LLM Agent Harnesses

LLM: Large Language Model

An LLM agent harness surrounds a language model with the mechanisms needed to act: tool interfaces, control loops, context and memory management, persistent state, and execution environments. Common approaches include ReAct-style loops, explicit workflow graphs, typed tool contracts, sandboxes, and verification or recovery steps; newer systems also adapt harnesses using execution traces, search, or reinforcement learning. Researchers evaluate these systems on coding, web, terminal, scientific, and interactive tasks, measuring not only success but also cost, reliability, and safety. A central tension is how much structure helps: added guidance can improve control and recovery, yet may constrain useful behavior, fail to transfer across models or tasks, or make comparisons hard to interpret without specifying the harness.

Latest papers

389 papers118 papers in the last four weeks, up 103% on the four weeks before. 0.7% of all new papers.

Applying Anthropic Primitives at Large Enterprises: Harness Paradigm for Knowledge Work
Aug 20, 2026George Juraj SalapaLLM Agent Harnesses LLM Agent Orchestration
G.S. s.r.o., PwC Austria
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
Bounded Loops: Pre-Run Spend Bounds, Proved Termination, and Verified Completion for Agent Harnesses
Aug 20, 2026Varun Pratap Bhardwaj, Garima Singh, Arun Pratap BhardwajLLM Agent Harnesses Runtime Enforcement for AI Agents
Qualixar / Independent Researcher, India · Independent Researcher, India
EnvHarness: Awakening Static Worlds for Agent Learning
Aug 20, 2026Chengsong Huang, Zifeng Wang, Rujun Han +14LLM Agent Harnesses Agent Harness Optimization
Washington University in St. Louis · Google Cloud AI Research · University of North Carolina at Chapel Hill +1
StagedWorkspace: A Versioned Workspace for Knowledge-Work Agents
Aug 18, 2026Yining Hua, Hongbin Na, Yifan Zhou +3LLM Agents LLM Agent Evaluation
Harvard University · Raycaster AI · University of Technology Sydney +2
HarnessRisk: A Lifecycle-Oriented Benchmark for Agent Harness Safety
Aug 18, 2026Yajing Bai, Jinhao Duan, Jie Peng +4AI Agent Safety Benchmarks LLM Agent Safety
University of North Carolina at Chapel Hill · University of Central Florida · Michigan State University
Agent Lightning v1.0: Towards Harnessed Agentic RL
Aug 18, 2026Zhiyuan He, Siwei Zhang, Zhiwen Zhou +7Agentic RL LLM Agent Training
Microsoft · Fudan University · Zhejiang University +1
Task-Aware Harness Provisioning for LLM Agents in Mission-Critical Infrastructure Operations
Aug 18, 2026Liangtao Lin, Qingang Zhang, Zhaomeng Zhu +2Agent Harness Optimization LLM Agent Harnesses
Nanyang Technological University Singapore
SkillEffect: Checked Lowering for Memory-Bounded Agent Tools
Aug 17, 2026Yinuo Wang, Yiyu ShiRuntime Enforcement for AI Agents Tool-Using Agents
Department of Computer Science & Engineering University of Notre Dame Notre Dame, IN, USA
ClawGym II: Exploring Black-Box RL on Agent Harness
Aug 17, 2026Huatong Song, Fei Bai, Ming Yang +17Reinforcement Learning RL for LLM Agents
Gaoling School of Artificial Intelligence, Renmin University of China · IQuest Research
The Working Set of a Coding Agent: Coherence Debt in Repository-Scale Tasks
Aug 17, 2026Bardia Mohammadi, Lars Klein, Aman Chadha +2Repository-Level Code Generation LLM Agent Memory
Max Planck Institute for Software Systems · EPFL · Apple +1
Mint-Agent: Introducing Finance-Native Agentic Foundation Models
Aug 17, 2026Mint-Agent Team, Kun Wang, Gavin Zhang +7LLM Agent Training Autonomous Research Agents
Mint Agent Team · Nanyang Technological University · Shanghai AI-Finance School, East China Normal University +1
AstronOS: A Unified Execution Model and Runtime for Long-Horizon Agentic Systems
Aug 17, 2026Zhenhang Nie, Gui Zheng, Xudong Sun +2Long-Horizon Agent Tasks LLM Agent Harnesses
iFLYTEK Co., Ltd., Hefei, China · iFLYTEK Co., Ltd., Hefei, Anhui, China.
Deploying Frontier Agentic Technology in MOOSEnger, a Multiphysics-Capable AI Assistant
Aug 16, 2026Zaid Abulawi, Mengnan Li, Guillaume Giudicelli +2Tool-Augmented Language Model Agents LLM Agent Harnesses
Department of Nuclear Engineering, Texas A&M University (TAMU), TX, USA · Idaho National Laboratory (INL), ID, USA
Training Agents to Evolve with Their Harness: TaoLive Digital Avatar Agent Technical Report
Aug 16, 2026TaoLive AIGC LLM Team, Yuhan Sun, Wenhao Lin +7LLM Agent Training LLM Agent Harnesses
TaoLive AIGC LLM Team
Agent Gym: A Framework for Continuous Evaluation and Evolution of LLM Agents Through Human-in-the-Loop Feedback
Aug 16, 2026Pouya Ghiasnezhad Omran, Michael Zimmermann, Duncan Cambridge +2LLM Agent Evaluation LLM Agent Self-Improvement
Google Cloud
Kozuchi Agent: A Language-Agnostic Open-Weight Agent for Software Repair
Aug 16, 2026Mehdi Bahrami, Kosaku Kimura, Satoshi Munakata +24Software Engineering Agents Automated Program Repair
Fujitsu Research of America Santa Clara, CA, USA · Fujitsu Research Kawasaki, Japan · Fujitsu Research of Europe Slough, UK +1
StateM: Reaching 95.3% Raw Accuracy, or a $15 Frontier Run, on Terminal-Bench 2.1 via Harness Scaling
Aug 15, 2026Ziheng Qin, Yaxin Lu, Zhangyang Atlas Wang +1LLM Agent Harnesses Long-Horizon LLM Agents
Somewhere on the Earth
Evo-Harness: Context-to-Harness Skill Compilation for Self-Evolving Agents
Aug 15, 2026Tianxin Wei, Zhan Shi, Minhua Lin +14Continual Learning for LLM Agents LLM Agent Skill Learning
University of Illinois Urbana-Champaign · Amazon · The Pennsylvania State University +3
MedClaw: Heuristic Agent Harness for Long-Horizon Surgical Video Reasoning
Aug 14, 2026Yingying Fan, Penghui Du, Leyan Zhu +10Surgical Video Understanding LLM Agent Harnesses
School of Automation and Intelligence, Beijing Jiaotong University, Beijing 100044, China · Department of Electrical and Computer Engineering, University of Maryland, College Park, MD 20742, USA · University of British Columbia +2
HELIX: Model-Harness Co-evolution for Recursive Self-Improvement
Aug 14, 2026Tianyu Fan, Chao HuangLLM Agent Harnesses Agent Harness Optimization
In3 Lab, The University of Hong Kong
Capability Sheaves for Compositional Agent-Harness Repair: Controlled Quotients and a Real-Repository Stress Test
Aug 13, 2026Saveliy BatruinAutomated Program Repair Sheaf Theory
Independent researcher
Beyond Handcrafted Security: Towards Self-Evolving Defense for LLM Agents
Aug 13, 2026Jiajun Ruan, Peiyang Li, Yukun Chen +2LLM Agent Security LLM Agent Safety
University of Minnesota · Ant Group · Tsinghua University +1
@skills: Attention is all you have
Aug 12, 2026Li Yin, Zhi Li, Zhan Shi +4LLM Agent Skill Retrieval LLM Agent Harnesses
Atlas · The University of Texas at Austin
SHE: Trajectory-driven Safety Harness Evolution for LLM Agents
Aug 10, 2026Wanying Qu, Qinghua Mao, Yu Li +12Agent Harness Optimization LLM Agent Safety
Shanghai Artificial Intelligence Laboratory · Shanghai Jiao Tong University · Fudan University +1
STAIR: Effective Incident Response Using an End-to-End Agentic Planning Framework
Aug 10, 2026Hanlin Jiang, Jionghao Huang, Shaofei Li +6Autonomous Cyber Defense LLM Agent Orchestration
Key Laboratory of High Confidence Software Technologies, Peking University, Ministry of Education · Southeast University · Huawei Technologies Co., Ltd.
From Prompt to Harness: Coderlet from Scratch
Aug 10, 2026Mengfan LiCoding Agents Agent Harness Optimization
Evidence-Calibrated Runtime Reconstruction for Agent Skills Across Heterogeneous Coding Agents
Aug 9, 2026Xueping GaoAI Coding Agents AI Agent Monitoring
Alibaba Cloud, Hangzhou, China
The Scaffolding Matters More Than the Interface: A Controlled Comparison of MCP and CLI Tool Use Across Seven Agent Scaffoldings, Five Language Models, and One Software Task
Aug 9, 2026Marc Alier Forment, María José Casañ Guerrero, Francisco José García-Peñalvo +1AI Coding Agents LLM Agent Evaluation
Universitat Politècnica de Catalunya (UPC), Barcelona, Spain · Universidad de Salamanca (USAL), Salamanca, Spain · Universidad del País Vasco / Euskal Herriko Unibertsitatea (UPV/EHU), Donostia-San Sebastián, Spain
$A^2E$ : An End-to-End Agent Auditing Engine
Aug 7, 2026Haoning Wang, Mingxun Zhang, Chenyue Yu +4Tool-Use Evaluation AI Agent Auditing
Shanghai Artificial Intelligence Laboratory

Aug 20, 2026J/K move · Enter open · S save

Applying Anthropic Primitives at Large Enterprises: Harness Paradigm for Knowledge Work

George Juraj Salapa

G.S. s.r.o., PwC Austria

LLM Agent Harnesses LLM Agent Orchestration AI Agent Governance Business Process Automation Tool Access Control for LLM Agents

Open paper AskPDF arXiv

Frontier models have collapsed the cost of writing custom code: a niche problem a specialist sees in their own domain now costs an afternoon. The cost of reviewing and maintaining that code hasn't collapsed. Each solution drifts from the next; understanding one means reading its codebase from scratch. Large enterprises build something centrally governed instead: at worst an off-the-shelf product, at best a graph-orchestration framework wired bespoke per use case, or a low-code platform used as the orchestrator. These are custom every time and limited in scope. Enterprises don't weigh a third option that escapes both constraints: the harness paradigm. Recent work treats the coding-agent harness as enterprise infrastructure rather than a coding tool, converging on three findings: harnesses suffice at the task level and outperform more elaborate architectures on enterprise work (arXiv:2604.00073, arXiv:2604.13107); harness choice accounts for most of the variance in agent benchmark results, more than model choice does (arXiv:2605.23950); and the gap between that finding and enterprise adoption is governance (arXiv:2605.10223, arXiv:2605.18747). We propose an architecture that closes that gap. One harness runs unmodified as the backbone; the code stays identical across every deployment, so reviewing what gets built collapses to reading its instructions file. Section 4 gives four mechanisms: credential-scoped tooling, where each backend gets one generic request tool and a scoped credential instead of a hand-built method; authorization logic outside the harness, so one artifact runs as a cron backbone, a chat-surface engine, and a terminal tool; registration is a side effect of pushing code, collapsing an audit a review of a text file. Built on microcc (https://pypi.org/project/micro-cc/), our reference harness.