[Resource Suggestion] Adding MAGMA: A Multi-Graph based Agentic Memory Architecture

[FredJiang0324]

Hi DEEP-PolyU Team,

Thank you for your incredible work on the survey "Graph-based Agent Memory: Taxonomy, Techniques, and Applications". This repository is a vital resource for the community as we move toward more structured and efficient agent memory systems.

I would like to suggest adding our recent work, MAGMA, to your Awesome-GraphMemory list. Our architecture aligns perfectly with your taxonomy, particularly in the Hybrid Graph Architectures and Memory Evolution sections.

Paper Title: MAGMA: A Multi-Graph based Agentic Memory Architecture for AI Agents
arXiv: https://arxiv.org/abs/2601.03236
GitHub: https://github.com/FredJiang0324/MAGMA

Alignment with your Taxonomy
Memory Storage (Hybrid/Multi-Graph): MAGMA utilizes a multi-graph structure (Entity, Temporal, and Topical) to represent agent experiences.
Memory Retrieval: It supports query-aware, multi-hop reasoning over the graph to refine retrieved raw chunks into query-focused memory.
Memory Evolution: MAGMA features proactive memory extraction and structured consolidation of interaction history.

We believe MAGMA provides a concrete implementation of the "Graph-based perspective" discussed in your survey. It would be an honor to have it included in your repository.

Thank you again for your contribution to the field!

[Qing145]

Dear Fred,

Thanks for your great work. I've added it to the paper list. BTW, we will be grateful if you could cite our survey in your research.

@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}
}

[FredJiang0324]

Thank you very much for adding our paper to the list — we truly appreciate it.

Our MAGMA work was completed in December last year. If you happen to revise or update your manuscript in the future, we would be very honored if you consider citing it where appropriate.

We also want to thank you for your survey it provides a very clear and structured taxonomy for the community. We are currently working on an ongoing system-oriented analysis survey in the agent memory space, and we have already included your survey in our references. We expect to finish it in the coming weeks.

Looking forward to more interactions in this area!

Dear Fred,

Thanks for your great work. I've added it to the paper list. BTW, we will be grateful if you could cite our survey in your research.

@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}
}