Claude Code is Anthropic’s AI assistant for the terminal. I can’t write code, yet I built this site with it. Here are the commands I use, grouped by situation.
Slash commands
Type / in the prompt to open the command list.
Everyday
Command
What it does
/help
Show help and every available command
/clear
Clear the conversation and start fresh (use it between tasks)
/compact
Compress the conversation history to save context
/resume
Resume an earlier conversation
/rewind
Go back to an earlier step, including undoing code edits
/exit
Quit Claude Code
Settings and models
Command
What it does
/model
Switch model (Opus / Sonnet / Haiku)
/fast
Toggle fast mode
/config
Open settings (theme, default model and more)
/permissions
Choose which actions no longer need your approval
/statusline
Customize the status line
/login
Sign in; /logout signs out
Projects and memory
Command
What it does
/init
Generate a CLAUDE.md guide for the current project
/memory
Edit memory files so Claude remembers your preferences
/add-dir
Give Claude access to another folder
/context
See how much context is in use
Usage and diagnostics
Command
What it does
/cost
Token spend for this conversation
/usage
Plan usage and limits
/status
Account, model and version details
/doctor
Check the installation for problems
/export
Export the conversation
Advanced
Command
What it does
/mcp
Manage MCP servers (connections to outside tools)
/agents
Manage subagents
/hooks
Set up automations, such as “run X every time a file changes”
/plugin
Install and manage plugins
/code-review
Review code changes for bugs
/loop 5m task
Repeat a task on an interval
Keyboard shortcuts
Keys
What it does
Esc
Interrupt what Claude is doing
EscEsc
Press twice to go back to an earlier message
Shift + Tab
Cycle modes: normal → auto-accept edits → plan mode
↑↓
Browse what you typed before
\ + Enter
New line in the prompt
Ctrl + C
Cancel input; press twice to quit
Ctrl + V
Paste an image, such as a screenshot
Input prefixes
Prefix
What it does
Example
@
Reference a file or folder
@plan.pdf summarize this
!
Run a shell command directly
! ls
/
Run a slash command
/clear
Launching from the terminal
Before you’re inside Claude Code, run these in the terminal:
claude # start an interactive sessionclaude "question" # start with a first questionclaude -c # continue the last conversationclaude -r # pick an earlier conversation to resumeclaude -p "question" # answer once and exit, handy in scriptsclaude --model opus # start with a specific modelclaude update # update to the latest version
Habits for beginners
Run /clear before a new task so the old conversation doesn’t interfere.
For bigger tasks, press Shift + Tab into plan mode. Claude proposes a plan and waits for your approval.
If Claude heads the wrong way, press Esc, and undo unwanted edits with /rewind.
When a long conversation makes Claude forgetful, run /compact.
Run /init once at the start of a project so Claude gets up to speed quickly every time.
Commands change between versions. For the definitive list, type /help on your own machine.