MENTOR: Mitigating Identity Drift in Dynamic Role-Playing via Dual-Chain Structured Memory

Long-context LLM agents increasingly serve multiple users or personas within a single session, requiring stable identity and knowledge boundaries under frequent switching. We identify a common failure mode, identity drift, where models conflate user-specific states and leak information across roles. On BEAM-SWITCH, a benchmark for controlled multiuser switching, performance consistently degrades as switching intensifies, even when responses remain fluent and locally coherent. We propose MENTOR, a cognitive architecture that mitigates identity drift without finetuning.MENTOR uses a Dual-Chain Memory Mechanism: a Global Chain (G) for long-term event logging and isolated Role Chains (R r ) as per-role working memories, supported by a semantic Knowledge Graph (K) that filters and verifies role-admissible information before generation.Across six LLM families, MENTOR improves the overall score (Avg) from 0.46 to 0.75 on average (+0.29 absolute), with substantial gains in identity adherence and knowledge fidelity.