AutoSkill: Experience-Driven Lifelong Learning via Skill Self-Evolution

In practical LLM applications, users repeatedly express stable preferences and requirements---such as reducing hallucinations, following institutional writing conventions, or avoiding overly technical wording---yet such interaction experience is seldom consolidated into reusable knowledge. As a result, LLM agents often fail to accumulate personalized capabilities across sessions. We present AutoSkill, an experience-driven lifelong learning framework that enables LLM agents to automatically derive, maintain, and reuse skills from dialogue and interaction traces. AutoSkill abstracts reusable skills from user experience, supports their continual self-evolution, and dynamically injects relevant skills into future requests without retraining the underlying model. Designed as a model-agnostic plug-in layer, it remains compatible with existing LLMs while introducing a standardized skill representation for sharing and transfer across agents, users, and tasks. By turning ephemeral interaction experience into explicit, reusable, and composable capabilities, AutoSkill provides a practical path toward lifelong personalized agents and personal digital surrogates.