HarnessMDP: Learning to Control LLM Agent Harnesses with Offline Reinforcement Learning
arXiv:2607.05458v3 Announce Type: replace-cross Abstract: The execution harness is a critical component of LLM agents, shaping how the agent interacts with its environment and organizes multi-step execution. Prior work has shown that improving tools, workflows, and other harness components can substantially improve agent performance. This suggests that the harness can be treated as a learnable control layer that adapts execution decisions from experience, rather than as a predetermined executio…