# New Papers Target the Weakest Point in Agent Deployments: Cross-Domain Tool Abuse in the Model Context Protocol

One arXiv submission proposes a control layer called AEGIS for resource abuse across trust domains, a second argues that end-to-end success rates hide where long-horizon security agents actually fail.

- Published: 2026-08-24T07:20:23.472Z
- Canonical: https://polylog.news/ai/2026-08-24/new-papers-target-the-weakest-point-in-agent-deployments-cro
- Publisher: Polylog (AI desk)
- Section: tech
- Sources: [arXiv cs.CR (AEGIS)](https://arxiv.org/abs/2608.20481), [arXiv cs.CR (long-horizon security agents)](https://arxiv.org/abs/2608.20563), [schlarp.com](https://schlarp.com/posts/everything-i-own-owned/)

Two submissions posted to arXiv on 24 August address the same structural weakness using two different approaches. AEGIS targets cross-domain resource abuse in the Model Context Protocol (MCP), the open JSON-RPC standard that lets large lang…

This story is for subscribers. Read it in full at https://polylog.news/ai/2026-08-24/new-papers-target-the-weakest-point-in-agent-deployments-cro (subscription information: https://polylog.news/pricing).