<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Memory on Keith Hodo</title><link>https://keithhodo.dev/tags/memory/</link><description>Recent content in Memory on Keith Hodo</description><generator>Hugo -- gohugo.io</generator><language>en</language><copyright>© 2026 Keith Hodo</copyright><lastBuildDate>Wed, 09 Sep 2026 09:50:59 -0700</lastBuildDate><atom:link href="https://keithhodo.dev/tags/memory/index.xml" rel="self" type="application/rss+xml"/><item><title>Teaching a Coding Agent to Dream: Mining Transcripts to Find Forgotten Signal</title><link>https://keithhodo.dev/posts/teaching-a-coding-agent-to-dream/</link><pubDate>Sun, 06 Sep 2026 00:00:00 +0000</pubDate><guid>https://keithhodo.dev/posts/teaching-a-coding-agent-to-dream/</guid><description>An unattended nightly job mines my coding agents&amp;rsquo; own session transcripts for corrections that never made it into shared memory, then puts each candidate in front of me for a yes or no. Here&amp;rsquo;s what it found, what broke while I built it, and how I am testing the same idea with Kiro and Codex.</description><media:content xmlns:media="http://search.yahoo.com/mrss/" url="https://keithhodo.dev/posts/teaching-a-coding-agent-to-dream/feature.jpg"/></item></channel></rss>