<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Agent on jimyag's Blog</title><link>https://jimyag.com/blog/agent/</link><description>jimyag 的个人博客，记录 Kubernetes、KubeVirt、Kube-OVN、Firecracker、Go 与云原生工程实践。</description><generator>Hugo</generator><language>zh-CN</language><copyright>Copyright © 2019, jimyag.</copyright><lastBuildDate>Wed, 30 Sep 2026 22:27:54 +0800</lastBuildDate><atom:link href="https://jimyag.com/blog/agent/index.xml" rel="self" type="application/rss+xml"/><item><title>Sandbox 之外，Agent 安全还需要什么：读一位 OpenAI 安全工程师的长文</title><link>https://jimyag.com/posts/ai-capability-jump-security-what-to-change/</link><pubDate>Wed, 30 Sep 2026 22:03:57 +0800</pubDate><guid>https://jimyag.com/posts/ai-capability-jump-security-what-to-change/</guid><description>结合 OpenAI 安全工程师 Joe 的个人文章与 DeepSeek DSec 论文中的案例，讨论 Agent 的隔离、工具权限、评测完整性、审计证据和事故响应。</description></item></channel></rss>