Remember When It Matters: Proactive Memory Agent for Long-Horizon Agents
Abstract
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 recent trajectory and deciding whether to inject a memory-grounded reminder or remain silent. The module is plug-and-play with frontier action agents and existing agent harnesses. Across Terminal-Bench 2.0 and τ^2-Bench, it improves pass@1 for both weaker and stronger action agents, with gains of +8.3 pp on Terminal-Bench and +6.8 pp on τ^2-Bench. Ablations show that selective intervention outperforms passive bank exposure, always-on injection, advisor-only guidance, and general retrieval. As an early step toward open-weight memory policies, we train Qwen3.5-27B on SETA using SFT and GRPO, improving validation reward and achieving partial transfer to Terminal-Bench.
Community
This is an automated message from the Librarian Bot. I found the following papers similar to this paper.
The following papers were recommended by the Semantic Scholar API
- SelfMem: Self-Optimizing Memory for AI Agents (2026)
- AdMem: Advanced Memory for Task-solving Agents (2026)
- EvoMemBench: Benchmarking Agent Memory from a Self-Evolving Perspective (2026)
- ECHO: Prune to act, trace to learn with selective turn memory in agentic RL (2026)
- Learning Agent-Compatible Context Management for Long-Horizon Tasks (2026)
- Learning to Retrieve: Dual-Level Long-Term Memory for Text-to-SQL Agents (2026)
- OPD-Evolver: Cultivating Holistic Agent Evolver via On-Policy Distillation (2026)
Please give a thumbs up to this comment if you found it helpful!
If you want recommendations for any Paper on Hugging Face checkout this Space
You can directly ask Librarian Bot for paper recommendations by tagging it in a comment: @librarian-bot recommend
Anyone have the repo for this project? Github repo the paper links is empty.
What surprised me most about this paper is how obvious the problem is once you name it. "Behavioral state decay" — the fact that an agent's own past decisions, diagnoses, and open subgoals drift out of effective context as the trajectory grows — is something I've definitely felt debugging long-running agents but never had a clean term for. The architecture is refreshingly practical: a separate memory agent that runs alongside the action agent, writing structured updates into a bank the action agent can query. No retraining, no prompt engineering tricks, just a dedicated process for surfacing what matters when it matters. I'd want to see how this holds up when the memory agent itself starts accumulating its own long context — does it need a memory agent for the memory agent? But for the common case of agents that lose the plot after 50+ turns, this feels like the right kind of fix.
Get this paper in your agent:
hf papers read 2607.08716 Don't have the latest CLI?
curl -LsSf https://hf.co/cli/install.sh | bash Models citing this paper 0
No model linking this paper
Datasets citing this paper 1
Spaces citing this paper 0
No Space linking this paper
Collections including this paper 0
No Collection including this paper