https://xuezhaojun.github.io/Recent content on Zhao XueHugoen-usSat, 18 Apr 2026 00:00:00 +0000https://xuezhaojun.github.io/collections/go-concurrency/go-goroutine-gmp/Sat, 18 Apr 2026 00:00:00 +0000https://xuezhaojun.github.io/collections/go-concurrency/go-goroutine-gmp/不讲干巴巴的八股文。从一个真实的 Kubernetes controller 内存暴涨 + 延迟飙高的线上事故出发，用 pprof 数据倒推 GMP 调度模型的每个概念，让你理解为什么要学这些东西。https://xuezhaojun.github.io/collections/go-concurrency/go-channel/Sat, 18 Apr 2026 00:00:00 +0000https://xuezhaojun.github.io/collections/go-concurrency/go-channel/一个 K8s admission webhook 在高峰期频繁超时，但单个请求处理逻辑明明很快。问题出在 channel 的使用方式上。从这个事故出发，拆解 channel 的底层结构、发送接收流程、select 实现，以及那些年我们踩过的 channel 坑。https://xuezhaojun.github.io/posts/repo-as-agent-en/Wed, 15 Apr 2026 00:00:00 +0000https://xuezhaojun.github.io/posts/repo-as-agent-en/What if an AI Agent’s identity, skills, and knowledge were all just files in a Git repository? This post introduces a pattern we’ve used in practice — no new frameworks, just directories, files, and Git to build production-grade AI Agents.https://xuezhaojun.github.io/posts/repo-as-agent/Wed, 15 Apr 2026 00:00:00 +0000https://xuezhaojun.github.io/posts/repo-as-agent/如果一个 AI Agent 的身份、技能和知识全都是 Git 仓库里的文件呢？本文介绍一种我们实践过的模式——不引入任何新框架，只用目录、文件和 Git 来构建生产级 AI Agent。https://xuezhaojun.github.io/resume/Mon, 01 Jan 0001 00:00:00 +0000https://xuezhaojun.github.io/resume/Red Hat 项目经历详述https://xuezhaojun.github.io/about/Mon, 01 Jan 0001 00:00:00 +0000https://xuezhaojun.github.io/about/About mehttps://xuezhaojun.github.io/about-zh/Mon, 01 Jan 0001 00:00:00 +0000https://xuezhaojun.github.io/about-zh/个人简介