Multi-agent LLM systems are increasingly used to solve complex tasks through decomposition, debate, specialization, and ensemble reasoning. However, these systems are usually evaluated in terms of robustness: whether performance is preserved under perturbation. This paper studies a different question: whether semantic stress exposes structured variation that could support future antifragile learning. We introduce CAFE (Cognitive Antifragility Framework for Evaluation), a statistical framework for detecting antifragility-compatible regimes in multi-agent architectures. CAFE models a controlled expected distribution of semantic stressors, reconstructs an architecture-specific observed effective stress distribution from multi-dimensional judge signals, and compares both distributions using a distributional Jensen Gap under a convex stress potential. A positive gap does not imply immediate performance improvement; instead, it indicates a convex-expansive deformation of the observed stress distribution, suggesting that the architecture exposes learnable stress structure. We evaluate CAFE on a banking-risk analysis benchmark with five multi-agent architectures: flat, hierarchical, debate, meta-adaptive, and ensemble. Across all architectures, semantic stress reduces average judged quality by roughly one third. Yet all architectures exhibit positive distributional Jensen Gaps with bootstrap confidence intervals above zero. These results show that immediate quality degradation can coexist with statistically detectable antifragility-compatible stress geometry. CAFE is therefore not an antifragile learner itself, but a measurement layer for identifying when and where antifragility learning may be worth applying.

翻译：基于大语言模型的多智能体系统日益广泛地通过分解、辩论、专业化与集成推理来求解复杂任务。然而，这类系统通常基于鲁棒性进行评估：即在扰动下性能是否保持不变。本文研究了一个不同的问题：语义压力是否能暴露可支撑未来反脆弱性学习的结构化变异。我们提出CAFE（反脆弱性认知评估框架），一个用于检测多智能体架构中反脆弱兼容状态的统计框架。CAFE对语义压力的受控期望分布进行建模，从多维评判器信号中重构架构特定的观测有效压力分布，并在凸压力势下基于分布性Jensen间隙比较两种分布。正向间隙不直接表明即时性能提升，而是指示观测压力分布发生凸扩张形变，表明该架构暴露了可学习的压力结构。我们在银行业风险分析基准上对五种多智能体架构（扁平式、层级式、辩论式、元自适应式和集成式）评估了CAFE。在所有架构中，语义压力使平均评判质量降低约三分之一。然而，所有架构均呈现正向分布性Jensen间隙，其自助法置信区间均大于零。这些结果表明，即时质量退化可与统计显著的反脆弱性兼容压力几何结构共存。因此CAFE本身并非反脆弱学习器，而是用于识别反脆弱学习可能适用时机与场景的测量层。