<?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>Agents on Muhammad Hassan Raza</title><link>https://mhassan.dev/tags/agents/</link><description>Recent content in Agents on Muhammad Hassan Raza</description><generator>Hugo</generator><language>en-US</language><lastBuildDate>Fri, 25 Sep 2026 10:00:00 +0500</lastBuildDate><atom:link href="https://mhassan.dev/tags/agents/index.xml" rel="self" type="application/rss+xml"/><item><title>Agent Middleware: What Kept Obelisk's Agents in Line</title><link>https://mhassan.dev/blog/langgraph-multi-agent-middleware/</link><pubDate>Wed, 08 Apr 2026 10:00:00 +0500</pubDate><guid>https://mhassan.dev/blog/langgraph-multi-agent-middleware/</guid><description>Obelisk runs more than fifteen specialist agents. The agents were the easy part. Model choice, context, cost, retries and approvals all ended up in a middleware layer, and that layer is what made it work.</description></item></channel></rss>