https://imsj-blog.pages.dev/tags/prompting/Recent content in Prompting on imsjHugoenSun, 09 Aug 2026 00:00:00 +0000https://imsj-blog.pages.dev/posts/control-claude-code/Sun, 09 Aug 2026 00:00:00 +0000https://imsj-blog.pages.dev/posts/control-claude-code/<p><strong>TL;DR:</strong> Set <code>opusplan</code> as your default model, then phrase every task as five things — <em>what you want</em>, <em>what must be preserved</em>, <em>what must not be touched</em>, <em>how to approach it</em>, and <em>what &ldquo;done&rdquo; means</em>. You don&rsquo;t need enormous prompts. You need clear boundaries and a clear definition of done.</p> <figure><img src="https://imsj-blog.pages.dev/posts/control-claude-code/claude-code-best-practices.jpg" alt="Best practices for Claude Code documentation"><figcaption> <p>Anthropic&rsquo;s <a href="https://www.anthropic.com/engineering/claude-code-best-practices">Best practices for Claude Code</a> guide.</p> </figcaption> </figure> <h2 id="steps">Steps</h2> <ol> <li> <p><strong>Default to <code>opusplan</code>.</strong> Opus thinks and plans, Sonnet writes the code. It&rsquo;s the single easiest optimisation for larger coding tasks.</p>