<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel>
<title>Cauld &amp; Clark · Engineering notes</title><link>https://cauldnclark.com/blogs</link>
<description>How we build software, agents, and the context they need.</description><language>en</language>
<atom:link href="https://cauldnclark.com/blogs/rss.xml" rel="self" type="application/rss+xml"/>
<lastBuildDate>Wed, 14 Oct 2026 11:40:10 GMT</lastBuildDate>
<item><title>Small judgments, real work: how we use Jev in Hermes</title><link>https://cauldnclark.com/blogs/how-we-use-jev-for-agent-judgments</link><guid isPermaLink="true">https://cauldnclark.com/blogs/how-we-use-jev-for-agent-judgments</guid><description>A practical look at our Jev judgment layer: skill hints, content warnings, memory decisions, and engineering checks. What is active, what is on demand, and what we stopped routing.</description><pubDate>Wed, 14 Oct 2026 11:40:10 GMT</pubDate><category>Jev</category><category>TypeSafe</category><category>AI agents</category><category>Engineering</category></item>
<item><title>Before the edit: a Neo4j code graph for Hermes</title><link>https://cauldnclark.com/blogs/neo4j-code-graph-for-ai-agents</link><guid isPermaLink="true">https://cauldnclark.com/blogs/neo4j-code-graph-for-ai-agents</guid><description>How we give coding agents a queryable map of files, functions, data access, and affected pages—and why we still verify it against the source.</description><pubDate>Wed, 14 Oct 2026 09:30:00 GMT</pubDate><category>Neo4j</category><category>AI agents</category><category>Code intelligence</category></item>
</channel></rss>