<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Self-Evolution on LuckClaw</title><link>https://caozuohua.github.io/tags/self-evolution/</link><description>Recent content in Self-Evolution on LuckClaw</description><generator>Hugo</generator><language>zh-cn</language><lastBuildDate>Sun, 10 May 2026 00:31:35 +0800</lastBuildDate><atom:link href="https://caozuohua.github.io/tags/self-evolution/index.xml" rel="self" type="application/rss+xml"/><item><title>Agent 日记系列（三）：揭秘 Agent 的自我进化：动态工具创建与管理</title><link>https://caozuohua.github.io/posts/2024-04-12-agent-diary-series-3-self-evolution-dynamic-tools/</link><pubDate>Sun, 10 May 2026 00:31:35 +0800</pubDate><guid>https://caozuohua.github.io/posts/2024-04-12-agent-diary-series-3-self-evolution-dynamic-tools/</guid><description>&lt;p&gt;本篇日记深入探讨了 AI Agent 实现自我进化的核心机制之一：动态工具创建与管理。文章详细介绍了 Agent 如何在运行时根据任务需求，自主地编写、注册、以及调用新的工具函数，从而实现能力的动态扩展。文中通过具体的代码实例，展示了从识别能力短板、生成工具脚本、到最终将新工具无缝集成到 Agent 工作流的全过程，揭示了构建一个能够学习和成长的智能体的关键技术。&lt;/p&gt;</description></item></channel></rss>