Claude Code Command Guide

Claude Code Command Guide | Generated by AI

Home 2026.02

I’ll search for the most current information about Claude Code’s command options from Anthropic’s documentation.Let me fetch more detailed information about the command-line options.Based on the documentation, here’s a comprehensive introduction to Claude Code command options:

Claude Code Command Options

Basic Commands

Starting Claude Code:

claude - Launch interactive session in current directory
claude "query" - Run single query and return to shell
claude -p "query" - Print mode (non-interactive, scriptable output)

Session Management:

claude --continue - Resume most recent conversation
claude -r "session-id" "query" - Resume specific session by ID
claude --resume - Interactive menu to select previous session
claude update - Update to latest version
claude mcp - Configure Model Context Protocol servers

Key CLI Flags

Working Directories:

--add-dir <path> [<path2>...] - Add additional directories for Claude to access
```
claude --add-dir ../apps ../lib
```

Permissions:

--allowedTools - Tools allowed without prompting

claude --allowedTools "Bash(git log:*)" "Read"

--disallowedTools - Tools to deny without prompting
--permission-mode <mode> - Set permission mode: normal, auto-accept, or plan
```
claude --permission-mode plan
```

System Prompts (4 options):

--system-prompt - Complete control, removes default instructions
--system-prompt-file <path> - Load custom prompt from file
--append-system-prompt - Add instructions while keeping defaults (recommended)
--append-system-prompt-file <path> - Append from file while keeping defaults

Output Formats:

-p, --print - Non-interactive print mode
--output-format json - JSON output for scripting/automation
--verbose - Detailed logging

Custom Agents:

--agents <json> - Define custom subagents

claude --agents '{
  "code-reviewer": {
    "description": "Expert code reviewer",
    "prompt": "Focus on code quality and security",
    "tools": ["Read", "Grep", "Glob", "Bash"],
    "model": "sonnet"
  }
}'

Interactive Commands (Slash Commands)

Available when Claude Code is running:

Session & Configuration:

/help - Show available commands
/config - Open settings interface
/login - Switch accounts
/vim - Enable vim keybindings

Workflow:

/resume - Resume previous conversation
/compact - Compact conversation history
/init - Initialize new session
/terminal-setup - Configure Shift+Enter for terminal

Permissions:

/allowed-tools - Configure tool permissions

Plugins & Extensions:

/plugin - Manage plugins interactively
/hooks - View/manage hooks
/context - View current context usage

Custom Commands: Any skill you create becomes a /command - type / to see all available

Plugin Commands (CLI)

# Install plugin
claude plugin install <name>@<marketplace> [--scope user|project|local]

# Remove plugin
claude plugin uninstall <name>

# Enable/disable
claude plugin enable <name>
claude plugin disable <name>

# Update
claude plugin update <name>

Permission Modes

Toggle with Shift+Tab during session:

Normal Mode - Claude asks permission for each action
Auto-Accept Mode (⏵⏵) - Automatically accept file edits
Plan Mode (⏸) - Claude plans with read-only operations

Environment Variables

CLAUDE_CODE_DISABLE_BACKGROUND_TASKS=1 - Disable background tasks
CLAUDE_CODE_TASK_LIST_ID=<name> - Use named task list
CLAUDE_CODE_ENABLE_TASKS=false - Revert to old TODO list
MAX_THINKING_TOKENS - Enable extended thinking permanently
CLAUDE_BASH_MAINTAIN_PROJECT_WORKING_DIR=1 - Reset to project dir after each command
SLASH_COMMAND_TOOL_CHAR_BUDGET - Adjust skill character budget

Interactive Features

Input:

\ + Enter - Insert newline
Shift+Enter - Insert newline (most terminals)
Ctrl+R - Reverse search command history
Tab - Toggle extended thinking mode
!<command> - Run bash command directly

Navigation:

Arrow keys - Command history
Ctrl+O - Toggle verbose mode

Common Workflow Examples

# Explore codebase in plan mode
claude --permission-mode plan "Analyze authentication system"

# Continue work from previous session
claude --continue "Finish implementing OAuth"

# Script automation
claude -p --output-format json "Run tests and report failures" | jq

# Add context from multiple directories
claude --add-dir ../shared ../config "Refactor API handlers"

# Custom instructions for session
claude --append-system-prompt "Always write type-annotated Python"

For more details on any specific feature, you can ask Claude Code directly - it has access to its own documentation and can answer questions about its capabilities.

Back Donate