Guardrails

Use this skill when the user wants a rule their AI agent must follow in a codebase, asks what rules, checks or hooks a repository already enforces or whether anything stops a mistake from landing, wants a rule turned into a lint, semgrep or hook check, or a repository needs its default checks set up. Use it on: 'make a rule for X', 'add a rule', 'always do X', 'never do Y', 'ban this pattern', 'the agent keeps making this mistake', 'set up guardrails', 'what rules are in my repo', 'do I have guardrails in place', 'can this be a semgrep rule', 'write a lint rule for X', 'we have no linting', 'add a commit gate', 'run the tests when claude finishes', 'which of our CLAUDE.md…