claude-code-primitives

name	claude-code-primitives
description	Reference for Claude Code's own mechanics — slash commands, skills, subagents, hooks — and how they route together. Use when creating or editing a slash command, skill, subagent, or hook, or when deciding which primitive fits a task.

Claude Code Primitives

• Routing: Commands initiate, subagents verify, hooks gate.
• Slash Commands (.claude/commands/): Short, repeatable actions. Canonical text of prompts lives here (e.g., /qspec, /tdd, /qcheck).
• Skills (.claude/skills/): Complex multi-step workflows. Write descriptions as triggers ("when should I fire?"). Provide goals, constraints, and a "Gotchas" section.
• Subagents (.claude/agents/): Spawn for fanning out or parallel reads. Skip for <3 tool calls. Read-only agents use haiku; write agents use sonnet/opus with isolation: worktree.
• Agent teams vs subagents: Use subagents for scoped delegation inside one workstream. Use teams when work should split across longer-lived sessions that coordinate.
• Hooks: Enforce standards mechanically (e.g., PostToolUse, Stop). Hook types: shell commands, prompt hooks, MCP tool hooks.
• Hooks receive event data as JSON on stdin. Read fields with jq, for example .tool_input.file_path; do not rely on $CLAUDE_FILE_PATH style variables unless explicitly configured.