yuxin-kang

yuxin-kang/yuxin-agent-skills

Skill Claude CodeCodex

Record or update a Leo Lab experiment after the user has Play-tested a policy and supplied the actual result plus a run, job, checkpoint, W&B, or experiment identifier. Always compare the current experiment with its prior baseline, including reward/training parameters, environment I/O, and network architecture, then…

yuxin-kang/yuxin-agent-skills

Skill Claude CodeCodex

Find, screen, deduplicate, and summarize newly released robotics papers for daily or ad-hoc research digests. Use when asked for the latest robotics, humanoid, legged locomotion, loco-manipulation, robot learning, perception, sim-to-real, whole-body control, or safety papers; when maintaining a recurring paper watch…