这篇论文用Terminal-Bench 2.0测试三种Agent优化方法，发现只有RELAI-VCL能持续累积提升，终身平均76.4%远超其他。做智能体持续学习的人必看。
大多数智能体优化研究仅测试一次性固定基准，忽略连续优化场景。本文使用Terminal-Bench 2.0的困难任务设计两阶段持续学习评估，对比GEPA、Meta Harness和RELAI-VCL三种方法。静态评估中三者均提升基线（58.7%），但引入新任务后分化明显：RELAI-VCL既正向转移又持续改进，每阶段通过率最高，终身平均达76.4%，远超GEPA（66.0%）、Meta Harness（64.6%）和基线（58.7%）。关键发现是优化增益仅在优化循环内置回归控制时才能累积。
Do Agent Optimizers Compound? A Continual-Learning Evaluation on Terminal-Bench 2.0
Most reported gains from agent-optimization methods are one-shot: an agent is optimized against a fixed benchmark and the resulting improvement is reported as if it were a stable property of the method. This does not test the setting that matters for deployed agents, where optimization is applied recursively as new failures and new tasks appear over time. The central question this raises is whether optimizer-driven gains compound: after an agent has been optimized once, can it be optimized again on newly arrived tasks without eroding the gains the first round produced? We study this question with a two-phase continual-learning evaluation built from hard tasks in Terminal-Bench 2.0, comparing three approaches to agent-harness optimization (GEPA, Meta Harness, and RELAI's Verifiable Continual Learning, RELAI-VCL) under identical optimization budgets. All three methods improve over the baseline agent in the conventional, static, single-phase setting. However, once new tasks are introduced, the methods diverge sharply: GEPA's optimized agent transfers below the unoptimized baseline, Meta Harness transfers well but fails to improve further once given a second optimization budget, and RELAI-VCL is the only method that both transfers positively to unseen tasks and continues improving after those tasks are folded into the optimization objective, reaching the highest pass rate at every evaluated stage and the highest lifelong average pass rate overall (76.4% vs. 66.0% for GEPA, 64.6% for Meta Harness, and 58.7% for the baseline). Our key observation was that optimization gains compounded only when regression control was built into the optimization loop, providing an inductive bias against shortcut solutions that fail to generalize.