@Minh Do, it's a way of thinking about using AI to build practical knowledge and tools. Here's the most recent article in the series that points to all the others and the GitHub repository. https://x.com/garrytan/status/2053127519872614419

@Salvatore Odesco, I completely agree with @Kj Rainey's answer. Here's a mental model that may help. There are skills in your home ($HOME) folder that apply to every project ($HOME/.claude/skills). If you have skills that apply to a bunch of projects, put them there. Then there are skills that only apply to a particular client. Those go inside the client folder. This helps to manage what gets loaded when starting claude.

@Verónica Gordo, the problem is that the installer put the claude app in a place zsh doesn't know about. This adds the location into your zshrc file (one time) so when zsh starts it knows where to look. echo 'export PATH="$HOME/.local/bin:$PATH"' >> ~/.zshrc source ~/.zshrc

Interesting question, @Ben Mo. Why do you ask?

Good news from Anthropic today: > Effective today, we are: 1) Doubling Claude Code’s 5-hour rate limits for Pro, Max, and Team plans; 2) Removing the peak hours limit reduction on Claude Code for Pro and Max plans; and 3) Substantially raising our API rate limits for Opus models.

Agreed, @Federico Martelli, this is good news. The deal was with xAI. They're sunsetting Colossus I and Anthropic bought access to the excess capacity.