Slash commands in OpenAI Codex CLI provide quick access to navigation, context management, and special features without leaving your terminal session. Mastering these commands lets you work faster and maintain better control over your AI-assisted coding workflow.
Understanding Slash Commands
Slash commands are special instructions that start with a forward slash (/). Unlike regular prompts that Codex interprets as coding requests, slash commands trigger specific built-in behaviors. You can use them at any point during an interactive session.
To use a slash command, simply type it at the prompt and press Enter:
codex> /help
Navigation and Session Commands
These commands help you manage your Codex session and navigate the interface.
/help
Displays all available slash commands with brief descriptions. This is your reference when you forget a command name:
codex> /help
The output lists every command available in your current Codex version, including any custom commands you have configured.
/clear
Resets the conversation context entirely. Use this when you want to start fresh without exiting and restarting Codex:
codex> /clear
After clearing, Codex has no memory of previous exchanges. This is useful when switching between unrelated tasks.
/quit or /exit
Ends your Codex session and returns you to the regular terminal prompt:
codex> /quit
You can also use Ctrl+D to exit.
Context Management Commands
Managing context is essential for long coding sessions. These commands help you optimize token usage and maintain relevant history.
/compact
Summarizes your conversation history to reduce token usage while preserving important context:
codex> /compact
When Codex compacts your history, it creates a condensed summary of key decisions, code changes, and context. Use this when:
- You are approaching context limits and responses slow down
- Your session has been running for a while with many exchanges
- You want to preserve context but reduce token costs
/history
Displays your conversation history within the current session:
codex> /history
This shows the exchanges between you and Codex, which is helpful for reviewing what has been discussed or understanding how context has accumulated.
/undo
Reverts the most recent change Codex made to your files:
codex> /undo
This is a safety net when Codex makes an edit you did not intend. It works with the built-in version tracking that Codex maintains during sessions.
Mode and Model Commands
These commands let you adjust how Codex operates during a session.
/mode
Changes the approval mode, which controls how much autonomy Codex has when executing commands and making changes:
codex> /mode
Codex presents a menu of available modes:
- suggest - Codex only suggests changes, you must approve each one
- auto-edit - Codex can edit files automatically but asks before running commands
- full-auto - Codex operates with full autonomy (use with caution)
/model
Switches to a different AI model mid-session without restarting:
codex> /model
A selection menu appears with available models like gpt-5.2-codex and gpt-5.1-codex-mini. This is useful when you want to use a cheaper model for simple tasks or a more powerful one for complex reasoning.
/status
Shows your current configuration including active model, approval mode, and token usage:
codex> /status
Special Feature Commands
Codex includes several commands that activate dedicated workflows.
/review
Launches a dedicated code review agent that analyzes your staged changes:
codex> /review
The review agent examines your git staging area and provides feedback on code quality, potential bugs, and improvement suggestions. This is particularly useful before committing changes.
/resume
Continues a previous session from where you left off:
codex> /resume
Codex maintains transcript history that allows you to pick up exactly where you stopped, including full context from the previous session.
/diff
Shows the differences between the current file state and before your session began:
codex> /diff
This helps you track what has changed during your Codex session without leaving the tool.
Keyboard Shortcuts
Alongside slash commands, Codex supports several keyboard shortcuts for common actions:
| Shortcut | Action |
|---|---|
| Ctrl+C | Cancel current operation |
| Ctrl+D | Exit Codex session |
| Ctrl+L | Clear screen (keeps context) |
| Up/Down | Navigate command history |
| Tab | Autocomplete file paths |
These shortcuts work throughout your session and complement slash commands for efficient navigation.
Creating Custom Commands
You can define custom slash commands in your ~/.codex/config.toml file to automate repetitive workflows:
# ~/.codex/config.toml
[[custom_commands]]
name = "test"
prompt = "Run the test suite and summarize any failures"
[[custom_commands]]
name = "docs"
prompt = "Update the documentation for the function I'm currently editing"
[[custom_commands]]
name = "security"
prompt = "Review this code for security vulnerabilities, focusing on input validation and authentication"
After adding custom commands, they become available as slash commands:
codex> /test
codex> /docs
codex> /security
Custom commands can include any prompt text, making them powerful shortcuts for tasks you perform frequently.
Tips for Efficient Command Usage
- Start sessions with /status to verify your model and mode settings are correct for the task
- Use /compact proactively rather than waiting for context limit errors
- Create custom commands for your most common workflows to save typing
- Use /mode suggest when working on critical code where you want full control
- Run /review before commits to catch issues before they enter your codebase
- Combine /clear with mode changes when switching between very different tasks
Next Steps
- Learn about switching models for detailed model selection guidance
- Explore code review workflows with the /review command
- Understand session resumption to pick up where you left off
- Configure Codex settings for custom commands