SKT: Skill-Use Training at Scale via Verified Synthetic Data Generation
Published:
Authors: Zelin Tan, Yiqun Zhang, Hao Li, Zhiyao Cui, Hejia Geng, Shao Zhang, Hangfan Zhang, Yang Chen, Xiaosong Wang, Lilong Wang, Zhenfei Yin, Shuyue Hu, Chen Zhang, and Lei Bai.
SKT is a verified synthesis pipeline for constructing skill-grounded tasks and executable trajectories from large collections of agent skills. Using 2,000 public skills, it produces 4,000 task packages and 27,164 verified trajectories, together with the held-out SkillEval benchmark.