Complex reasoning in tool-augmented agent frameworks is inherently long-horizon, causing reasoning traces and transient tool artifacts to accumulate and strain the bounded working context of large language models. Without explicit memory mechanisms, such accumulation disrupts logical continuity and undermines task alignment. This positions memory not as an auxiliary efficiency concern, but as a core component for sustaining coherent, goal-directed reasoning over long horizons. We propose MemoBrain, an executive memory model for tool-augmented agents that constructs a dependency-aware memory over reasoning steps, capturing salient intermediate states and their logical relations. Operating as a co-pilot alongside the reasoning agent, MemoBrain organizes reasoning progress without blocking execution and actively manages the working context. Specifically, it prunes invalid steps, folds completed sub-trajectories, and preserves a compact, high-salience reasoning backbone under a fixed context budget. Together, these mechanisms enable explicit cognitive control over reasoning trajectories rather than passive context accumulation. We evaluate MemoBrain on challenging long-horizon benchmarks, including GAIA, WebWalker, and BrowseComp-Plus, demonstrating consistent improvements over strong baselines.

翻译：在工具增强的智能体框架中，复杂推理本质上是长视野的，这导致推理轨迹和临时工具产物不断累积，从而对大型语言模型有限的工作上下文容量造成压力。若缺乏显式的记忆机制，此类累积会破坏逻辑连续性并损害任务对齐。这使得记忆不再是一个辅助性的效率问题，而是成为在长视野中维持连贯、目标导向推理的核心组件。我们提出了 MemoBrain，一种用于工具增强智能体的执行记忆模型，它在推理步骤之上构建了一个具备依赖感知能力的记忆，捕获关键的中间状态及其逻辑关系。MemoBrain 作为推理智能体的协同驾驶者运行，在不阻塞执行的情况下组织推理进展，并主动管理工作上下文。具体而言，它在固定的上下文预算下，修剪无效步骤、折叠已完成的子轨迹，并保留一个紧凑、高显著性的推理主干。这些机制共同实现了对推理轨迹的显式认知控制，而非被动的上下文累积。我们在具有挑战性的长视野基准测试（包括 GAIA、WebWalker 和 BrowseComp-Plus）上评估了 MemoBrain，结果表明其相对于强基线模型取得了持续的改进。