用注意力机制检测野外恶意Agent技能，为AI安全提供新思路。
arXiv:2606.23416v1 Announce Type: cross Abstract: LLM agents increasingly load skills, file-based packages of natural-language instructions written by…
从智能体轨迹中提炼可迁移技能，提出一种基于局部经验的知识蒸馏新方法，助力通用智能体学习。
arXiv:2603.25158v5 Announce Type: replace Abstract: Large Language Model (LLM) agents increasingly rely on domain-specific skills, yet manually author…
这篇论文提出Skill-RM框架，通过智能体技能统一异构评估标准，推动AI评估体系革新。
arXiv:2606.03980v1 Announce Type: cross Abstract: Reward models (RMs) provide critical feedback signals for LLM post-training, notably in reinforced f…
多目标优化新方案：Chebyshev退火算法让智能体技能训练更高效，兼顾精度与泛化能力。
arXiv:2605.19330v1 Announce Type: cross Abstract: LLM agents organize behavior through skills - structured natural-language specifications governing h…