<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:dc="http://purl.org/dc/elements/1.1/">
  <channel>
    <title>Notes · Harry Zhao (English)</title>
    <link>https://me.tech-harry.top/blog/index.en.html</link>
    <description>Engineering notes on automotive embedded software, AutoSAR toolchains, and putting AI agents into real engineering workflows.</description>
    <language>en</language>
    <lastBuildDate>Fri, 06 Nov 2026 11:00:00 +0800</lastBuildDate>
    <atom:link href="https://me.tech-harry.top/blog/feed-en.xml" rel="self" type="application/rss+xml"/>

    <item>
      <title>Not configured looks exactly like no problem at all</title>
      <link>https://me.tech-harry.top/blog/cdd-arxml-consistency.en.html</link>
      <guid isPermaLink="true">https://me.tech-harry.top/blog/cdd-arxml-consistency.en.html</guid>
      <pubDate>Fri, 06 Nov 2026 11:00:00 +0800</pubDate>
      <dc:creator>Harry Zhao</dc:creator>
      <category>AUTOSAR</category>
      <category>UDS</category>
      <category>CDD</category>
      <category>Diagnostics</category>
      <description>Building a CDD x ARXML diagnostic consistency checker: eight checks skipped themselves in silence while the report still read like a complete review, because "the report never mentioned CAN ids" and "the CAN ids are fine" are the same string on a terminal. Plus how the hardware acceptance filter decides which of the two files is wrong, and why a template default must never be treated as configuration.</description>
    </item>

    <item>
      <title>Three paths for one CAN frame</title>
      <link>https://me.tech-harry.top/blog/com-send-receive-chain.en.html</link>
      <guid isPermaLink="true">https://me.tech-harry.top/blog/com-send-receive-chain.en.html</guid>
      <pubDate>Thu, 05 Nov 2026 19:00:00 +0800</pubDate>
      <dc:creator>Harry Zhao</dc:creator>
      <category>AUTOSAR</category>
      <category>Com</category>
      <category>CAN</category>
      <description>Reading a PuHua AutoSAR CP build: sending, transmit confirmation and reception are three independent paths driven by three different main functions. Confirmation is not nested inside the send call. Includes the one PduR boolean that decides whether the loop closes.</description>
    </item>

    <item>
      <title>AI search engines could not see me</title>
      <link>https://me.tech-harry.top/blog/geo-audit.en.html</link>
      <guid isPermaLink="true">https://me.tech-harry.top/blog/geo-audit.en.html</guid>
      <pubDate>Thu, 05 Nov 2026 18:00:00 +0800</pubDate>
      <dc:creator>Harry Zhao</dc:creator>
      <category>GEO</category>
      <category>Structured Data</category>
      <category>AI Search</category>
      <description>A GEO audit of my own blog and GitHub repository: the English half did not exist as far as a crawler was concerned, structured data scored 12, s-maxage did nothing on Workers, and a real bug fix moved the score by zero. The last 17 points cannot be reached by writing code.</description>
    </item>

    <item>
      <title>105 errors. 99 of them were not real.</title>
      <link>https://me.tech-harry.top/blog/arxml-dbc-consistency.en.html</link>
      <guid isPermaLink="true">https://me.tech-harry.top/blog/arxml-dbc-consistency.en.html</guid>
      <pubDate>Thu, 05 Nov 2026 00:00:00 +0800</pubDate>
      <dc:creator>Harry Zhao</dc:creator>
      <category>AUTOSAR</category>
      <category>CAN FD</category>
      <category>Static analysis</category>
      <description>Building an AUTOSAR ARXML × DBC consistency checker for an AI agent. The first version reported 105 errors on a real project; checking them one by one, 99 were false — the matrix described a whole 11-node network and most frames had nothing to do with this ECU. False positives are worse than misses: nobody reads past line one hundred.</description>
    </item>

    <item>
      <title>Designing an engineering-grade skill for an AI agent</title>
      <link>https://me.tech-harry.top/blog/dbc-matrix-audit-skill.en.html</link>
      <guid isPermaLink="true">https://me.tech-harry.top/blog/dbc-matrix-audit-skill.en.html</guid>
      <pubDate>Wed, 04 Nov 2026 00:00:00 +0800</pubDate>
      <dc:creator>Harry Zhao</dc:creator>
      <category>AI Agent</category>
      <category>Skill engineering</category>
      <category>AUTOSAR</category>
      <description>Requirement boundaries, architecture, skill shape and quality system: the complete design process behind an AUTOSAR DBC audit skill, with a reusable five-step method.</description>
    </item>
  </channel>
</rss>
