Escaping the Context Bottleneck: Active Context Curation for LLM Agents via Reinforcement Learning

• URL: http://arxiv.org/abs/2604.11462v1
• Date: Mon, 13 Apr 2026 13:39:17 GMT
• Title: Escaping the Context Bottleneck: Active Context Curation for LLM Agents via Reinforcement Learning
• Abstract summary: We introduce a symbiotic framework that decouples context management from task execution.<n>Our architecture pairs a lightweight, specialized policy model, ContextCurator, with a powerful frozen foundation model, TaskExecutor.<n>On WebArena, our framework improves the success rate of Gemini-3.0-flash from 36.4% to 41.2% while reducing token consumption by 8.8%.<n>On DeepSearch, it achieves a 57.1% success rate, compared with 53.9%, while reducing token consumption by a factor of 8.
• Score: 6.577332846728808
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Large Language Models (LLMs) struggle with long-horizon tasks due to the "context bottleneck" and the "lost-in-the-middle" phenomenon, where accumulated noise from verbose environments degrades reasoning over multi-turn interactions. To address this issue, we introduce a symbiotic framework that decouples context management from task execution. Our architecture pairs a lightweight, specialized policy model, ContextCurator, with a powerful frozen foundation model, TaskExecutor. Trained via reinforcement learning, ContextCurator actively reduces information entropy in the working memory. It aggressively prunes environmental noise while preserving reasoning anchors, that is, sparse data points that are critical for future deductions. On WebArena, our framework improves the success rate of Gemini-3.0-flash from 36.4% to 41.2% while reducing token consumption by 8.8% (from 47.4K to 43.3K). On DeepSearch, it achieves a 57.1% success rate, compared with 53.9%, while reducing token consumption by a factor of 8. Remarkably, a 7B ContextCurator matches the context management performance of GPT-4o, providing a scalable and computationally efficient paradigm for autonomous long-horizon agents.

Related papers

• UI-Mate: Advancing Open-Weight Foundation GUI Agents with In-Context Demonstrations [68.29650478987561]
  UI-Mate is a foundation GUI agent that integrates an environment-grounded training stack with in-context demonstration learning.<n>It sets a new open-weight state of the art on general computer-use benchmarks, scoring 77.0% on OSWorld and 66.2% on WindowsAgentArena.
  arXiv  Detail & Related papers  (2026-08-16T20:59:59Z)
• Context Contamination in LLM Analysis of Network Security Logs: Poison with Passive Prompt Injection and Mitigation Evaluation [8.339789704552706]
  Large Language Models are increasingly deployed in Security Operations Centers for log analysis tasks.<n>We demonstrate that adversaries can embed prompt injection payloads in log-generating fields that persist in storage.<n>We present LogInject, a systematic framework for evaluating these threats.
  arXiv  Detail & Related papers  (2026-07-16T02:15:35Z)
• LocalNav: Distilling Frontier VLMs and Embodied RL for On-Device Object Goal Navigation [53.88286427894783]
  Vision Language Models (VLMs) have emerged in the robotic domain as a powerful tool that enables environmental perception with language context.<n>Yet, their computational footprint typically confines them to cloud execution, hindering low-latency inference with local deployment on resource-constrained robots.<n>We present a distillation strategy that transfers complex spatial-semantic reasoning from large frontier models into a lightweight, 4B- parameter local VLM for edge execution on embedded GPU devices.
  arXiv  Detail & Related papers  (2026-06-26T09:11:59Z)
• DEFENGRAPH: Knowledge Graph-Enhanced LLMs for Blue Team Cyber Defense [10.619506659651456]
  Large Language Models (LLMs) show promise for supporting decision-making in cybersecurity, but their reliability in high-stakes, time-evolving environments remains limited.<n>We introduce DEFENGRAPH, an LLM-driven assistant designed to support human defenders during cybersecurity incidents.
  arXiv  Detail & Related papers  (2026-06-19T03:10:01Z)
• TokenMizer: Graph-Structured Session Memory for Long-Horizon LLM Context Management [0.0]
  We present TokenMizer, an open-source proxy system that models large language model sessions as a typed knowledge graph.<n>It produces resume blocks averaging 78 tokens (range: 42-124) - 2x smaller than evaluated baselines.<n>Across all sessions, TokenMizer mean task recall 51.0%, decision recall 46.6%, and file recall 58.7%.
  arXiv  Detail & Related papers  (2026-06-04T16:12:28Z)
• LongSeeker: Elastic Context Orchestration for Long-Horizon Search Agents [63.75550089045995]
  Long-horizon search agents must manage a rapidly growing working context as they reason, call tools, and observe information.<n>We propose that effective context management should be adaptive: parts of the agent's trajectory are maintained at different levels of detail depending on their current relevance to the task.<n>We develop LongSeeker, a long-horizon search agent fine-tuned from Qwen3-30B-A3B on 10k synthesized trajectories.
  arXiv  Detail & Related papers  (2026-05-06T17:54:16Z)
• LightThinker++: From Reasoning Compression to Memory Management [61.2260619973687]
  Large language models (LLMs) excel at complex reasoning, yet their efficiency is limited by the surging cognitive overhead of long thought traces.<n>We propose LightThinker, a method that enables LLMs to dynamically compress intermediate thoughts into compact semantic representations.<n>We evolve the framework into LightThinker++, introducing Explicit Adaptive Memory Management.
  arXiv  Detail & Related papers  (2026-04-04T10:46:09Z)
• GLANCE: Gaze-Led Attention Network for Compressed Edge-inference [10.229095428511654]
  Real-time object detection in AR/VR systems faces critical computational constraints, requiring sub-10,ms latency within tight power budgets.<n>Inspired by biological foveal vision, we propose a two-stage pipeline that combines differentiable weightless neural networks for ultra-efficient gaze estimation with attention-guided region-of-interest object detection.
  arXiv  Detail & Related papers  (2026-03-16T15:52:52Z)
• CoT4Det: A Chain-of-Thought Framework for Perception-Oriented Vision-Language Tasks [53.88194225946438]
  Chain-of-Thought for Detection (CoT4Det) is a simple but efficient strategy that reformulates perception tasks into three interpretable steps.<n>We show that CoT4Det significantly improves perception performance without compromising general vision language capabilities.
  arXiv  Detail & Related papers  (2025-12-07T05:26:30Z)
• Revisiting Long-context Modeling from Context Denoising Perspective [47.73358822870211]
  Long-context models (LCMs) can locate implicit critical information within the context for further prediction.<n>Recent research reveals that LCMs are often susceptible to contextual noise, i.e., irrelevant tokens, that can mislead model attention.<n>We propose an effective metric, the Integrated Gradient (IG) score, to detect and quantify the noise information within the context.
  arXiv  Detail & Related papers  (2025-10-07T12:32:23Z)
• An Automated Attack Investigation Approach Leveraging Threat-Knowledge-Augmented Large Language Models [17.220143037047627]
  Advanced Persistent Threats (APTs) compromise high-value systems to steal data or disrupt operations.<n>Existing methods suffer from poor platform generality, limited generalization to evolving tactics, and an inability to produce analyst-ready reports.<n>We present an LLM-empowered attack investigation framework augmented with a dynamically adaptable Kill-Chain-aligned threat knowledge base.
  arXiv  Detail & Related papers  (2025-09-01T08:57:01Z)
• Overflow Prevention Enhances Long-Context Recurrent LLMs [81.71585057993074]
  A recent trend in LLMs is developing recurrent sub-quadratic models that improve long-context processing efficiency.<n>We investigate leading large long-context models, focusing on how their fixed-size recurrent memory affects their performance.<n>Our experiments reveal that, even when these models are trained for extended contexts, their use of long contexts remains underutilized.
  arXiv  Detail & Related papers  (2025-05-12T17:45:05Z)
• Benchmarking Reasoning Robustness in Large Language Models [76.79744000300363]
  We find significant performance degradation on novel or incomplete data.<n>These findings highlight the reliance on recall over rigorous logical inference.<n>This paper introduces a novel benchmark, termed as Math-RoB, that exploits hallucinations triggered by missing information to expose reasoning gaps.
  arXiv  Detail & Related papers  (2025-03-06T15:36:06Z)
• INTACT: Inducing Noise Tolerance through Adversarial Curriculum Training for LiDAR-based Safety-Critical Perception and Autonomy [0.4124847249415279]
  We present a novel framework designed to enhance the robustness of deep neural networks (DNNs) against noisy LiDAR data.<n>IntACT combines meta-learning with adversarial curriculum training (ACT) to address challenges posed by data corruption and sparsity in 3D point clouds.<n>IntACT's effectiveness is demonstrated through comprehensive evaluations on object detection, tracking, and classification benchmarks.
  arXiv  Detail & Related papers  (2025-02-04T00:02:16Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.