<?xml version="1.0" encoding="UTF-8"?><rss version="2.0"><channel><title>Yi Liu</title><description>Welcome to my posts!</description><link>https://yl.do/</link><item><title>Prompt Engineering: A Practical Guide From One Sentence to a Testable Model Interface</title><link>https://yl.do/en/posts/prompt-engineering-guide/</link><guid isPermaLink="true">https://yl.do/en/posts/prompt-engineering-guide/</guid><description>Starting from a support-ticket classification case, this covers how to design prompts, which frameworks to reach for, how to evaluate them, and how retrieval, reasoning, agents, and safety boundaries plug into the basic pipeline.</description></item><item><title>Context Engineering: Helping an Agent See What It Needs to See Across Many Turns</title><link>https://yl.do/en/posts/context-engineering/</link><guid isPermaLink="true">https://yl.do/en/posts/context-engineering/</guid><description>Using a refund agent as a running example, this covers what context engineering manages, how it differs from prompt engineering, and how to write, select, compress, and isolate context in practice.</description></item><item><title>Markdown Syntax Guide</title><link>https://yl.do/en/posts/markdown-syntax-guide/</link><guid isPermaLink="true">https://yl.do/en/posts/markdown-syntax-guide/</guid><description>A complete overview of the Markdown, MDX, code block, math, callout, and card syntax supported by this blog.</description></item><item><title>Jev: When AI Stops Talking and Starts Judging</title><link>https://yl.do/en/posts/jev/</link><guid isPermaLink="true">https://yl.do/en/posts/jev/</guid><description>TypeSafe AI just released Jev, a model that skips text generation and turns program state directly into a structured judgment. What it solves, and how it relates to Structured Output and agents.</description></item></channel></rss>