<?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 Feng&#39;s Notes</title>
    <link>https://ofeng.org/tags/memory/</link>
    <description>Recent content in Memory on Feng&#39;s Notes</description>
    <generator>Hugo</generator>
    <language>en</language>
    <lastBuildDate>Sun, 12 Apr 2026 09:00:00 +0100</lastBuildDate>
    <atom:link href="https://ofeng.org/tags/memory/index.xml" rel="self" type="application/rss+xml" />
    <item>
      <title>How SOTA Agent Systems Manage Sessions and Memory</title>
      <link>https://ofeng.org/posts/agent-memory-comparison/</link>
      <pubDate>Sun, 12 Apr 2026 09:00:00 +0100</pubDate>
      <guid>https://ofeng.org/posts/agent-memory-comparison/</guid>
      <description>&lt;p&gt;“Agent memory” sounds like one feature, but in practice it is at least four different problems: session state, durable memory, project context, and recall strategy. The current generation of agent systems does not solve these in the same way. OpenClaw treats memory as file-backed knowledge plus retrieval tools; Hermes Agent separates bounded persistent memory from a searchable session archive; Codex CLI leans on local transcripts, layered project instructions, and skills; Claude Code combines persistent &lt;code&gt;CLAUDE.md&lt;/code&gt; rules with auto memory and resumable sessions.&lt;/p&gt;</description>
    </item>
  </channel>
</rss>
