Remember When It Matters: Proactive Memory Agent for Long-Horizon Agents
In long-horizon tasks, decision-relevant state is often scattered across an expanding trajectory, while the action agent must surface it and act. As trajectories grow, task requirements, environment facts, prior attempts, diagnoses, and open subgoals can be buried in the context window or pushed beyond it, failing to influence decisions when needed. We call this failure mode "behavioral state decay". We study memory as an active intervention mechanism rather than passive retrieval. A separate memory agent runs alongside an unmodified action agent, updating a structured memory bank from the rec
Lineage graph
Paper → model → repo connections mined from source citations (Tier-1 exact match).
Why these links exist
- Linked via arxiv authorYifan Wu →
Remember When It Matters: Proactive Memory Agent for Long-Horizon Agents
- Linked via arxiv authorLizhu Zhang →
Remember When It Matters: Proactive Memory Agent for Long-Horizon Agents
- Linked via arxiv authorYuhang Zhou →
Remember When It Matters: Proactive Memory Agent for Long-Horizon Agents
- Linked via arxiv authorMingyi Wang →
Remember When It Matters: Proactive Memory Agent for Long-Horizon Agents
- Linked via arxiv authorBo Peng →
Remember When It Matters: Proactive Memory Agent for Long-Horizon Agents
- Linked via arxiv authorSerena Li →
Remember When It Matters: Proactive Memory Agent for Long-Horizon Agents
- Linked via arxiv authorXiangjun Fan →
Remember When It Matters: Proactive Memory Agent for Long-Horizon Agents
- Linked via arxiv authorZhuokai Zhao →
Remember When It Matters: Proactive Memory Agent for Long-Horizon Agents