MemoBrain: Executive Memory as an Agentic Brain for Reasoning

TL;DR

MemoBrain introduces an executive memory system for tool-augmented agents, enabling coherent long-horizon reasoning by managing and organizing intermediate states and logical relations, thus improving performance on complex benchmarks.

Contribution

This paper presents MemoBrain, a novel memory mechanism that constructs dependency-aware memory for reasoning steps, enhancing logical continuity and task alignment in long-horizon agent reasoning.

Findings

Outperforms strong baselines on GAIA, WebWalker, BrowseComp-Plus benchmarks.

Effectively manages reasoning context by pruning, folding, and preserving salient states.

Demonstrates improved logical coherence and task success rates.

Abstract

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.