guardscope：测量 LLM 智能体系统中运行时控制实际检查了哪些行为
guardscope: Measuring Which Actions Runtime Controls Check in LLM Agent Systems
AI 导读
研究者发布 guardscope 框架，让企业运行时控制（如 agent gateway）对 LLM 智能体工具调用行为的检查覆盖率可被验证。该框架分两阶段：静态阶段预测控制可从源头检查 15 条行动路径中的哪些，实证阶段在 8 种拓扑设计的部署上运行固定工作负载，结合内核级真值将每个行为分类为已检查、绕过或被阻止。
正文 · 原文
Abstract:Large-language model (LLM) agents execute actions on external systems through tool calls, and enterprises deploy runtime controls, such as agent gateways, to check this behavior in real time. However, a control can only evaluate the actions that cross the point where it is placed, and an agent can act through more paths than the tool-call channel; no existing work measures this coverage against kernel-level ground truth. In this paper, we present guardscope, a framework that makes this coverage measurable. guardscope works in two stages: a static stage that predicts which of an agent's 15 action paths a control can check from its source, and an empirical stage that runs a fixed workload on a deployment that realizes one of eight topology designs and classifies every executed action as checked, bypassed, or prevented by reconciling the control's decision records against kernel-level ground truth. guardscope makes a control's coverage claim verifiable: a runtime control developer can publish which paths its product inspects under a kernel-verified protocol. We conducted experiments with two agent harnesses and real runtime controls, including Microsoft's agent governance toolkit and agentgateway. guardscope is available at this https URL.
| Comments: | Code is available at this https URL |
| Subjects: | Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2611.15223 [cs.AI] |
| (or arXiv:2611.15223v1 [cs.AI] for this version) | |
| https://doi.org/10.48550/arXiv.2611.15223 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Kentaroh Toyoda [view email]
[v1]
Wed, 11 Nov 2026 09:16:29 UTC (487 KB)
来源：arXiv cs.AI · arxiv.org