<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Web on Bryce Alan Hayes — Security Research</title><link>https://brycealanhayes.com/tags/web/</link><description>Recent content in Web on Bryce Alan Hayes — Security Research</description><generator>Hugo</generator><language>en-us</language><lastBuildDate>Sun, 20 Sep 2026 00:00:00 +0000</lastBuildDate><atom:link href="https://brycealanhayes.com/tags/web/index.xml" rel="self" type="application/rss+xml"/><item><title>AI Coding Agent Workbench: Setup and Field Practice</title><link>https://brycealanhayes.com/research/ai-agent-workbench/</link><pubDate>Sun, 20 Sep 2026 00:00:00 +0000</pubDate><guid>https://brycealanhayes.com/research/ai-agent-workbench/</guid><description>Wrapping coding agents like Claude Code in four engineering layers — rules, memory, capabilities, and a ledger — turns them into a pipeline that keeps producing gradeable reports against authorized targets: a field methodology spanning cold start through Web / mini-program / APP target types.</description></item></channel></rss>