RSR: Scaling Trajectories for Complex Tasks through Recursive Self-Rewrite

Zongxia Li‡, Yucheng Shi‡, Zhongzhi Li‡, Junyao Yang, Ruhan Wang, Chengsong Huang, Fuxiao Liu, Haitao Mi, Jordan Boyd-Graber, Leowei Liang

‡ Core contribution (marked † in the source paper)

RSR planner, critic and executor pipeline for rewriting verified trajectories

Recursive Self-Rewrite turns successful trajectories discovered under diverse harnesses into reusable capabilities under a general harness. It expands 2,001 trajectories to 11,094 and improves Qwen-3.8-27B pass@3 from 57.0% to 74.2% on Terminal-Bench 2.

Paper arXiv Model