{"slug": "forging-more-tasks-into-better-harnesses-agentic-harness-evolution-via-scale", "title": "Forging More Tasks into Better Harnesses: Agentic Harness Evolution via Successive Scale Racing", "summary": "A new arXiv paper (2610.28815v1) proposes Successive Scale Racing (SSR), a framework that adaptively searches for optimal evidence combinations in each evolution round when optimizing an AI agent's instructions, tools, and control flow. Across three benchmarks and six configurations, SSR reached state-of-the-art results, and as the task pool grew from 40 to 200 tasks its pass rate rose from 64.3% to 68.3% while baseline methods' performance dropped; SSR's racing mechanism also cut validation executions by 60% while identifying near-optimal candidates. The authors report that existing agentic harness evolution methods degrade as the task pool expands due to suboptimal evidence selection, a failure mode SSR is designed to correct.", "body_md": "arXiv:2610.28815v1 Announce Type: new \nAbstract: Agentic harness evolution optimizes an agent's instructions, tools, and control flow by using its task execution experiences as evolving evidence. While more evolution tasks should intuitively improve performance, our preliminary study reveals a paradoxical failure mode: the performance of existing methods degrades as the task pool expands. Our further analysis demonstrates that this degradation stems from suboptimal evidence selection during optimization, as different evidence combinations lead to substantially different evolution outcomes. To address this issue, we propose Successive Scale Racing (SSR), a framework that adaptively searches for optimal evidence combinations in each evolution round. SSR constructs diverse evidence combinations based on the corresponding task quality and coverage, and efficiently races the resulting candidate harnesses under a small compute budget. Empirically, across three distinct benchmarks, SSR achieves state-of-the-art results in all six configurations. Notably, as the task pool grows from 40 to 200, SSR steadily improves the pass rate from 64.3% to 68.3%, while the baselines' performances all drop. Furthermore, its racing mechanism identifies near-optimal candidates while reducing validation executions by 60%.", "url": "https://wpnews.pro/news/forging-more-tasks-into-better-harnesses-agentic-harness-evolution-via-scale", "canonical_source": "https://arxiv.org/abs/2610.28815", "published_at": "2026-10-23 04:00:00+00:00", "updated_at": "2026-10-23 04:14:01.341572+00:00", "lang": "en", "topics": ["ai-agents", "ai-research", "artificial-intelligence", "machine-learning"], "entities": ["Successive Scale Racing", "arXiv"], "also_reported_by": [], "alternates": {"html": "https://wpnews.pro/news/forging-more-tasks-into-better-harnesses-agentic-harness-evolution-via-scale", "markdown": "https://wpnews.pro/news/forging-more-tasks-into-better-harnesses-agentic-harness-evolution-via-scale.md", "text": "https://wpnews.pro/news/forging-more-tasks-into-better-harnesses-agentic-harness-evolution-via-scale.txt", "jsonld": "https://wpnews.pro/news/forging-more-tasks-into-better-harnesses-agentic-harness-evolution-via-scale.jsonld"}}