claude-code/separation-of-concerns

separation of concerns Shipped

What this lens looks for

Each skill, rule, and agent should have one reason to change; if describing what a skill/rule does requires "and," consider splitting; applies at every scale — a skill that does planning AND execution AND reporting is three skills

What its verifier checks

Each skill/rule/agent has a single stated purpose; skill description does not contain "and" joining unrelated concerns; workflow steps are separated into distinct agents rather than one agent doing everything