The paper investigates how optimizing the harness of language model agents—without changing model weights—affects performance. It distinguishes between gains that improve success on tasks already within reach (elicitation) and gains that broaden the range of solvable tasks (expansion). By comparing optimized harnesses to their pre‑optimization versions across static and dynamic tasks, the study finds that benefits can reverse at higher attempt budgets and that aggregate gains may mask shortcomings on harder tasks, highlighting the need for detailed diagnostics.
Large Language Models (LLMs) have driven rapid progress in autonomous agents, yet standard evaluations remain confined to static task solving. An emerging frontier is harness evolution---the agent's capacity to autonomously optimize its own operating harness.
arXiv:2610.28201v1 Announce Type: new Abstract: Recursive self-improvement (RSI) of agent harnesses has emerged as an effective approach to enhancing the capabilities of large language model (LLM) ag...
arXiv:2610.10426v1 Announce Type: new Abstract: Terminal-agent capability depends jointly on model weights and the runtime harness that formats prompts, binds tools, and handles error recovery. Exist...
arXiv:2608. 20169v1 Announce Type: cross Abstract: We present a novel approach to efficient LLM agent harness optimization through adaptive validation task selection.
Terminal-agent capability depends jointly on model weights and the runtime harness that formats prompts, binds tools, and handles error recovery. Existing harness-model co-evolution approaches improve...
The paper introduces Harness-Delta Attribution (HDA), a protocol that breaks down the performance gains from harness evolution into overfitting, test-time scaling, and residual generalizable improvement. Applying HDA to 11 executor models and three harness-evolution frameworks across various tasks, the study finds that over half of the search-set gains are due to overfitting or scaling, with limited transfer to held-out tasks. The results highlight the need for attribution-aware evaluation in future harness evolution research.
The paper investigates the limitations of post-training AI agents that can autonomously train large language models. It distinguishes between execution-level capability—making adjustments within a chosen training strategy—and strategy-level capability—revising the overall approach based on new evidence. Analysis of many public post-training runs shows that agents lock into a strategy early and then only perform local tweaks, regardless of task. Experiments with experience scaffolds, human guidance, and extra compute improve execution but do not enable strategy reevaluation, indicating that agents lack a mechanism to spontaneously reassess their strategy during training.
The paper introduces Self-HPO, a recursive framework that alternates between optimizing the harness and the policy of large language model agents for coding tasks. By letting the current policy propose new harness candidates based on its failures, the method aligns harness evolution with policy improvement, leading to consistent performance gains across three benchmarks. Experiments show that Self-HPO raises the average resolve rate of Qwen3.5-35B-A3B from 57.8% to 64.8% and improves reliability and capability metrics, with benefits that generalize to out‑of‑distribution settings.
The paper introduces a self‑evolving harness framework where a frozen language‑model agent first solves tasks and then edits its own harness based on run records. Using a 49‑line seed harness, the evolved harness improves average scores on in‑distribution benchmarks by 4.48 points and on out‑of‑distribution benchmarks by 12.64 points, surpassing Codex on the former and matching it on the latter. Continued evolution on a specific out‑of‑distribution benchmark further raises performance, and the study analyzes emergent mechanisms such as output truncation and history compaction.
The paper introduces Rollout-Time Intervention (RTI), a method that uses reward-side evidence during rollouts to transform unsuccessful trajectories into successful ones, thereby providing learning signals for tasks beyond a policy’s intrinsic exploration limits. RTI filters out trajectories with excessive distributional shifts, allowing these off-policy interventions to guide policy optimization effectively. Experiments on five benchmarks show that RTI consistently outperforms existing RL baselines, enabling post-trained policies to approach the performance of stronger large language models and surpass their original capability boundaries.
The study evaluates multi-harness reinforcement learning (RL) for coding agents by training two Qwen3-8B RL variants—one with rewards normalized within each harness and one across harnesses—using identical replay data. Results show modest gains over the supervised fine-tuning baseline (+0.48 and +0.77 percentage points for within and cross normalization, respectively), but no consistent advantage of cross normalization on unseen harnesses. Performance differences across harnesses are much larger than the training gains, indicating that harness dependence remains a significant challenge.
arXiv:2607. 26722v1 Announce Type: cross Abstract: Harness plays a critical role in large language model agent performance, and building a high-performing harness requires substantial expert effort.