This one helps you write better agent definitions for Claude Code so it knows when to delegate work to custom agents. The key insight here is that your agent description acts as a trigger pattern, so wording matters more than you'd think. It covers how to structure descriptions, configure tool access, and design prompts that actually get Claude to hand off tasks appropriately. With 361 installs and some security audit warnings to note, it's clearly solving a real pain point. Worth grabbing if you're building custom agents and finding that Claude either ignores them completely or delegates at the wrong times.
npx skills add https://github.com/jezweb/claude-skills --skill agent-development