| title | Terminal-RSI: Recursive Self-Improving Terminal Agents via Recursive Self-Rewrite and Policy-Refine |
|---|---|
| authors | Junyao Yang, Zongxia Li, Yucheng Shi, Zhongzhi Li, Ruhan Wang, Ruiwen Zhou, Haitao Mi, Shuicheng Yan, Leowei Liang |
| date | 2026-10-15 |
| venue | arXiv (preprint, not peer reviewed) |
| id | arXiv:2610.22192 |
| affiliation | Tencent |
| upvotes | 37 (Hugging Face Daily Papers, 2026-11-01) |
| links | abs pdf hf |
| abstract | Converting verifier outcomes into useful policy updates and partial progress into complete task resolution is crucial for terminal agents. We present Terminal-RSI, which alternates Recursive Self-Rewrite (RSR) of verified trajectories with critic-inheriting Recursive Policy-Refine (RPR). We formulate Turn-Boundary Skip-Observation GAE (TB-GAE) to retain token-specific baselines while applying decay across interaction turns. Family Dense Reward (FDR) calibrates semantic progress across task families; Success-Prioritized Reward (SPR) separates strict success from bounded partial credit. A Warm-up Critic Model, recursive critic handoff, and stage-specific learning rates coordinate the two refinement phases. The five-stage sequence from Qwen3.8-27B base model through RSR-1, RPR-1, RSR-2, and RPR-2 yields Terminal-RSI-27B: 19.84% on Terminal-Bench 4.0, 86.52% on Terminal-Bench 2.1, 16.95% on Terminal-Bench 3.0, and 79.00% on Terminal-Bench Hard. On LHTB, process reward rises from 0.21 to 0.37 and strict completion from 0/46 to 5/46; SWR100 success doubles from 3.00% to 6.00%. By recursively coupling verified trajectory rewriting with critic-inheriting policy refinement, each retained stage consolidates discovered behaviors and progressively advances the agent from partial local repairs to strict terminal verification. |
cite
@misc{yang2026terminal,
title={Terminal-RSI: Recursive Self-Improving Terminal Agents via Recursive Self-Rewrite and Policy-Refine},
author={Junyao Yang and Zongxia Li and Yucheng Shi and Zhongzhi Li and Ruhan Wang and Ruiwen Zhou and Haitao Mi and Shuicheng Yan and Leowei Liang},
year={2026},
eprint={2610.22192},
archivePrefix={arXiv},
url={https://arxiv.org/abs/2610.22192}
}