Large language models (LLMs) are fluent on open-ended tasks, yet in agentic settings, where a system must plan, use tools, and act over extended horizons, fluency does not ensure reliable delivery. We trace this gap to three coupled structural failures: errors propagate without localization, worst-case perturbations go unevaluated, and accumulated knowledge is never invalidated. We argue these share a root cause: abductive, counterfactual, meta-inductive, corrective, and inductive reasoning pull a shared context in incompatible directions. We introduce Reflective Adversarial Pareto Search (R-APS), to our knowledge the first method addressing all three failures jointly via reasoning-mode decomposition, allocating each reasoning mode its own context and orchestrating interaction across three timescales: staged compositional reasoning with a typed validation critic (failure localization), sensitivity-guided counterfactual stress-testing as a first-class Pareto objective (robustness), and meta-inductive rule extraction with explicit invalidation (persistent memory). R-APS requires no fine-tuning and operates on a frozen LLM purely via structured protocol design. We evaluate on planar mechanism synthesis (robotics, prosthetics, mechanical design), with every candidate checked by a kinematic solver. On 32 target trajectories, R-APS delivers robustness certificates 3.5x tighter than uniform-perturbation baselines, 46% faster iterations-to-first-admission, and 2.1x Chamfer-distance reduction over Enum+GA while jointly controlling bar-count and worst-case robustness. Small 4B reasoning-specialized models prove competitive with general-purpose 70B backbones inside the protocol, suggesting structured protocols can partially offset model scale.

翻译：大型语言模型（LLM）在开放式任务中表现流畅，但在需要系统规划、使用工具并在长周期内行动的智能体场景中，流畅性无法保证可靠性。我们将这一差距归因于三个相互关联的结构性缺陷：错误传播缺乏定位、最坏情况扰动未被评估、累积知识从未失效。我们认为这些问题的根源在于：溯因推理、反事实推理、元归纳推理、纠错推理和归纳推理对共享上下文施加了不相容的驱动力。为此，我们提出反思性对抗帕累托搜索（R-APS），据我们所知，这是首个通过推理模式分解联合解决上述三种缺陷的方法——为每种推理模式分配独立上下文，并在三个时间尺度上协调交互：带类型验证评判器的分阶段组合推理（故障定位）、将灵敏度引导的反事实压力测试作为一阶帕累托目标（鲁棒性）、以及带显式失效机制的元归纳规则提取（持久记忆）。R-APS无需微调，仅通过结构化协议设计即可在冻结LLM上运行。我们在平面机构综合（机器人学、假肢设计、机械设计）任务上进行评估，每个候选解均通过运动学求解器验证。在32条目标轨迹上，R-APS提供的鲁棒性证书比均匀扰动基线严格3.5倍，首次准入迭代速度提升46%，在联合控制杆数约束与最坏情况鲁棒性的条件下，相较于枚举+遗传算法实现了2.1倍的倒角距离缩减。协议内小型4B推理专用模型可与通用70B骨干模型竞争，表明结构化协议可部分抵消模型规模需求。