给 AI 配的鞍子，现在能自己调了
大模型在长任务上总翻车，不是它笨，是没人给它配好鞍——外部那套提示词、工具配置和控制逻辑，全靠人手工调，又贵又慢。这篇把「调鞍子」变成了自动的：让 AI 自己从失败记录里找病根、把鞍子当代码改、改完先验证再上马。在三个硬核长任务基准上，自动配的鞍子比人工基线分别高出 9.0、9.6、10.0 个百分点。它不是你明天能用上的东西，但方向很明确：与其逼模型变强，不如让套在模型外面的那层壳自己进化。
📄 原文摘要(英文)
LLM agents remain unreliable on long-horizon tasks, where small local failures can compound over extended interactions and lead to overall task failure. Although external harnesses can substantially improve robustness, harness design remains a manual and expensive process that requires searching over a large space of prompts, tool configurations, and control logic. We propose AutoSaddler, an automatic harness optimization framework that formulates harness improvement as an offline learning problem and iteratively updates the harness using failure signals from mini-batches. AutoSaddler combines failure-trace diagnosis, structured patch generation that treats the harness as code, and validation-based update selection. Experiments on GAIA2, SWE-Bench Pro, and Terminal-Bench 2.0 show that AutoSaddler substantially improves agent performance over the corresponding base harnesses, achieving gains of 9.0, 9.6, and 10.0 percentage points, respectively. Ablation studies further suggest that effective harness optimization benefits from three ingredients: deep debugging rather than shallow reflection, targeted modifications rather than unconstrained editing, and generalization-aware selection rather than trajectory-specific repair. Together, these results suggest that automatic harness optimization is a promising path toward more performant and reliable agent systems.