<?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>Developer&#39;s Log on Codeslog</title>
    <link>https://www.codeslog.com/en/series/developers-log/</link>
    <description>Recent content in Developer&#39;s Log on Codeslog</description>
    <image>
      <title>Codeslog</title>
      <url>https://www.codeslog.com/images/default-cover.png?v=20260114T1436</url>
      <link>https://www.codeslog.com/images/default-cover.png?v=20260114T1436</link>
    </image>
    <generator>Hugo -- 0.168.0</generator>
    <language>en-US</language>
    <lastBuildDate>Tue, 28 Jul 2026 00:47:04 +0900</lastBuildDate>
    <atom:link href="https://www.codeslog.com/en/series/developers-log/index.xml" rel="self" type="application/rss+xml" />
    <item>
      <title>Which Pages Do You Audit Out of Thousands?</title>
      <link>https://www.codeslog.com/en/posts/page-sampling-algorithm/</link>
      <pubDate>Tue, 28 Jul 2026 00:47:04 +0900</pubDate>
      <guid>https://www.codeslog.com/en/posts/page-sampling-algorithm/</guid>
      <description>How do you pick which pages to audit out of thousands? A deterministic algorithm on WCAG-EM 2.0 stratified sampling — URL clustering finds representative pages without fetching a page body.</description>
    </item>
    <item>
      <title>The Zombie Task Stuck in &#39;Scanning&#39; — Vercel Deployment Protection Was the Culprit</title>
      <link>https://www.codeslog.com/en/posts/vercel-deployment-protection-zombie-task/</link>
      <pubDate>Fri, 24 Jul 2026 01:45:00 +0900</pubDate>
      <guid>https://www.codeslog.com/en/posts/vercel-deployment-protection-zombie-task/</guid>
      <description>Right after launch, two audits sat stuck &amp;#39;scanning&amp;#39; for hours. Two fixes I was sure of missed, and only the third found the real culprit — Vercel Deployment Protection was blocking an internal call with a 401. A serverless queue debugging story.</description>
    </item>
    <item>
      <title>An Open-Source Web Accessibility Checker — The A11y Check Story</title>
      <link>https://www.codeslog.com/en/posts/a11ychk-accessibility-audit/</link>
      <pubDate>Thu, 23 Jul 2026 13:28:00 +0900</pubDate>
      <guid>https://www.codeslog.com/en/posts/a11ychk-accessibility-audit/</guid>
      <description>Meet A11y Check, an open-source web accessibility checker. It audits against WCAG 2.2 and KWCAG together, finds violations, and hands you an AI-ready fix guide — plus a compliance badge, and why the engine is open source.</description>
    </item>
    <item>
      <title>Presenting at an Accessibility Seminar — A Journey Toward a Happier Web: The ModuWeb Story</title>
      <link>https://www.codeslog.com/en/posts/moduweb-accessibility-seminar/</link>
      <pubDate>Thu, 23 Apr 2026 22:30:00 +0900</pubDate>
      <guid>https://www.codeslog.com/en/posts/moduweb-accessibility-seminar/</guid>
      <description>A recap of presenting ModuWeb, an open-source overlay accessibility tool, at an AI Accessibility seminar in Seoul — plus an honest take on overlay-tool criticism and other speakers&amp;#39; talks.</description>
    </item>
    <item>
      <title>Questions I Asked at the AI Public Service Evaluation</title>
      <link>https://www.codeslog.com/en/posts/ai-evaluation-review/</link>
      <pubDate>Wed, 31 Dec 2025 01:51:00 +0900</pubDate>
      <guid>https://www.codeslog.com/en/posts/ai-evaluation-review/</guid>
      <description>My experience as a citizen evaluator for an AI agent competition. Recording the questions I asked about accessibility, failure response, and communication design.</description>
    </item>
    <item>
      <title>A $35 Evaluation Panel — Why I&#39;m Still Going to Seoul</title>
      <link>https://www.codeslog.com/en/posts/evaluation-panel-seoul/</link>
      <pubDate>Tue, 30 Dec 2025 01:07:37 +0900</pubDate>
      <guid>https://www.codeslog.com/en/posts/evaluation-panel-seoul/</guid>
      <description>A full day of my time, travel paid out of pocket, and just 50,000 won in return — yet I&amp;#39;m heading to Seoul as a citizen evaluator for an AI agent competition. A reflection on why a seemingly losing choice can plant seeds for the future.</description>
    </item>
  </channel>
</rss>
