Keeps you from making outdated assumptions about what Claude can and can't do. Training data goes stale fast, and this gives you a current reference for both Claude AI (web/app) and Claude Code (terminal/IDE) so you can check actual capabilities before claiming something works or doesn't. Honestly just a sanity check tool. If you're writing documentation, answering questions about Claude's features, or debugging why something isn't working as expected, pull this up first. Saves you from confidently stating wrong information based on old knowledge. Simple reference skill that does one thing well.
npx skills add https://github.com/jezweb/claude-skills --skill claude-capabilities