<?xml version="1.0" encoding="UTF-8"?><rss version="2.0"><channel><title>quanh.dev</title><description>AI + systems engineering blog</description><link>https://quanh.dev/</link><item><title>Why your agent needs a skill graph, not a prompt chain</title><link>https://quanh.dev/posts/skill-graph-routing/</link><guid isPermaLink="true">https://quanh.dev/posts/skill-graph-routing/</guid><description>Declarative routing is dead. TOML DAGs give agents deterministic, auditable decision paths that prompt chains can&apos;t match.</description><pubDate>Wed, 08 Apr 2026 00:00:00 GMT</pubDate></item><item><title>MCP beyond hello world: production patterns for tool servers</title><link>https://quanh.dev/posts/mcp-beyond-hello-world/</link><guid isPermaLink="true">https://quanh.dev/posts/mcp-beyond-hello-world/</guid><description>Model Context Protocol is more than a spec — it&apos;s an architecture. Here&apos;s how to build tool servers that don&apos;t fall over at scale.</description><pubDate>Fri, 03 Apr 2026 00:00:00 GMT</pubDate></item><item><title>Agent evals that actually matter: beyond vibe checks</title><link>https://quanh.dev/posts/agent-evals-that-matter/</link><guid isPermaLink="true">https://quanh.dev/posts/agent-evals-that-matter/</guid><description>Most agent evals test the wrong thing. Here&apos;s a framework for measuring what matters: reliability, cost, and user trust.</description><pubDate>Sat, 28 Mar 2026 00:00:00 GMT</pubDate></item><item><title>Context window economics: every token has a cost you&apos;re ignoring</title><link>https://quanh.dev/posts/context-window-economics/</link><guid isPermaLink="true">https://quanh.dev/posts/context-window-economics/</guid><description>Your 200K context window isn&apos;t free real estate. Here&apos;s how to budget tokens across system prompt, tools, memory, and conversation.</description><pubDate>Fri, 20 Mar 2026 00:00:00 GMT</pubDate></item><item><title>Multi-agent handoff patterns that don&apos;t lose context</title><link>https://quanh.dev/posts/multi-agent-handoff/</link><guid isPermaLink="true">https://quanh.dev/posts/multi-agent-handoff/</guid><description>Handing off between agents is where most multi-agent systems fail. Three patterns for keeping context intact across the handoff boundary.</description><pubDate>Sat, 14 Mar 2026 00:00:00 GMT</pubDate></item><item><title>The good, the bad, and the ugly of AI agents in 2026</title><link>https://quanh.dev/posts/good-bad-ugly-agents/</link><guid isPermaLink="true">https://quanh.dev/posts/good-bad-ugly-agents/</guid><description>An honest take on where agents actually are: what works, what&apos;s hype, and what&apos;s genuinely dangerous.</description><pubDate>Sun, 08 Mar 2026 00:00:00 GMT</pubDate></item></channel></rss>