The paper introduces RECAP, a framework for deciding when to retain, compress, or discard parts of a long-horizon agent’s history to improve future performance. RECAP uses matched context interventions, execution-relational prediction, and independent policy certification to evaluate the value of a failure block relative to the current execution. Experiments across five backbones show that RECAP can edit a significant portion of retained checkpoints while maintaining high continuation success and reducing prompt costs.
Software engineering (SWE) agents resolve repository-level issues through long trajectories that grow increasingly expensive as context accumulates. Failed runs tend to be longer and exhibit redundant exploration or looping, suggesting that some failures may be detectable before completion.
The paper introduces a diagnostic framework for long‑horizon security LLM agents that uses checkpoints to distinguish failures occurring before and after a model’s capability is exposed, and applies controlled interventions to pinpoint upstream bottlenecks. The methodology is tested on four task families—delayed reuse of discovered information, reuse of observed state, recovery from failed strategies, and decision making after uncertain outcomes—revealing that many failures happen before the agent observes the state it later needs to reuse. Experiments with Gemini 2.5 Flash and Gemini 3.7 Flash show that targeted protocol‑disambiguation guidance can significantly alter state observation rates and that the primary source of failure can shift across model generations, underscoring the need for fine‑grained failure diagnostics rather than relying solely on overall task success.
arXiv:2609. 04875v1 Announce Type: cross Abstract: Long-running LLM agents are stateful: beyond the transcript they accrete compressed summaries, plaintext memory, pending tool plans, and, under every serving API, a KV cache.
The paper investigates why constrained updates in supervised fine‑tuning (SFT) still lead to forgetting by developing a functional‑path analysis that decomposes changes in old loss into controlled direction, gradient staleness, probe mismatch, and finite‑step effects. It introduces a conditional refresh budget and identifies a restoring direction provided by informative replay, showing that the placement of constraints relative to Adam and the granularity of constraints significantly affect retention across code‑label audits. The study also explores checkpointed generation, replay schedules, and sequential‑task comparisons, demonstrating that subspace regularization, projection, and supervised replay differentially influence backward transfer and content retention.
The paper introduces an admission‑audit protocol for continual embodied agents, arguing that update admission should balance error control with retained learning opportunities within a fixed interaction budget. It critiques a range‑based confidence gate for failing to certify unchanged old‑task behavior, and proposes a paired‑binomial construction that reduces this burden when outcome disagreements are rare. Experiments on a one‑step pushing diagnostic show that fresh paired checks admit a significant portion of updates while the range‑based gate admits none, and a learned‑dynamics stress test helps distinguish model bias from feedback‑selection error.
The paper investigates how personalized agents decide to use, ignore, update, or query retrieved user memory before acting on a task. An empirical audit protocol is developed to test structured intermediate outputs, revealing that while exposing state definitions improves accuracy, an explicit state-output field does not significantly enhance policy accuracy for large language models. The study also shows that example-level accuracy overstates consistency, with full four‑way family success being rare, and that providing benchmark‑associated state labels merely conditions predictions rather than proving internal fidelity.
The paper investigates how memory systems can answer a current query correctly yet fail to retain distinctions needed for later updates. Using a paired‑history audit, the authors evaluate 24 history pairs across six synthetic mechanisms and two model backends, achieving perfect reveal accuracy on DeepSeek and high accuracy on GLM. Record‑level audits reveal specific failures in structured reveal memories and frontier late‑reference adequacy, and the authors test a label‑equivariant repair that only partially restores correctness.
arXiv:2608. 03222v1 Announce Type: cross Abstract: Software engineering (SWE) agents resolve repository-level issues through long trajectories that grow increasingly expensive as context accumulates.
Agent-Controlled Forgetting for Tool-Using Agents: Reversible Context Curation in Practice explores a method where an agent selectively replaces detailed tool results with concise notes while archiving the originals for recovery. Using a Python harness, the approach reduces prompt tokens from 912,492 to 231,951 and cuts cumulative input tokens by 50%, yielding significant API cost savings. However, the method increases request count, slightly slows execution, and its effectiveness varies across different workloads.
arXiv:2607. 21635v1 Announce Type: new Abstract: Personal agents maintain memories, learned skills, tool configurations, and policy state that evolve with each user.
arXiv:2609.36526v1 Announce Type: cross Abstract: Long-horizon agents require context compression to manage growing interaction histories. Compression quality, however, is ultimately determined by do...
The paper introduces ReTRAIL, a framework that allows long‑horizon agents to rewind the environment to earlier states while preserving forward progress in knowledge. By training a shared policy with State‑Anchored GRPO and a branch‑advancement reward, ReTRAIL enables agents to recover from unproductive states and reuse insights from abandoned exploration. Experiments on MineSweeper, Sokoban, ALFWorld, and ScienceWorld show that this combined rewinding and experience‑reuse strategy outperforms simple rewind primitives and improves long‑horizon task solving.