0. How AI Broke the Technical Manual (What Changed in the Claude Ecosystem)

<p>In this special preamble episode of <strong>Beyond Prompting</strong>, we examine how rapid AI development forced a complete rewrite of technical documentation and operator manuals. Author Sho Shimoda breaks down why <em>Running Claude</em> replaces earlier guides following roughly 180 software releases in just six months. We explore four fundamental shifts reshaping the operational ecosystem:</p><ul><ul><li><strong>Model Tiers &amp; Effort:</strong> The lineup now spans five tiers—Mythos, Fable, Opus, Sonnet, and Haiku—with native 1M-token context windows and an "Effort" setting replacing traditional temperature sampling dials.</li></ul><ul><li><strong>Cloud-First Cowork:</strong> Cowork sessions now execute in sandboxed cloud virtual machines, turning the desktop application into a local broker for disk, browser, and screen access.</li></ul><ul><li><strong>Stateless MCP:</strong> The updated Model Context Protocol (2026-07-28) removed connection handshakes and persistent sessions so every request stands alone.</li></ul><ul><li><strong>Auto Mode &amp; Agent SDK:</strong> Default permission postures transitioned to classifier-driven "Auto Mode", while the Agent SDK allows developers to embed execution loops directly into custom applications.</li></ul></ul><p>Finally, we discuss why operating AI requires verifiable code repositories, live errata tracking, and a steadfast core philosophy: <strong>execution over explanation, context over scale, and workflow over conversation</strong>.<em>(Note for listeners: This episode covers the Preface of Sho Shimoda's book </em><strong>RUNNING CLAUDE: The Operator’s Guide to Chat, Cowork and Claude Code</strong><em>, which is available on Amazon as the successor to </em><strong>Master Claude: Chat, Cowork and Code</strong><em>).</em></p>