MemHarness: Memory Is Reconstructed, Not Replayed

Source

• Source name: arXiv
• Source kind: paper
• Authors / organization: Rong Wu, Daocheng Fu, Licheng Wen, Xuemeng Yang, Shu Zou, Jianbiao Mei, Yuxin Wang, Hairong Zhang, Yu Yang, Tao Hu, Cong Zhang, Botian Shi, Pinlong Cai
• Venue: arXiv: cs.AI
• Published: 2026-07-30
• Collected: 2026-07-31
• URL: http://arxiv.org/abs/2607.28272v1
• PDF: https://arxiv.org/pdf/2607.28272v1
• DOI: UNKNOWN

Plain-Language Brief

Researchers have developed MemHarness, a framework to improve large language model agents by enabling them to adapt and reconstruct past experiences based on the current context, instead of simply replaying stored memories. This approach outperforms static memory-augmented baselines in experiments and shows robustness in out-of-distribution scenarios.

What the Document Says

• The authors propose MemHarness, a framework that equips large language model (LLM) agents to actively harness and reconstruct past experiences based on the present context.
• At each decision step, a unified policy model critiques and reconstructs the retrieved experience conditioned on the current state.
• This reconstructive ability emerges through end-to-end training with GRPO.
• Experiments on ALFWorld and WebShop show that MemHarness outperforms pure RL and static memory-augmented baselines.
• The reconstruction objective prevents negative transfer and serves as latent guidance during training, improving the agent's intrinsic reasoning capabilities.
• The framework demonstrates strong robustness in out-of-distribution (OOD) scenarios.

Key Concepts Mentioned

• MemHarness
• Large Language Model (LLM)
• Reconstructive ability
• End-to-end training
• GRPO
• ALFWorld
• WebShop
• Out-of-distribution (OOD) scenarios

Claims Made by the Document

• MemHarness improves large language model agents.
• The reconstructive ability of MemHarness emerges through end-to-end training with GRPO.
• MemHarness outperforms pure RL and static memory-augmented baselines in experiments.
• The reconstruction objective prevents negative transfer and improves intrinsic reasoning capabilities.

Evidence Used

• The authors present MemHarness as a solution to enhance large language model agents.
• Details about the end-to-end training with GRPO are provided.
• Results from experiments on ALFWorld and WebShop are shown, demonstrating the superiority of MemHarness over other baselines.

Methods / Approach

UNKNOWN

Limitations Stated by the Authors

UNKNOWN

Limitations Noted by the Agent

Agent interpretation — not from the source.

UNKNOWN

Open Questions

UNKNOWN

Notes for Future Q&A

May include interpretation. Labeled where it does.

UNKNOWN

Raw Abstract / Excerpt

Retrieving past experiences has become a common strategy to enhance large language model agents. However, most existing memory-augmented agents treat retrieved experiences as static records to be replayed verbatim, injecting them into the context regardless of whether they align with the agent's current situation. This ``replay'' paradigm ignores the gap between the abstract, general nature of stored experience and the concrete, ever-changing states encountered at decision time, frequently causing negative transfer. In contrast, humans rarely recall past experiences verbatim; instead, they reorganize and adapt retrieved memories to fit the present context. Inspired by this, we propose MemHarness, a framework that equips LLM agents to actively harness and reconstruct past experiences based on the present context. At each decision step, a unified policy model critiques and reconstructs the retrieved experience conditioned on the current state, producing context-grounded guidance before acting. This reconstructive ability emerges naturally through end-to-end training with GRPO. Experiments on ALFWorld and WebShop show that MemHarness substantially outperforms pure RL and static memory-augmented baselines, demonstrating strong robustness in out-of-distribution (OOD) scenarios. Furthermore, our analyses reveal that this reconstruction objective not only prevents negative transfer but also serves as latent guidance during training, fundamentally improving the agent's intrinsic reasoning capabilities.