<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:content="http://purl.org/rss/1.0/modules/content/"><channel><title>Spec-Driven-Development on Mohanad Abu-Nayla</title><link>https://hannody.github.io/tags/spec-driven-development/</link><description>Recent content in Spec-Driven-Development on Mohanad Abu-Nayla</description><generator>Hugo</generator><language>en-US</language><lastBuildDate>Sat, 01 Aug 2026 00:00:00 +0000</lastBuildDate><atom:link href="https://hannody.github.io/tags/spec-driven-development/index.xml" rel="self" type="application/rss+xml"/><item><title>Spec-Driven Development with AI Agents: Constitutions, Checkpoints, and Handoffs</title><link>https://hannody.github.io/posts/agents/spec-driven-development-with-ai-agents/</link><pubDate>Sat, 01 Aug 2026 00:00:00 +0000</pubDate><guid>https://hannody.github.io/posts/agents/spec-driven-development-with-ai-agents/</guid><description>&lt;p&gt;AI coding agents are strangely lopsided. Hand one a well-scoped task and it writes better code, faster, than I do. Hand it a &lt;em&gt;large&lt;/em&gt; feature, the kind that takes days and touches a dozen files across several layers, and the failure mode changes. The context window fills up and quality quietly degrades (AKA &lt;strong&gt;context rot&lt;/strong&gt;). A fresh session, missing the history, confidently invents an API that never existed. And even when the code is good, you&amp;rsquo;re left staring at a 3,000-line diff you can&amp;rsquo;t meaningfully review.&lt;/p&gt;</description></item></channel></rss>