5arXiv cs.CL (Computation and Language)·
MemTools: Unified interoperability framework for agent memory system research
MemTools is a new research framework that decouples agent memory system components from deployment environments, addressing fragmentation in existing memory architectures. It standardizes the memory lifecycle via declarative data contracts, separates benchmark datasets from execution protocols, and provides a unified interface for symbolic, neural, and multimodal memory representations. Empirical evaluations demonstrate the framework enables systematic isolation of memory design variables across heterogeneous systems. The work targets a recognized gap in agent infrastructure research where tight coupling has impeded principled comparison and composition.
Related guides (2)
Related events (8)
Systematic evaluation of 12 agent memory systems from a data management perspective
A new arXiv preprint proposes an analytical framework decomposing agent memory into four core modules—representation/storage, extraction, retrieval/routing, and maintenance—and evaluates 12 representative memory systems across five benchmark workloads spanning 11 datasets. The study finds no single architecture dominates across scenarios; effectiveness depends on alignment between memory structure and workload bottleneck. Fine-grained ablation studies quantify effects on retrieval precision, update correctness, and long-horizon stability, and reveal that localized maintenance is more cost-efficient than global reorganization. Code is publicly released.
UniMem: Self-routing framework for complementary episodic-to-parametric memory in LLM agents
UniMem is a proposed framework for autonomous memory management in LLM agents that addresses the stability-plasticity dilemma in boundary-agnostic task streams. The system uses learnable routing tokens to coordinate between an episodic retrieval buffer (for novel/sparse tasks) and expandable parametric memory blocks (for recurring patterns), inspired by human memory consolidation. Experiments on long-horizon streaming task sequences show an average 4.0 EM point gain over baselines across three backbone models. The work is relevant to continual learning and agent memory architecture research.
FluxMem: Connectivity-Evolving Memory Framework for LLM Agents
FluxMem proposes a heterogeneous graph-based memory framework for LLM agents that continuously evolves its topology through three stages: initial connection formation, feedback-driven refinement, and long-term consolidation. Unlike static memory repositories, FluxMem repairs missing links, prunes interference, aligns abstraction granularity, and distills successful trajectories into reusable procedural circuits. The system is guided by a single metric for memory generalizability and evolutionary maturity, achieving state-of-the-art results on LoCoMo, Mind2Web, and GAIA benchmarks.
MemCon: Learned adaptive memory management for LLM agents via MDP-based policy
Researchers introduce Memory as a Controlled Process (MemCon), a framework that models LLM agent memory operations as a Markov Decision Process and learns an online policy for adaptive retrieval, plan injection, and memory consolidation. The system is backend-agnostic, wraps existing memory implementations, and uses a lightweight tabular contextual bandit with UCB exploration requiring no pretraining and no additional LLM calls. Evaluated across 6 benchmarks, 3 agent frameworks, and 3 LLM backbones, MemCon outperforms static memory baselines by up to 15.2 points in task success while reducing token consumption by 5–20%.
MemOS: Self-Evolving Memory OS for LLM Agents with Hybrid Retrieval and Token Savings
MemOS is an open-source TypeScript project providing a memory operating system layer for LLM and AI agents, featuring ultra-persistent memory, hybrid retrieval, and cross-task skill reuse. The project claims 35.24% token savings through its memory management approach. It has accumulated 9,329 GitHub stars with moderate daily momentum (+67). The system targets agent memory persistence and efficiency as a foundational infrastructure component.
AutoMem: Automated framework trains LLMs to manage memory as a learnable cognitive skill
AutoMem is a new framework that treats memory management in LLMs as a trainable skill, using two optimization loops: one that iteratively revises memory structure via trajectory review by a strong LLM, and one that distills good memory decisions into direct training signal for the agent model. Evaluated on three long-horizon procedurally generated games (Crafter, MiniHack, NetHack), optimizing memory alone yielded 2x-4x performance improvements, bringing a 32B open-weight model competitive with frontier systems like Claude Opus 4.5 and Gemini 3.1 Pro Thinking. The work draws on cognitive science concepts of metamemory and demonstrates that memory management is an independently learnable, high-leverage capability for long-horizon agentic tasks.
MemLens: Value-aware memory management system with Shapley-style evaluation for LLM agents
MemLens is a proposed memory management system for LLM-based agents that treats memory records as first-class data objects with explicit value scoring, contrasting with existing systems that handle all interaction records uniformly. The system uses Shapley-style evaluation to assess memory utility and provides an interactive analytics dashboard exposing the full memory lifecycle including storage, retrieval, and response generation. A study-copilot application demonstrates the system's ability to compare memory strategies across response quality, retrieval latency, and token consumption metrics.
agentmemory: Persistent Memory for AI Coding Agents
agentmemory is an open-source TypeScript library providing persistent memory for AI coding agents, designed based on real-world benchmarks. The repository has accumulated 13,772 total stars with a notable single-day gain of 1,626 stars, indicating strong community traction. It targets the agent tool ecosystem by addressing memory continuity across coding agent sessions.