docs-claude-code-features
Claude Code Docs — Slash Commands, Skills, Plan Mode, Checkpointing, Output Styles
URL: https://code.claude.com/docs/en/slash-commands Publisher: Claude Code Docs Retrieved: 2026-07-08
Summary
Official Claude Code documentation for the interactive-feature surface of the certification: slash commands, skills, plan mode, checkpointing/rewind, and output styles. Key facts: /en/slash-commands now resolves to the Skills page (/en/skills) because custom commands have been merged into skills — a file at .claude/commands/deploy.md and a skill at .claude/skills/deploy/SKILL.md both create /deploy. Skills use a SKILL.md with YAML frontmatter (name, description, allowed-tools, disable-model-invocation, user-invocable, argument-hint, arguments, model, effort, context: fork, agent, paths, hooks, disallowed-tools, shell) plus a markdown body; support $ARGUMENTS/$ARGUMENTS[N]/$N/$name substitution, !`cmd` bash injection, and @file references; discovery is by directory (personal ~/.claude/skills/, project .claude/skills/, plugin, enterprise) with progressive disclosure. Built-in commands are catalogued at /en/commands (many bundled ones marked Skill or Workflow). Plan mode is one of the permission modes (default/acceptEdits/plan/auto/dontAsk/bypassPermissions) — reads-only, no edits until you approve a plan; enter via Shift+Tab, /plan, or --permission-mode plan. Checkpointing auto-captures code before each edit; /rewind (aliases /checkpoint, /undo) or double-Esc on empty input opens the rewind menu; bash-command file changes are NOT tracked. Output styles (Default, Explanatory, Learning, Proactive) modify the system prompt; set via /config or the outputStyle setting; custom styles live in .claude/output-styles/. Version-specific behaviors captured from the live docs on 2026-07-08 (marked ⚠ verify in the note). Content fetched from code.claude.com/docs.