Multi-Harness Reinforcement Learning for Coding Agents: Training Gains and Harness Dependence

arXiv:2609.04518 · cs.AI · Submitted 2026-09-03 · Read on arXiv

cs.AI

Submitted: 2026-09-03

Updated: 2026-10-09

Comments: 21 pages, 1 figure, 10 tables. Preprint

Code: https://github.com/Aider-AI/aider

License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/

Terminology

• Adaptation — 1× in this paper · explained in 1 episode
• Credit assignment — 1× in this paper · explained in 3 episodes
• GRPO — 1× in this paper · explained in 6 episodes
• Reinforcement Learning — 1× in this paper · explained in 4 episodes
• Reinforcement Learning (RL) — 1× in this paper · explained in 25 episodes

Sources

• HarnessX: A Composable, Adaptive, and Evolvable Agent Harness Foundry
• Mach-Mind-4-Flash Technical Report
• KAT-Coder-V2.5 Technical Report
• Tmax: A simple recipe for terminal agents
• The Interplay of Harness Design and Post-Training in LLM Agents
• Kimi K3: Open Frontier Intelligence
• Nemotron 3 Super: Open, Efficient Mixture-of-Experts Hybrid Mamba-Transformer Model for Agentic Reasoning
• Orchard: An Open-Source Agentic Modeling Framework
• Hindsight Credit Assignment for Long-Horizon LLM Agents
• Polar: Agentic RL on Any Harness at Scale
• OpenForgeRL: Train Harness-native Agents in Any Environment

Related papers

• MAVEN-T: Reinforced Heterogeneous Distillation for Real-Time Multi-Agent Trajectory Prediction
• Model Discovery Agent: LLM-assisted Bayesian experiment design for data-efficient discovery of mechanistic world models
• Agents' Last Exam
• The Clinician's Veto: Navigating Trust, Liability, and Uncertainty in Autonomous AI Prescribing
• MindHelper: Closed-Loop Embodied Mental-State Reasoning for Precision Intervention
• Incumbent Advantage: Brand Bias and Cognitive Manipulation Dynamics in LLM Recommendation Systems