<?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>Claude Code on AI Insights Lab</title><link>https://cskwork.github.io/tags/claude-code/</link><description>Recent content in Claude Code on AI Insights Lab</description><generator>Hugo</generator><language>en-us</language><lastBuildDate>Sat, 12 Sep 2026 08:37:08 +0900</lastBuildDate><atom:link href="https://cskwork.github.io/tags/claude-code/index.xml" rel="self" type="application/rss+xml"/><item><title>A Claude Code setup that actually runs every day: settings.json, CLAUDE.md, hooks, skills, MCP</title><link>https://cskwork.github.io/posts/claude-code-setup-settings-hooks-skills-mcp/</link><pubDate>Sat, 12 Sep 2026 08:00:00 +0900</pubDate><guid>https://cskwork.github.io/posts/claude-code-setup-settings-hooks-skills-mcp/</guid><description>The Claude Code configuration I run daily, file by file. Permission allow/deny lists that survive auto mode, the real hook JSON shape, what belongs in CLAUDE.md versus a hook, skills and subagents, MCP scopes, and headless mode for CI.</description></item></channel></rss>