This lets you set up guardrails for Claude Code by writing markdown files with YAML frontmatter that pattern match against bash commands, file edits, agent stops, or prompts. You drop these files in `.claude/hookify.{rule-name}.local.md` and define regex patterns that either warn or block when they match. The use cases are obvious: stop yourself from running `rm -rf /`, catch leftover `console.log` statements before commit, protect `.env` files, or require a checklist before the agent stops. It's straightforward regex matching with simple operators like `contains` and `starts_with`, nothing fancy. Worth setting up a few rules early so you don't have to babysit the agent as much later.
npx skills add https://github.com/anthropics/claude-plugins-official --skill writing-hookify-rules