ModularRSI: Modular and Generalizable Recursive Harness Self-Improvement

Recent work extends Recursive Self-Improvement (RSI) to agent harnesses for long-horizon coding and terminal tasks, enabling agents to improve their execution mechanisms from experience. However, achieving and demonstrating generalizable harness RSI remains challenging. First, existing approaches often evolve harnesses directly on evaluation benchmarks or subsets drawn from them, making it difficult to distinguish reusable harness improvements from benchmarkspecific adaptation. Second, updates derived from individual trajectories can entangle systematic harness deficiencies with instance-specific reasoning and solution details, leading to task-specific modifications that transfer poorly to unseen tasks. Third, even when recurring behavioral deficiencies are identified, localizing them to the responsible components within a monolithic harness remains difficult. Whole-harness optimization can therefore entangle unrelated mechanisms and produce changes that are difficult to attribute and validate. We propose ModularRSI, a benchmark-disjoint, contrastive, and modular framework for generalizable harness evolution. ModularRSI contrasts successful and failed trajectories for the same task and aggregates evidence across tasks to identify recurring behavioral deficiencies.

Introduction. CLI agents have achieved remarkable performance on complex software engineering and terminalbased tasks (Jimenez et al., 2024a; Deng et al., 2025; Merrill et al., 2026b; Hong et al., 2026). Beyond foundation models, their effectiveness increasingly depends on agent harnesses that govern execution, tool interaction, context management, and environment feedback. Recent work has therefore explored recursive self-improvement (RSI) of harnesses, allowing agents to refine these mechanisms from execution experience. However, achieving generalizable harness RSI remains challenging because task-level outcomes provide only coarse supervision for harness evolution. We identify three coupled challenges. First, there is a data-level challenge in obtaining high-quality evolution experience. Harness RSI requires executable long-horizon terminal tasks with reliable environments and correctness feedback, but constructing such an evolution dataset at sufficient scale and diversity is costly and difficult.

Discussion / Conclusion. We present ModularRSI, a benchmark-disjoint and modular framework for generalizable harness self-improvement. By contrasting execution trajectories and evolving harness modules independently, ModularRSI identifies reusable mechanism-level improvements while reducing task-specific adaptation. Experiments on TerminalBench 2.0 and SWE-Bench Verified demonstrate consistent gains across unseen tasks, domains, and foundation models. Our study has several limitations. First, we do not conduct a dedicated ablation that isolates the contribution of contrastive trajectory analysis, although our trajectory analysis and case studies provide supporting evidence. In addition, due to computational cost, our main evolution experiments use only a subset of the 2,000 curated evolution instances. We leave broader comparisons and largerscale studies to future work.