LLM-based recommendation agents personalize what they know through evolving per-user semantic memory, yet how they perform reasoning relies on a universal, static system prompt shared across all users. When a recommendation fails, the agent only updates its memory, but never examines or revises the reasoning that leads to the failure. To close this gap, we propose SAGER (Self-Evolving Agent for Personalized Recommendation), the first framework to equip each user with a dedicated, continuously evolving policy skill. SAGER incorporates three tailored mechanisms for recommendation scenarios: (i) a two-representation skill architecture constrained by a Cognitive Injection Boundary, decoupling a rich evolution substrate from a minimal inference-time injection; (ii) a contrastive differential chain-of-thought engine that diagnoses ranking errors against unchosen near-misses under a refine-not-replace principle; and (iii) skill-augmented listwise reasoning that mitigates the Score-Tie Bottleneck, enabling the slim skill to serve as an active tie-breaker. Evaluated on four public benchmarks, SAGER achieves state-of-the-art accuracy. Furthermore, plugging its policy skill layer into three structurally distinct host agents delivers consistent 9%–12% gains on the H@1 metric. These results indicate that policy skill evolution provides architecture-agnostic improvements that are largely complementary to memory accumulation.

Personalized news headline generation has emerged as a promising avenue for optimizing user engagement and satisfaction in online news consumption. However, prevailing methods predominantly adopt an indirect modeling paradigm, inferring user preferences from related tasks rather than capturing their direct headline preferences. Direct modeling remains underexplored due to two primary challenges: (1) the scarcity of large-scale personalized headline datasets and (2) the risk of factual inaccuracies when models overly cater to subjective user preferences. To address these challenges, we propose USG, a novel multi-agent user simulator headline generation framework. USG introduces three specialized agents: the User provides subjective feedback to guide headline personalization; the Editor enforces editorial standards (i.e., factual consistency and no clickbait); and the Writer iteratively refines candidates via a Critique-and-Refine algorithm guided by feedback from the User and Editor agents. The multi-agent interactions yield optimal personalized headlines and reasoning chains distilled from the interaction trajectories, forming a high-quality dataset. This dataset directly supervises the fine-tuning of a standalone generator. Extensive experiments demonstrate that USG outperforms the strongest baseline by nearly 6% on personalization, achieving state-of-the-art performance on both personalization and factual consistency, while enabling efficient end-to-end inference. Our code is available at https://github.com/Leah-HKBU/USG.

The evolution of recommender systems has shifted from traditional collaborative filtering to LLM-based agentic systems, which rely on semantic user and item memories to make predictions. However, existing agents maintain these memories in isolation. This overlooks crucial collaborative signals, such as user-item co-engagements and peer relationships across the community, which significantly limits their ability to uncover hidden preferences and accurately infer user needs, particularly for data-sparse users. To bridge this gap, we introduce collaborative memory, a paradigm that connects isolated semantics to enable the sharing of relational insights. Yet, naively utilizing collaborative memory causes severe context overload and introduces noise to downstream LLMs, alongside prohibitive computational costs. To resolve this, we propose MemRec, a framework that architecturally decouples memory management from reasoning. MemRec introduces a dedicated, lightweight language model LM_Mem to efficiently manage and synthesize a dynamic collaborative memory graph in the background. It provides only distilled, high-signal contexts to a downstream, heavyweight large language model (LLM_Rec) for the final recommendation. Extensive experiments on four benchmarks demonstrate that MemRec achieves state-of-the-art performance. Code: https://github.com/rutgerswiselab/memrecHomepage: https://memrec.weixinchen.com