<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Regression Testing on Andreas Nissen</title><link>https://andreasnissen.dev/tags/regression-testing/</link><description>Recent content in Regression Testing on Andreas Nissen</description><generator>Hugo</generator><language>en-us</language><copyright>© Andreas Nissen</copyright><lastBuildDate>Mon, 31 Aug 2026 00:00:00 +0000</lastBuildDate><atom:link href="https://andreasnissen.dev/tags/regression-testing/index.xml" rel="self" type="application/rss+xml"/><item><title>Skill Routing Is a Reliability Problem</title><link>https://andreasnissen.dev/writing/skill-routing-is-reliability/</link><pubDate>Mon, 31 Aug 2026 00:00:00 +0000</pubDate><guid>https://andreasnissen.dev/writing/skill-routing-is-reliability/</guid><description>&lt;p&gt;A skill can contain excellent instructions and still make the agent less reliable.&lt;/p&gt;&#10;&lt;p&gt;If it activates for the wrong request, competes with a nearby skill, loads stale policy, or assumes permissions the task does not grant, the failure begins before the workflow runs.&lt;/p&gt;&#10;&lt;p&gt;That makes skill routing an evaluation problem.&lt;/p&gt;&#10;&lt;h2 id="the-description-is-an-interface"&gt;The description is an interface&lt;/h2&gt;&#10;&lt;p&gt;Most agent harnesses select a skill before loading its full instructions. The selector often sees a name and short description, then decides which deeper workflow deserves context.&lt;/p&gt;</description></item></channel></rss>