<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:content="http://purl.org/rss/1.0/modules/content/"><channel><title>Agent-Eval on 兜忆轩 | Cedric's Compiler Notes</title><link>https://douyixuan.github.io/tags/agent-eval/</link><description>Recent content in Agent-Eval on 兜忆轩 | Cedric's Compiler Notes</description><image><title>兜忆轩 | Cedric's Compiler Notes</title><url>https://douyixuan.github.io/images/papermod-cover.png</url><link>https://douyixuan.github.io/images/papermod-cover.png</link></image><generator>Hugo -- 0.146.0</generator><language>en-us</language><lastBuildDate>Wed, 07 Oct 2026 08:00:00 +0800</lastBuildDate><atom:link href="https://douyixuan.github.io/tags/agent-eval/index.xml" rel="self" type="application/rss+xml"/><item><title>AI Native Digest 2026-10-07</title><link>https://douyixuan.github.io/posts/ai-native/2026-10-07/</link><pubDate>Wed, 07 Oct 2026 08:00:00 +0800</pubDate><guid>https://douyixuan.github.io/posts/ai-native/2026-10-07/</guid><description>今天关注 AI-native 软件工程从“Agent 能写代码”走向“机器速度下可治理、可验证”的两个方向：GitLab 把 agentic workflow、上下文、供应链安全、成本与证据链整合成 governed software factory；Wiz 则展示了如何用 harness、持续 benchmark 和稳定 finding lifecycle，把非确定性的 AI 代码分析变成可运营的工程系统。</description></item><item><title>AI Native Digest 2026-10-06</title><link>https://douyixuan.github.io/posts/ai-native/2026-10-06/</link><pubDate>Tue, 06 Oct 2026 08:00:00 +0800</pubDate><guid>https://douyixuan.github.io/posts/ai-native/2026-10-06/</guid><description>今天关注 Agent 从“能跑一次”走向可长期运行的两个基础能力：GitHub Agentic Workflows 用 work queue、ledger、safe outputs 和审计把自治工作流做成可恢复、可治理的系统；ThinkingBox 则用最终数据库状态和 20 次重复运行检验 Agent 是否真的可靠完成任务。</description></item></channel></rss>