<?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>AI Agents on Forrest Chai</title><link>http://forrestchai.com/tags/ai-agents/</link><description>Recent content in AI Agents on Forrest Chai</description><generator>Hugo</generator><language>en-us</language><lastBuildDate>Sat, 11 Apr 2026 00:00:00 +0000</lastBuildDate><atom:link href="http://forrestchai.com/tags/ai-agents/index.xml" rel="self" type="application/rss+xml"/><item><title>Beyond Prompt Engineering: Context, Harness, and the Product Architecture of AI Agents</title><link>http://forrestchai.com/posts/beyond-prompt-engineering/</link><pubDate>Sat, 11 Apr 2026 00:00:00 +0000</pubDate><guid>http://forrestchai.com/posts/beyond-prompt-engineering/</guid><description>Prompt engineering has been demoted from the whole problem to one layer of the stack. Context engineering decides what the model can think with. Harness engineering decides whether that thinking becomes durable work.</description></item><item><title>REST API + Skill Documents vs MCP: Two Strategies for Connecting AI Agents to Backend Capabilities</title><link>http://forrestchai.com/posts/rest-api-vs-mcp/</link><pubDate>Tue, 07 Apr 2026 00:00:00 +0000</pubDate><guid>http://forrestchai.com/posts/rest-api-vs-mcp/</guid><description>REST APIs are not inherently weak contracts. MCP is not inherently superior. The real comparison requires distinguishing three layers of interface quality — prose documentation, machine-readable specification, and protocol-native tool discovery.</description></item></channel></rss>