognitive memory engine for LLMs and agents
• Memori: A system that gives your AI agents a structured, persistent memory layer
• MemMachine: A multi-layered memory system for AI agents
• Memary: A longterm memory for autonomous agents
• Graphiti: A framework for building and querying temporally-aware knowledge graphs, specifically tailored for AI agents operating in dynamic environments
• Memvid: A single-file memory layer for AI agents with instant retrieval and long-term memory
• Anda Hippocampus: A native graph-based memory system for autonomous AI agents, featuring bio-inspired sleep-based memory consolidation and powered by KIP & AndaDB
• Omnigraph: Typed graph database for agent memory. S3-native, Rust, traversal + vector + BM25 in one runtime, Git-style branch/merge.
• taOSmd: Local-first, benchmarked agent memory on an append-only transcript, with a typed temporal knowledge graph where corrected facts supersede old ones via invalidation, plus hybrid vector + BM25 retrieval, tuned for small local models

📃 Citation

@article{yang2026graph,
  title={Graph-based Agent Memory: Taxonomy, Techniques, and Applications},
  author={Yang, Chang and Zhou, Chuang and Xiao, Yilin and Dong, Su and Zhuang, Luyao and Zhang, Yujing and Wang, Zhu and Hong, Zijin and Yuan, Zheng and Xiang, Zhishang and others},
  journal={arXiv preprint arXiv:2602.05665},
  year={2026}
}