Neura Market
Agent-Memory-Paper-List logo

Agent-Memory-Paper-List

Free

Curated papers on memory architectures for LLM agents — long-term, short-term, attention mechanisms (2026) ![](https://img.shields.io/github/stars/Shichun-Liu/Agent-Memory-Paper-List?style=flat-square)

FreeFree tier
Type
Open Source
Visit Website GitHub (40 stars)

About Agent-Memory-Paper-List

Agent-Memory-Paper-List is a curated repository of research papers on memory architectures for foundation model-based agents. It provides a comprehensive survey—'Memory in the Age of AI Agents'—that organizes agent memory through three unified lenses: Forms (token-level, parametric, latent), Functions (factual, experiential, working memory), and Dynamics (formation, evolution, retrieval). The repository distinguishes agent memory from related concepts like RAG and Context Engineering, and includes an up-to-date paper list covering topics such as event-centric memory, multimodal memory, scalable agentic memory, and more. It is actively maintained with community contributions and has garnered over 2,300 stars on GitHub.

Key Features

Curated paper list on agent memory for AI agents
Unified taxonomy: Forms (token-level, parametric, latent), Functions (factual, experiential, working memory), Dynamics (formation, evolution, retrieval)
Distinguishes agent memory from RAG and Context Engineering
Includes survey paper: 'Memory in the Age of AI Agents'
Regularly updated with recent works (e.g., 2025-2026 papers)
Open source repository with community contributions and stars

Pros & Cons

Pros
  • Comprehensive and organized taxonomy covering multiple dimensions
  • Actively maintained with frequent updates and community engagement
  • Open access and free to use
  • Distinguishes agent memory from overlapping concepts like RAG
  • Includes direct links to papers for easy access
Cons
  • Not a tool or library, only a curated list of references
  • Requires background knowledge in AI agents and memory systems
  • Limited to research survey, no implementation or code provided

Best For

Researching memory architectures for LLM agentsSurvey reference for academia and industryUnderstanding taxonomy and distinctions in agent memoryStaying updated on latest memory-related papers in AI agents

FAQ

What is agent memory?
Agent memory refers to the storage and retrieval mechanisms that enable foundation model-based agents to perform long-horizon reasoning, adapt continually, and interact effectively with complex environments. The survey categorizes it into forms (token-level, parametric, latent), functions (factual, experiential, working memory), and dynamics (formation, evolution, retrieval).
How is agent memory different from RAG?
The repository explicitly distinguishes agent memory from Retrieval-Augmented Generation (RAG) and Context Engineering. While RAG focuses on external knowledge retrieval, agent memory encompasses broader storage modalities (parametric, latent) and functions like experiential and working memory.
Is this repository actively maintained?
Yes, the repository has recent updates (as of 2026) and community contributions, with news including reaching 1,000 stars and updates to the survey paper.
Can I contribute papers to this list?
The repository is open source and welcomes contributions. The news section thanks the community for their valuable contributions and suggestions.