<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0"><channel><title><![CDATA[代码 agent 做仓库改动，最怕没有验收口径]]></title><description><![CDATA[<p dir="auto">代码 agent 帮改 bug，大家怎么验收？我现在感觉它能改，但经常改得太宽。</p>
]]></description><link>https://localaihub.com/topic/117/代码-agent-做仓库改动-最怕没有验收口径</link><generator>RSS for Node</generator><lastBuildDate>Wed, 03 Jun 2026 19:29:16 GMT</lastBuildDate><atom:link href="https://localaihub.com/topic/117.rss" rel="self" type="application/rss+xml"/><pubDate>Thu, 07 May 2026 18:48:00 GMT</pubDate><ttl>60</ttl><item><title><![CDATA[Reply to 代码 agent 做仓库改动，最怕没有验收口径 on Fri, 08 May 2026 20:12:00 GMT]]></title><description><![CDATA[<p dir="auto">这样现实。代码 agent 不是替代 review，是把改动草稿和验证流水线前移。</p>
]]></description><link>https://localaihub.com/post/1040</link><guid isPermaLink="true">https://localaihub.com/post/1040</guid><dc:creator><![CDATA[Grace]]></dc:creator><pubDate>Fri, 08 May 2026 20:12:00 GMT</pubDate></item><item><title><![CDATA[Reply to 代码 agent 做仓库改动，最怕没有验收口径 on Fri, 08 May 2026 18:12:00 GMT]]></title><description><![CDATA[<p dir="auto">我准备先在非核心 repo 上试，限制目录、限制 diff、必须跑复现测试。</p>
]]></description><link>https://localaihub.com/post/1039</link><guid isPermaLink="true">https://localaihub.com/post/1039</guid><dc:creator><![CDATA[小郑]]></dc:creator><pubDate>Fri, 08 May 2026 18:12:00 GMT</pubDate></item><item><title><![CDATA[Reply to 代码 agent 做仓库改动，最怕没有验收口径 on Fri, 08 May 2026 18:01:00 GMT]]></title><description><![CDATA[<p dir="auto">还有一个坑：agent 为了让测试过，会改测试去适配错误行为。要锁住测试或要求解释测试变更。</p>
]]></description><link>https://localaihub.com/post/1038</link><guid isPermaLink="true">https://localaihub.com/post/1038</guid><dc:creator><![CDATA[melo]]></dc:creator><pubDate>Fri, 08 May 2026 18:01:00 GMT</pubDate></item><item><title><![CDATA[Reply to 代码 agent 做仓库改动，最怕没有验收口径 on Fri, 08 May 2026 15:16:00 GMT]]></title><description><![CDATA[<p dir="auto">别忘了安全。不要让代码 agent 读取 <code>.env</code>、SSH key、生产 kubeconfig。读到了也不该进日志。</p>
]]></description><link>https://localaihub.com/post/1037</link><guid isPermaLink="true">https://localaihub.com/post/1037</guid><dc:creator><![CDATA[nora]]></dc:creator><pubDate>Fri, 08 May 2026 15:16:00 GMT</pubDate></item><item><title><![CDATA[Reply to 代码 agent 做仓库改动，最怕没有验收口径 on Fri, 08 May 2026 13:19:00 GMT]]></title><description><![CDATA[<p dir="auto">我们让 agent 每次提交前跑 <code>git diff --stat</code> 和相关测试，最后写变更摘要。还不错。</p>
]]></description><link>https://localaihub.com/post/1036</link><guid isPermaLink="true">https://localaihub.com/post/1036</guid><dc:creator><![CDATA[小曹]]></dc:creator><pubDate>Fri, 08 May 2026 13:19:00 GMT</pubDate></item><item><title><![CDATA[Reply to 代码 agent 做仓库改动，最怕没有验收口径 on Fri, 08 May 2026 11:26:00 GMT]]></title><description><![CDATA[<p dir="auto">执行日志要包含命令输出摘要。只写“测试通过”不够，至少要有命令名、失败点、修复后结果。</p>
]]></description><link>https://localaihub.com/post/1035</link><guid isPermaLink="true">https://localaihub.com/post/1035</guid><dc:creator><![CDATA[小唐]]></dc:creator><pubDate>Fri, 08 May 2026 11:26:00 GMT</pubDate></item><item><title><![CDATA[Reply to 代码 agent 做仓库改动，最怕没有验收口径 on Fri, 08 May 2026 08:26:00 GMT]]></title><description><![CDATA[<p dir="auto">长任务必须有恢复点。agent 改一半断了，下次要知道已验证什么、没验证什么，不然重复探索。</p>
]]></description><link>https://localaihub.com/post/1034</link><guid isPermaLink="true">https://localaihub.com/post/1034</guid><dc:creator><![CDATA[林小北]]></dc:creator><pubDate>Fri, 08 May 2026 08:26:00 GMT</pubDate></item><item><title><![CDATA[Reply to 代码 agent 做仓库改动，最怕没有验收口径 on Fri, 08 May 2026 06:50:00 GMT]]></title><description><![CDATA[<p dir="auto">我倒觉得架构也能让它参与，但只能出方案和风险，不能直接落地到主干。</p>
]]></description><link>https://localaihub.com/post/1033</link><guid isPermaLink="true">https://localaihub.com/post/1033</guid><dc:creator><![CDATA[陈一]]></dc:creator><pubDate>Fri, 08 May 2026 06:50:00 GMT</pubDate></item><item><title><![CDATA[Reply to 代码 agent 做仓库改动，最怕没有验收口径 on Fri, 08 May 2026 06:28:00 GMT]]></title><description><![CDATA[<p dir="auto">代码 agent 很适合做机械但有上下文的活，比如补测试、迁移 API 调用、修 lint。复杂架构决策还是要人盯。</p>
]]></description><link>https://localaihub.com/post/1032</link><guid isPermaLink="true">https://localaihub.com/post/1032</guid><dc:creator><![CDATA[半截薯条]]></dc:creator><pubDate>Fri, 08 May 2026 06:28:00 GMT</pubDate></item><item><title><![CDATA[Reply to 代码 agent 做仓库改动，最怕没有验收口径 on Fri, 08 May 2026 03:32:00 GMT]]></title><description><![CDATA[<p dir="auto">这个有用。还有禁止改锁文件，除非任务就是依赖升级。</p>
]]></description><link>https://localaihub.com/post/1031</link><guid isPermaLink="true">https://localaihub.com/post/1031</guid><dc:creator><![CDATA[qwer_asdf]]></dc:creator><pubDate>Fri, 08 May 2026 03:32:00 GMT</pubDate></item><item><title><![CDATA[Reply to 代码 agent 做仓库改动，最怕没有验收口径 on Fri, 08 May 2026 02:30:00 GMT]]></title><description><![CDATA[<p dir="auto">我们加了 diff budget。超过文件数或行数要说明原因，不然直接打回。</p>
]]></description><link>https://localaihub.com/post/1030</link><guid isPermaLink="true">https://localaihub.com/post/1030</guid><dc:creator><![CDATA[小傅]]></dc:creator><pubDate>Fri, 08 May 2026 02:30:00 GMT</pubDate></item><item><title><![CDATA[Reply to 代码 agent 做仓库改动，最怕没有验收口径 on Fri, 08 May 2026 00:49:00 GMT]]></title><description><![CDATA[<p dir="auto">我最怕“顺手重构”。一个小 bug 改 20 个文件，review 成本比自己改还高。</p>
]]></description><link>https://localaihub.com/post/1029</link><guid isPermaLink="true">https://localaihub.com/post/1029</guid><dc:creator><![CDATA[阿白]]></dc:creator><pubDate>Fri, 08 May 2026 00:49:00 GMT</pubDate></item><item><title><![CDATA[Reply to 代码 agent 做仓库改动，最怕没有验收口径 on Thu, 07 May 2026 22:25:00 GMT]]></title><description><![CDATA[<p dir="auto">SWE-agent 这类项目有个启发：让 agent 在真实 repo 里循环读文件、改文件、跑测试。但测试缺失时，它也会乱猜。</p>
]]></description><link>https://localaihub.com/post/1028</link><guid isPermaLink="true">https://localaihub.com/post/1028</guid><dc:creator><![CDATA[rootless]]></dc:creator><pubDate>Thu, 07 May 2026 22:25:00 GMT</pubDate></item><item><title><![CDATA[Reply to 代码 agent 做仓库改动，最怕没有验收口径 on Thu, 07 May 2026 20:02:00 GMT]]></title><description><![CDATA[<p dir="auto">给它 issue 不够，要给验收标准。比如失败测试、复现步骤、不能动的模块、期望行为。</p>
]]></description><link>https://localaihub.com/post/1027</link><guid isPermaLink="true">https://localaihub.com/post/1027</guid><dc:creator><![CDATA[Grace]]></dc:creator><pubDate>Thu, 07 May 2026 20:02:00 GMT</pubDate></item></channel></rss>