ACE: Pluggable Adaptive Context Elasticizer across Agents

Source

Source name: arXiv
Source kind: paper
Authors / organization: Ning Liao, Zihao Long, Xiaoxing Wang, Xue Yang, Yaoming Wang, Ziyuan Zhuang, Xunliang Cai, Rongxiang Weng, Junchi Yan
Venue: arXiv: cs.AI
Published: 2026-06-30
Collected: 2026-07-01
URL: http://arxiv.org/abs/2606.31564v1
PDF: https://arxiv.org/pdf/2606.31564v1
DOI: UNKNOWN

Plain-Language Brief

Researchers propose a new module called Adaptive Context Elasticizer (ACE) to address the challenges posed by growing trajectory lengths in large language models (LLMs) with fixed context windows. ACE is designed to be flexible and reversible, storing both raw messages and compressed abstractions for each historical step, and adaptively assigning each step an elastic type based on the current task state. The module improves performance over truncation and summarization baselines across four diverse agent frameworks.

What the Document Says

The paper presents Adaptive Context Elasticizer (ACE), a plug-and-play module designed to manage context for large language model (LLM) based agents.
ACE maintains a lossless message maintenance layer that stores both raw messages and compressed abstractions for each historical step.
A context orchestration layer adaptively assigns each step an elastic type as raw, abstract, or drop at every decision step.
The design ensures the main LLM always receives a compact yet information-rich context.
ACE has been adapted to four diverse agent frameworks: ReAct, DeepAgent, WebThinker, and MiroFlow.
Experiments show that ACE outperforms truncation and summarization baselines and brings consistent performance gains across all four agent frameworks.

Key Concepts Mentioned

Adaptive Context Elasticizer (ACE)
large language models (LLMs)
context management
trajectory lengths
truncation
summarization
ReAct
DeepAgent
WebThinker
MiroFlow

Claims Made by the Document

ACE addresses the limitations of existing context management techniques.
ACE ensures that the main LLM always receives a compact yet information-rich context.
ACE outperforms truncation and summarization baselines.

Evidence Used

The authors describe ACE's design and its benefits in addressing the limitations of existing context management techniques.
Experimental results showing that ACE outperforms truncation and summarization baselines are provided.

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

The increasing complexity of agentic tasks has led to rapidly growing trajectory lengths, which poses significant challenges for large language model (LLM) based agents with fixed context windows. Existing context management techniques, such as truncation and summarization, suffer from inherent inflexibility and irreversibility: once information is discarded or compressed, it cannot be recovered even when it becomes critically relevant in later decision steps. To address these limitations, we propose the Adaptive Context Elasticizer (ACE), a plug-and-play module that elastically orchestrates historical step information into the agent's context at each decision step. ACE maintains a lossless message maintenance layer that stores both raw messages and compressed abstractions for each historical step, while a context orchestration layer adaptively assigns each step an elastic type as raw, abstract, or drop, at every decision step based on the current task state. This reversible design ensures that the main LLM always receives a compact yet information-rich context. We adapt ACE to four diverse agent frameworks, including ReAct, DeepAgent, WebThinker, and MiroFlow, without training or architectural modifications. Experiments show that ACE consistently outperforms truncation and summarization baselines, and brings consistent performance gains across all four agent frameworks.