arXiv：cs.LG· Xin Xu, Siru Tao·· 3 小时前AI 评分33
智能体基础设施验证套件中的判别性夹具覆盖研究
Discriminating Fixture Coverage in Agent-Infrastructure Verification Suites
AI 导读
针对多会话智能体状态投影层的不变式套件，研究发现标准验证会放过一个删除事件身份去重的变异体。修复后的十二项检查套件对十个变异体仅杀死五个，存活者分两类：三个因无夹具提供差异化输入而从未激活，两个破坏套件预言机无法观测的内部状态。研究枚举七个判别维度并各补一个夹具后，五个存活者全部被杀死。
正文
Abstract:Invariant suites and runtime monitors increasingly gate agent deployment decisions, and the evidence offered for any particular suite is almost always a single observation: it passes an implementation believed correct and fails one believed broken. We measure what that observation is worth. Applying mutation analysis to an invariant suite for a multi-session agent state-projection layer, we first find that this standard validation certifies a suite in which a first-order mutant removing event-identity deduplication survives every check. We then freeze the repaired twelve-check suite, record its hash, and run it once against ten mutants specified by an adversarial reader who designed none of its fixtures: it kills five. Instrumenting the five survivors against the reference shows they fail in two distinct ways, not one. Three are never activated, because no fixture supplies an input on which the mutated code behaves differently at all. The other two corrupt internal state that no oracle in the suite can observe. The two modes need different repairs, and neither is visible from a pass/fail report. Treating the missing inputs as a coverage question, we enumerate seven discriminating dimensions of the input space, register in advance which are uncovered and which survivors they should explain, and add one fixture per uncovered dimension while reusing the existing oracles verbatim. All five survivors then die, each to the check written for its predicted dimension. We report this as a repair result on the same challenge set rather than a second held-out estimate, and give the artifact, including the frozen hash, the registered predictions, all mutants and the run logs, so the distinction is checkable.
| Comments: | 10 pages, 1 figure, 2 tables. NeurIPS 2026 Workshop: Who Verifies the Agents? Toward Reliable Agent Development |
| Subjects: | Software Engineering (cs.SE); Artificial Intelligence (cs.AI); Machine Learning (cs.LG) |
| Cite as: | arXiv:2610.02928 [cs.SE] |
| (or arXiv:2610.02928v1 [cs.SE] for this version) | |
| https://doi.org/10.48550/arXiv.2610.02928 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Xin Xu [view email]
[v1]
Fri, 2 Oct 2026 07:17:41 UTC (48 KB)
来源：arXiv：cs.LG · arxiv.org