<?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>Weave-Router on Liuji's Blog</title><link>https://liuji.cc/tags/weave-router/</link><description>Recent content in Weave-Router on Liuji's Blog</description><generator>Hugo</generator><language>zh-cn</language><lastBuildDate>Sat, 27 Jun 2026 10:00:00 +0800</lastBuildDate><atom:link href="https://liuji.cc/tags/weave-router/index.xml" rel="self" type="application/rss+xml"/><item><title>有人给 Claude Code 加了一个大脑：Weave Router 让 coding agent 自动选最便宜的模型</title><link>https://liuji.cc/posts/weave-router-coding-agent-cost/</link><pubDate>Sat, 27 Jun 2026 10:00:00 +0800</pubDate><guid>https://liuji.cc/posts/weave-router-coding-agent-cost/</guid><description>Weave Router 是一个代理层，插在 coding agent 和模型之间，50ms 内动态路由每条请求到最合适的模型。内部实测省了 40% token 成本，质量无明显差异。</description></item></channel></rss>