<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Arxiv-2606-26978 on Text Matrix</title><link>https://txtmix.com/tags/arxiv-2606-26978/</link><description>Recent content in Arxiv-2606-26978 on Text Matrix</description><generator>Hugo</generator><language>zh-cn</language><lastBuildDate>Thu, 23 Jul 2026 03:05:08 +0800</lastBuildDate><atom:link href="https://txtmix.com/tags/arxiv-2606-26978/index.xml" rel="self" type="application/rss+xml"/><item><title>To Run or Not to Run——ISSTA 2026 论文深度解读，LLM 程序修复的代码执行成本收益分析</title><link>https://txtmix.com/posts/tech/arxiv-2606-26978-code-execution-cost-effectiveness-llm-program-repair/</link><pubDate>Tue, 30 Jun 2026 20:56:00 +0800</pubDate><guid>https://txtmix.com/posts/tech/arxiv-2606-26978-code-execution-cost-effectiveness-llm-program-repair/</guid><description>&lt;h2 id="译序为什么这篇-issta-2026-论文值得完整读">译序：为什么这篇 ISSTA 2026 论文值得完整读&lt;/h2>
&lt;p>论文标题就一句话：&lt;strong>To Run or Not to Run&lt;/strong>——LLM 编程 Agent 是否应该默认执行代码？&lt;/p>
&lt;p>这是 2026 年所有 AI 编程助手（Claude Code / Codex / OpenCode / Cursor / OpenHands / SWE-agent）都默认开启的能力。你问任何 AI 工程师，TA 都会说&amp;quot;执行测试反馈是 agent 修复 bug 的关键&amp;quot;。&lt;/p></description></item></channel></rss>