Existing methods for AI psychological counselors predominantly rely on supervised fine-tuning using static dialogue datasets. However, this contrasts with human experts, who continuously refine their proficiency through clinical practice and accumulated experience. To bridge this gap, we propose an Experience-Driven Lifelong Learning Agent (\texttt{PsychAgent}) for psychological counseling. First, we establish a Memory-Augmented Planning Engine tailored for longitudinal multi-session interactions, which ensures therapeutic continuity through persistent memory and strategic planning. Second, to support self-evolution, we design a Skill Evolution Engine that extracts new practice-grounded skills from historical counseling trajectories. Finally, we introduce a Reinforced Internalization Engine that integrates the evolved skills into the model via rejection fine-tuning, aiming to improve performance across diverse scenarios. Comparative analysis shows that our approach achieves higher scores than strong general LLMs (e.g., GPT-5.4, Gemini-3) and domain-specific baselines across all reported evaluation dimensions. These results suggest that lifelong learning can improve the consistency and overall quality of multi-session counseling responses.

翻译：现有AI心理咨询方法主要依赖静态对话数据集进行监督微调，但这与人类专家通过临床实践和积累经验持续提升专业能力的特质存在显著差异。为弥合这一差距，我们提出了一种面向心理咨询的经验驱动终身学习智能体（PsychAgent）。首先，我们构建了专为纵向多轮交互设计的记忆增强规划引擎，通过持久化记忆与策略性规划确保治疗连续性。其次，为支持自主进化，我们设计了技能进化引擎，可从历史咨询轨迹中提取新型实践导向技能。最后，我们引入强化内化引擎，通过拒绝微调将进化后的技能整合至模型中，旨在提升跨场景的响应质量。对比分析表明，本方法在所有评估维度上均优于强通用大语言模型（如GPT-5.4、Gemini-3）及领域专属基线模型。研究结果证实，终身学习机制能有效提升多轮咨询中响应的一致性与整体质量。