Type: url
Original: https://claude.com/blog/steering-claude-code-skills-hooks-rules-subagents-and-more
Status: generated
Summary
An Anthropic blog post (June 18, 2026) comparing the seven ways of instructing Claude Code — CLAUDE.md files, rules, skills, subagents, hooks, output styles and appending the system prompt — by when each loads into context, how it behaves under compaction, what it costs in context and how much authority it carries. It gives a decision framework for where an instruction belongs, arguing in particular that behaviour which must always or never happen belongs in hooks, permissions or managed settings rather than in prompted instructions.