<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>SKILL.md on StackOnward</title><link>https://stackonward.com/tags/skill.md/</link><description>Recent content in SKILL.md on StackOnward</description><generator>Hugo</generator><language>en-US</language><lastBuildDate>Fri, 11 Sep 2026 04:52:00 +0800</lastBuildDate><atom:link href="https://stackonward.com/tags/skill.md/index.xml" rel="self" type="application/rss+xml"/><item><title>How to Write Claude Code Skills: Repeatable Workflows</title><link>https://stackonward.com/posts/agent-skill-project-workflow/</link><pubDate>Fri, 11 Sep 2026 04:40:00 +0800</pubDate><guid>https://stackonward.com/posts/agent-skill-project-workflow/</guid><description>&lt;p&gt;How to write Claude Code skills depends on whether the flow repeats across tasks, whether it still needs on-the-spot judgment, and whether the project can observe a result after it runs. Facts that stay true in a repository belong in &lt;a href="https://stackonward.com/posts/agents-md-claude-md-project-rules/"&gt;project rules&lt;/a&gt;. A one-off requirement belongs in the &lt;a href="https://stackonward.com/posts/coding-agent-task-prompt/"&gt;current task prompt&lt;/a&gt;. Actions that must run on an event belong in a hook. External services and data come from MCP or other tools.&lt;/p&gt;
&lt;p&gt;A full-stack project that often changes an existing API contract has to find the real producer, confirm approved field meanings, sync direct consumers, run existing checks, and observe a page or a call. The fields and the range change every time. The judgment order stays stable. That flow belongs in a project skill. The field names, paths, and acceptance result for this run still come from the current prompt.&lt;/p&gt;</description></item></channel></rss>