<?xml version="1.0" encoding="UTF-8"?><oembed><type>video</type><version>1.0</version><html>&lt;iframe src=&quot;https://www.loom.com/embed/921f92e07fed4b04b15634ed838f7455&quot; frameborder=&quot;0&quot; width=&quot;1920&quot; height=&quot;1440&quot; webkitallowfullscreen mozallowfullscreen allowfullscreen&gt;&lt;/iframe&gt;</html><height>1440</height><width>1920</width><provider_name>Loom</provider_name><provider_url>https://www.loom.com</provider_url><thumbnail_height>1440</thumbnail_height><thumbnail_width>1920</thumbnail_width><thumbnail_url>https://cdn.loom.com/sessions/thumbnails/921f92e07fed4b04b15634ed838f7455-4397163427a1707a.gif</thumbnail_url><duration>5516.88</duration><title>Building Prompts and Skills for AI Agents</title><description>This Loom explains how to build better prompts and turn them into reusable Claude skills using a critical thinking framework. The speaker distinguishes prompts as one time instructions versus skills as reusable “skills” taught to Claude, and walks through Claude settings for short global instructions plus an “effort control” update in Opus 4.8 with guidance to use higher effort only for pressure testing. A live walkthrough uses a prompt builder with six questions (output, purpose, constraints, goal, anti goal, example), then applies “cheat codes” like clarifying questions, pre mortems, and steel man and red team style pushback. The session emphasizes validating prompts by iterating through several versions before creating skills, and provides a skill audit concept for ongoing improvement.</description></oembed>