All 55+ Slash Commands Explained

The Command Landscape

Claude Code has over 55 slash commands organized into functional categories. You don't need to memorize all of them — but knowing the categories and the essential commands in each makes you dramatically more efficient.

Every slash command starts with / and is typed directly in the Claude Code prompt. Tab completion works for most commands.

Essential Commands

These are the commands you'll use in almost every session:

/help

Shows all available commands with brief descriptions. Your quick reference when you forget a command name.

/help

/clear

Wipes all conversation history. Complete reset. Claude forgets everything from the current session.

/clear

Use this when you're switching to a completely unrelated task and don't need any previous context.

/compact [hint]

Compresses conversation history into a condensed summary, freeing context window space while preserving key information.

/compact

# With a hint about what to preserve
/compact "keep the auth architecture decisions and the API schema"

Use this when your context window is getting full (~80% capacity) but you still need Claude to remember important decisions from earlier in the conversation.

/exit

Exits the Claude Code session. Same as Ctrl+D.

/exit

/status

Shows current session information: model, plan, authentication method, context usage, and version.

/status

Context and Cost Commands

/context

Detailed breakdown of your context window usage — how many tokens are consumed by conversation history, system prompts, CLAUDE.md, MCP tools, and file references.

/context

This is your token budget dashboard. Check it when sessions feel expensive or when you're approaching limits.

/cost

Shows the estimated cost of the current session in dollars (for API users) or tokens consumed (for subscribers).

/cost

/insights

Shows Claude Code's internal understanding of your project — what it knows about the codebase, patterns it has noticed, and context it has accumulated.

/insights

Planning and Model Commands

/plan

Asks Claude to create a plan before taking action. Instead of jumping into implementation, Claude outlines the steps first and waits for your approval.

/plan "add real-time notifications using WebSockets"

Claude will output a detailed plan — files to modify, the approach, potential risks — before writing any code.

/execute

Tells Claude to execute the plan that was previously created with /plan.

/execute

Used after reviewing and approving a /plan output.

/model [name]

Switch the active model. Takes effect on the next message.

/model opus      # Deep reasoning
/model sonnet    # Daily driver (default)
/model haiku     # Quick tasks
/model opusplan  # Opus plans, Sonnet implements

/fast

Quick toggle to Haiku for the next message, then switches back. Useful for a one-off quick task.

/fast

/effort [level]

Set the effort level: low, medium, high, or max.

/effort high     # For complex tasks
/effort low      # For simple, mechanical tasks

Session Commands

/fork

Branch the current conversation. Creates a copy of the session at the current point, letting you explore a different direction without losing the original.

/fork

Use this when you want to try an experimental approach but keep the option to go back to the current state.

/rewind

Go back to a previous point in the conversation, undoing messages and changes after that point.

/rewind

Opens a picker showing conversation checkpoints. Select one to rewind to that state.

/resume

Interactive session picker to resume a previous session.

/resume

/rename [name]

Rename the current session.

/rename "auth-refactor-v2"

/teleport

Jump to a specific point in conversation history.

/teleport

Multi-Agent Commands

These commands manage agent teams (Max feature):

/batch

Send the same prompt to multiple agents in parallel.

/batch "run the test suite"

/simplify

Review changed code across agents for reuse opportunities and quality improvements.

/simplify

/agents

List active sub-agents and their status.

/agents

/tasks

View and manage pending tasks across agents.

/tasks

/loop [interval] [command]

Run a command on a recurring interval.

/loop 5m /status    # Check status every 5 minutes
/loop 10m npm test  # Run tests every 10 minutes

Configuration Commands

/config

Open configuration options — edit settings, manage permissions, and adjust preferences.

/config

/allowed-tools

View and manage which tools Claude is allowed to use.

/allowed-tools

/permissions

View and edit the current permission settings.

/permissions

/hooks

Manage hooks — scripts that run before or after specific Claude Code events.

/hooks

/mcp

Manage MCP (Model Context Protocol) servers — add, remove, configure.

/mcp

/vim

Toggle vim-style keybindings in the input.

/vim

/voice

Toggle voice input mode.

/voice

Skill Commands

/btw

Report feedback or bugs to the Claude Code team.

/btw "the /compact command seems to lose function signatures"

/claude-api

Get help building applications with the Claude API or Anthropic SDK.

/claude-api

/debug

Enter debug mode for troubleshooting Claude Code issues.

/debug

In-Session Shortcuts

These aren't slash commands but are part of the prompt syntax:

@file — File References

Reference files directly in your prompt:

Look at @src/auth/jwt.ts and find the bug

!command — Shell Commands

Run a shell command and include output in context:

Fix the errors in !npm test 2>&1

The 15 Commands to Memorize

If you only memorize 15 commands, make it these:

| Command | What It Does | When to Use |

|---------|-------------|-------------|

| /help | List all commands | When you forget a command |

| /clear | Wipe all history | Switching to an unrelated task |

| /compact | Compress history | Context window getting full |

| /context | Show token usage | When sessions feel expensive |

| /cost | Show session cost | Budget monitoring |

| /model | Switch model | Need different capability level |

| /plan | Create a plan first | Before complex implementations |

| /execute | Execute the plan | After reviewing a /plan |

| /fork | Branch conversation | Trying an experimental approach |

| /rewind | Undo messages | Claude went wrong, want to go back |

| /resume | Pick a session | Resuming previous work |

| /status | Session info | Checking model, plan, auth |

| /effort | Set effort level | Matching computation to task |

| /mcp | Manage MCP servers | Setting up integrations |

| /exit | Leave session | Done working |

Key Takeaway

Claude Code's 55+ slash commands are organized into categories: essentials (/help, /clear, /compact), context management (/context, /cost), planning (/plan, /execute), model control (/model, /effort), session management (/fork, /rewind, /resume), multi-agent (/batch, /agents), and configuration (/config, /mcp, /hooks). You don't need all 55 on day one — start with the 15 most important commands and expand as your workflow demands.