Terminal Agents

Follow this topic Ask about this topic

Terminal agents use a shell, filesystem, and command-line tools to inspect and change digital environments, often across long workflows that require planning, debugging, and recovery. Common training approaches include supervised fine-tuning on task trajectories, reinforcement learning with executable verifiers, and synthetic generation of tasks and environments; some methods also teach agents to predict how environments respond to actions. Benchmarks such as Terminal-Bench measure task completion, while newer evaluations test real-world workflows, multimedia tasks, and whether agents follow relevant instructions without being misled by distractors. A central challenge is making results reliable and transferable, since scores can depend on the harness and runtime, and task success alone may not capture process quality or safe, task-aligned behavior.

Latest papers

31 papers6 papers in the last four weeks, up 100% on the four weeks before. 0.0% of all new papers.

MiRL: Error-Targeted Reinforcement Learning for Correcting Terminal Agent Misbehaviors
Oct 26, 2026Runchu Tian, Zaid Khan, Zun Wang +2RL for LLM Agents Reinforcement Learning
Department of Computer Science, University of North Carolina at Chapel Hill
Terminal-RSI: Recursive Self-Improving Terminal Agents via Recursive Self-Rewrite and Policy-Refine
Oct 15, 2026Junyao Yang, Zongxia Li, Yucheng Shi +6Terminal Agents Recursive Self-Improvement
AutoEnvScaling: Automating the Data Flywheel with Terminal Agents
Oct 9, 2026Simon Yu, Baolin Peng, Bo Liu +8LLM Agent Training Self-Evolving Agents
Northeastern University · Microsoft · Stanford University +2
CoTrace: Data Recipes for Training Terminal Agents with Harness-Model Co-Evolution
Oct 7, 2026Jixuan Chen, Jiaxin Zhang, Qinyuan Ye +11Terminal Agents LLM Agent Harnesses
University of California, San Diego · Salesforce AI Research · University of Washington
CUAWright: A Minimal Unified Interface for Digital Agents
Oct 2, 2026Yadong Lu, Theodore Lee, Yifei Li +5Computer-Use Agents LLM Agent Harnesses
Microsoft Research · National University of Singapore · The Ohio State University +1
Self-Supervised Scaling of Terminal Environments for Scientific Domains
Oct 2, 2026Zhongzhi Li, Yucheng Shi, Zongxia Li +8Terminal Agents LLM Agent Training
Tencent HY LLM Frontier · University of Georgia · University of Maryland, College Park +4
Mid-Harness: Scaling Actions Between Model and Harness for Terminal Agents
Sep 30, 2026Minki Kang, Ryo Hachiuma, Shaokun Zhang +8Terminal Agents Test-Time Scaling
NVIDIA · KAIST
T1: Terminal Agent Reinforcement Learning for Long-Horizon Tasks
Sep 10, 2026Junyao Yang, Yucheng Shi, Zhongzhi Li +4RL for Tool Use Long-Horizon Agent Evaluation
Tencent Hy Foundation Model · National University of Singapore · University of Georgia +2
Is Bash All You Need? An Empirical Study of Tool Interfaces for Enterprise Digital Worker Agents
Sep 10, 2026Hazel Mak, Susheel Suresh, Sahil Bhatnagar +3Terminal Agents LLM Agent Evaluation
Microsoft Corporation One Microsoft Way, Redmond, WA 98052, USA · Carnegie Mellon University 5000 Forbes Avenue, Pittsburgh, PA 15213, USA
Environment Evolution for Terminal Agents
Sep 3, 2026Zhiyuan Fan, Tinghao Yu, Yuanjun Cai +9Terminal Agents Curriculum RL
Hunyuan Team, Tencent
Learning Generalizable Behaviors for Terminal Agents
Aug 23, 2026Yihang Yao, Bo Pang, Xuan Phi Nguyen +3Terminal Agents Generalization in RL
Salesforce AI Research · Carnegie Mellon University
Terminal Agents: A Survey of AI Agents in Command-Line Environments
Aug 20, 2026Yi Bin, Xiaoyang Yuan, Haoxi Zeng +9Terminal Agents Tool-Using Agents
Tongji University, Shanghai, China · Shanghai Innovation Institute, Shanghai, China · Shanghai Jiao Tong University, Shanghai, China +1
FACET: Preserving Source Intent and Executable State in Terminal Task Synthesis
Aug 19, 2026Kou Shi, Zun Wang, Qisheng Su +10Terminal Agents LLM Agent Training
MoE Key Lab of BIPC, University of Science and Technology of China · Shanghai AI Laboratory · Fudan University
Recursive Synthesis for Long-Horizon Terminal Tasks
Aug 5, 2026Zhongzhi Li, Yucheng Shi, Zongxia Li +8Terminal Agents Synthetic Benchmark Generation
Tencent HY LLM Frontier · University of Georgia · University of Maryland, College Park +5
Meta-Task: Turning Terminal Task Synthesis into a Terminal Task for Scalable Agent Training
Jul 30, 2026Zhihong Pan, Jiyuan He, Kai Zhang +5Synthetic Data Generation Terminal Agents
State Key Laboratory of Cognitive Intelligence, University of Science and Technology of China · Meituan
NexForge: Scaling Executable Agent Tasks via Requirement-First Synthesis
Jul 15, 2026Jiarong Zhao, Zhikai Lei, Zhiheng Xi +5Software Engineering Agents Synthetic Data Generation
1East China Normal University · 2Shanghai Qiji Zhifeng Co., Ltd
SETA: Scaling Environments for Terminal Agents
Jul 12, 2026Qijia Shen, Zhiqi Huang, Vamsidhar Kamanuru +19Reinforcement Learning Terminal Agents
Imperial College London · University College London · SambaNova +5
TUA-Bench: A Benchmark for General-Purpose Terminal-Use Agents
Jun 26, 2026Shoufa Chen, Luyuan Wang, Xuan Yang +7Computer-Use Agent Benchmarks Computer-Use Agents
Meta AI · Duke University · Stanford University
Tmax: A simple recipe for terminal agents
Jun 22, 2026Hamish Ivison, Junjie Oscar Yin, Rulin Shao +3Synthetic Data Generation Terminal Agents
α Allen Institute for AI · ω University of Washington
LiteCoder-Terminal: Scaling Long-Horizon Terminal Environments for Learning Language Agents
May 28, 2026Xiaoxuan Peng, Kaiqi Zhang, Xinyu Lu +5RL for Language Models Terminal Agents
Chinese Information Processing Laboratory, Institute of Software, Chinese Academy of Sciences · University of Chinese Academy of Sciences
ECHO: Terminal Agents Learn World Models for Free
May 23, 2026Vaishnavi Shrivastava, Piero Kauffmann, Ahmed Awadallah +1World Model Learning Computer-Use Agents
Microsoft Research
TerminalWorld: Benchmarking Agents on Real-World Terminal Tasks
May 21, 2026Zhaoyang Chu, Jiarui Hu, Xingyu Jiang +8Computer-Use Agent Benchmarks Terminal-Agent Benchmarks
University College London · Nanjing University · Tencent
Terminal-World: Scaling Terminal-Agent Environments via Agent Skills
May 20, 2026Zihao Cheng, Hongru Wang, Zeming Liu +6Tool-Augmented Language Model Agents Terminal Agents
School of Computer Science and Engineering, Beihang University, Beijing, China · Independent Researcher · 3Beijing Institute of Technology +2
No More, No Less: Task Alignment in Terminal Agents
May 12, 2026Sina Mavali, David Pape, Jonathan Evertz +5Computer-Use Agent Benchmarks AI Alignment
CISPA Helmholtz Center for Information Security · ELLIS Institute Tübingen and MPI for Intelligent Systems · Tübingen AI Center
WildClawBench: A Benchmark for Real-World, Long-Horizon Agent Evaluation
May 11, 2026Shuangrui Ding, Xuanlang Dai, Long Xing +14LLM Agent Harnesses AI Agent Benchmarks
1Shanghai AI Laboratory, 2The Chinese University of Hong Kong · 1Shanghai AI Laboratory, 3Fudan University · 1Shanghai AI Laboratory, 4University of Science and Technology of China +8
MMTB: Evaluating Terminal Agents on Multimedia-File Tasks
May 8, 2026Chiyeong Heo, Jaechang Kim, Junhyuk Kwon +4Computer-Use Agent Benchmarks Audio-Visual Understanding
GSAI, POSTECH · National AI Research Lab · Krafton AI +1
What Makes a Good Terminal-Agent Benchmark Task: A Guideline for Adversarial, Difficult, and Legible Evaluation Design
Apr 30, 2026Ivan BercovichComputer-Use Agent Benchmarks Reward Hacking
Toward Scalable Terminal Task Synthesis via Skill Graphs
Apr 28, 2026Zhiyuan Fan, Tinghao Yu, Yuanjun Cai +8Terminal Agents LLM Agent Trajectory Synthesis
Hunyuan Team, Tencent
A Self-Evolving Framework for Efficient Terminal Agents via Observational Context Compression
Apr 21, 2026Jincheng Ren, Siwei Wu, Yizhi Li +8Terminal Agents Long-Horizon Agent Tasks
University of Manchester · 2MAP · 4HKUST(GZ) +2
Terminal Agents Suffice for Enterprise Automation
Mar 31, 2026Patrice Bechard, Orlando Marquez Ayala, Emily Chen +5Terminal Agents Business Process Automation
ServiceNow · Mila – Quebec AI Institute · Université de Montréal

Oct 26, 2026J/K move · Enter open · S save

MiRL: Error-Targeted Reinforcement Learning for Correcting Terminal Agent Misbehaviors

Runchu Tian, Zaid Khan, Zun Wang, Justin Chih-Yao Chen+1

Department of Computer Science, University of North Carolina at Chapel Hill

RL for LLM Agents Reinforcement Learning Terminal Agents Long-Horizon RL

Open paper AskPDF arXiv

Error-targeted learning gives a legible way to improve a model: identify specific weaknesses of the model, train the model to fix these weaknesses, and check whether the weaknesses are fixed. We study how to apply error-targeted learning to terminal agents, which solve long-horizon tasks specified in natural language by interacting with a command shell. Existing error-targeted methods are designed for single-turn tasks, and existing analyses of agent failures are used to evaluate and debug agents rather than to train them. We propose Misbehavior-driven RL (MiRL), which first profiles the errors of a policy to find generalizable misbehaviors: kinds of wrong steps that recur across many environments. MiRL then trains the policy with RL, starting rollouts at the steps where these misbehaviors occur and rewarding a rollout only if it passes the tests and avoids the misbehavior. On Qwen3.5-4B, eight misbehaviors account for 52.1% of wrong steps. After training with MiRL, the policy repeats the measured misbehaviors 43% to 57% less often on average on unseen prefixes. MiRL outperforms TMax-4B on Terminal-Bench Lite (51.8% vs. 47.4%) and matches it on Terminal-Bench 2.1 (21.3% vs. 20.6%), while using 35% of its rollouts. These results show that error-targeted learning can be applied to realistic, long-horizon terminal tasks.