<?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>SWE Agent on MessageDaily</title><link>https://inkeast.github.io/MessageDaily/tags/swe-agent/</link><description>Recent content in SWE Agent on MessageDaily</description><generator>Hugo</generator><language>zh-cn</language><lastBuildDate>Fri, 11 Sep 2026 00:00:00 +0000</lastBuildDate><atom:link href="https://inkeast.github.io/MessageDaily/tags/swe-agent/index.xml" rel="self" type="application/rss+xml"/><item><title>XAgent: eXecution-guided Agentic AI for GitHub Issues 精读</title><link>https://inkeast.github.io/MessageDaily/posts/2026-09-11-xagent-github-issues-paper-reading/</link><pubDate>Fri, 11 Sep 2026 00:00:00 +0000</pubDate><guid>https://inkeast.github.io/MessageDaily/posts/2026-09-11-xagent-github-issues-paper-reading/</guid><description>XAgent 把仓库级 issue 解决从&amp;rsquo;读描述猜位置&amp;rsquo;升级为&amp;rsquo;跑起来看分歧&amp;rsquo;：执行引导定位对 buggy/非 buggy 版本做差分执行分析+树编辑距离定位行为分歧点，上下文感知测试增强器生成超越 issue 描述的边缘用例验证补丁。SWE-bench-Lite 上 resolve rate 62.0%（SOTA）、函数级定位准确率 72.8%、每 issue 成本 $1.56（比前 SOTA 便宜 37%）、token 省 40%，并解决 7 个所有顶级基线（SWE-Agent/EXPEREPAIR/Refact）全部失败的问题。本文精读&amp;rsquo;静态描述→动态执行&amp;rsquo;范式转移的机制细节。</description></item></channel></rss>