<?xml version="1.0" encoding="UTF-8"?><rss version="2.0"><channel><title>树哥的地下室</title><description>林榕健的个人博客，记录 AI、数学、独立开发与生活。</description><link>https://machinegunlin.github.io/</link><language>zh-CN</language><item><title>Stanford CS146S Week 5：AI 增强命令行与终端自动化</title><link>https://machinegunlin.github.io/posts/cs146s-week5-ai-terminal/</link><guid isPermaLink="true">https://machinegunlin.github.io/posts/cs146s-week5-ai-terminal/</guid><description>从 AI 为什么回到终端，到 Warp 与 CLI Agent 的差别，梳理 Block、Workflow、多模型和可视化 diff 等能力。</description><pubDate>Wed, 23 Sep 2026 00:00:00 GMT</pubDate><category>AI 与技术</category><category>AI</category><category>Agent</category><category>Stanford</category></item><item><title>Stanford CS146S Week 4：Agent 自主程度与人机协作模式</title><link>https://machinegunlin.github.io/posts/cs146s-week4-agent-autonomy/</link><guid isPermaLink="true">https://machinegunlin.github.io/posts/cs146s-week4-agent-autonomy/</guid><description>从任务何时适合交给 Agent、如何用可验证的检查形成闭环，到 Subagent 隔离、对抗性审查与 Claude Code 的 system reminder，整理 Stanford CS146S Week 4 的阅读与作业实践。</description><pubDate>Tue, 22 Sep 2026 00:00:00 GMT</pubDate><category>AI 与技术</category><category>AI</category><category>Agent</category><category>Stanford</category></item><item><title>Stanford CS146S Week 3：AI 接管执行之后，人的判断力去哪了</title><link>https://machinegunlin.github.io/posts/cs146s-week3-human-judgment/</link><guid isPermaLink="true">https://machinegunlin.github.io/posts/cs146s-week3-human-judgment/</guid><description>从规格、上下文管理到工具设计，复盘 Stanford CS146S Week 3：AI 接管执行后，人如何把判断力放在更高杠杆的研究、规划与边界上，并记录股票行情 MCP Server 实作。</description><pubDate>Sun, 20 Sep 2026 00:00:00 GMT</pubDate><category>AI 与技术</category><category>AI</category><category>Agent</category><category>Stanford</category></item><item><title>Stanford CS146S Week 2：MCP 架构入门</title><link>https://machinegunlin.github.io/posts/cs146s-week2-mcp-architecture/</link><guid isPermaLink="true">https://machinegunlin.github.io/posts/cs146s-week2-mcp-architecture/</guid><description>从 Host、Client、Server 的分工，到 Tools、Resources、Prompts 三种 Primitive，梳理 MCP 怎样连接 AI 应用与外部能力，并记录 Week 2 的 Action Item Extractor 实作。</description><pubDate>Thu, 17 Sep 2026 00:00:00 GMT</pubDate><category>AI 与技术</category><category>AI</category><category>MCP</category><category>Stanford</category></item><item><title>Stanford CS146S Week 1：LLM 基础原理入门</title><link>https://machinegunlin.github.io/posts/cs146s-week1-llm-basics/</link><guid isPermaLink="true">https://machinegunlin.github.io/posts/cs146s-week1-llm-basics/</guid><description>跟着 Stanford CS146S 这门课的节奏，第一周学 LLM 基础——预训练、监督微调（SFT）、强化学习（RL）是怎么一步步把模型训练出来的，附六个动手练习的实际踩坑记录。</description><pubDate>Sat, 12 Sep 2026 00:00:00 GMT</pubDate><category>AI 与技术</category><category>AI</category><category>LLM</category><category>Stanford</category></item><item><title>欢迎来到树哥的地下室</title><link>https://machinegunlin.github.io/posts/welcome/</link><guid isPermaLink="true">https://machinegunlin.github.io/posts/welcome/</guid><description>这里会记录 AI、数学、独立开发与生活，也会成为这些文字长期存放的地方。</description><pubDate>Mon, 31 Aug 2026 00:00:00 GMT</pubDate><category>思考与生活</category><category>关于本站</category></item></channel></rss>