arXiv paper proposes two inference-time designs (FC, PS) to govern persistent memory in LLM agents

KEY POINTS

1. The paper (arXiv:2610.11188v1) frames persistent-memory governance for LLM agents as two decisions—admission (what recalled information enters context) and presentation (how admitted information is expressed)—and implements two inference-time, no-retraining designs: factor-compiled admission (FC) and permission-semantic admission (PS).
2. Evaluated across a four-backbone development suite and an external benchmark, FC and PS reduce pooled judge-assessed failure rates by 6.7 and 8.8 percentage points and cut development-set cross-domain leakage by up to 29.5 percentage points, while also increasing some personalization failures and leaving presentation-design differences statistically insignificant in corrected comparisons.
3. Distinguishing and separately evaluating admission versus presentation enables targeted, inference-time controls that measurably reduce certain safety failures (notably cross-domain leakage) without retraining, though challenges in preserving safe personalization remain.