Complete Command & Configuration Reference

This reference covers every slash command, built-in tool, settings key, CLI flag, keyboard shortcut, and authentication method in Gemini CLI. Use it as a lookup table when you need the exact syntax or behavior for a specific feature.

Slash commands are your primary interface for controlling Gemini CLI. They manage sessions, configure settings, inspect state, and trigger workflows.

SESSION MANAGEMENT
/clear           Clear conversation history and start fresh
/compact         Compress context to free tokens
/restore         Restore to a previous checkpoint
/chat            View session history, resume, or share sessions
/chat resume     Resume a previous conversation
/chat share      Export session as markdown or JSON

CONFIGURATION
/settings        View and modify all configuration
/init            Auto-generate GEMINI.md from codebase
/memory add      Add runtime context note
/hooks list      List all configured lifecycle hooks
/hooks enable    Enable a specific hook
/hooks disable   Disable a specific hook

INSPECTION
/help            Show all commands and usage
/tools           List all available tools (built-in + MCP)
/stats           Show token usage statistics
/stats model     Show current model and context allocation

OUTPUT
/copy            Copy last response to clipboard

Built-in tools are the core capabilities Gemini CLI uses to interact with your codebase and environment. These are always available without any configuration.

FILE OPERATIONS
read_file              Read file contents (supports offset/limit for large files)
write_file             Write content to a file (creates or overwrites)
edit                   Apply targeted edits to existing files
list_directory         List contents of a directory
glob                   Find files matching a glob pattern

SEARCH
search_file_content    Search file contents with regex (ripgrep-style)
codebase_investigator  Deep codebase analysis — traces dependencies, flows, architecture

EXECUTION
run_shell_command      Execute a shell command and return output

WEB
google_web_search      Search Google for current information
web_fetch              Retrieve content from a specific URL

MEMORY
save_memory            Persist notes to GEMINI.md
write_todos            Create and manage task lists

The settings.json file is the primary configuration mechanism. It lives in .gemini/settings.json for project scope or ~/.gemini/settings.json for global scope. Here are the key configuration fields.

{
  "model": "gemini-2.5-pro",
  "thinkingBudget": "auto",
  "chatCompression": "auto",
  "sandbox": "none",
  "trustedFolders": ["/home/user/projects/my-app"],
  "mcpServers": {},
  "includeTools": [],
  "excludeTools": [],
  "hooks": {}
}

CLI flags control behavior when launching Gemini CLI from the terminal. These override settings.json values for the current session.

CLI FLAGS
gemini                    Launch interactive mode
gemini -p "prompt"        Headless mode — single prompt, no interaction
--output-format json      Machine-readable JSON output (headless)
--output-format stream-json   Streaming JSON output (headless)
--sandbox docker          Run tool executions in Docker container
--sandbox podman          Run tool executions in Podman container
--yolo                    Auto-approve all tool executions (DANGER)
--model <name>            Override the default model
--verbose                 Show detailed execution logs

Keyboard shortcuts for the interactive terminal UI. These work during an active Gemini CLI session.

KEYBOARD SHORTCUTS
Ctrl+L       Clear screen (preserves conversation)
Ctrl+G       Open $EDITOR for multi-line input
Ctrl+T       Toggle tool descriptions in output
Ctrl+C       Cancel current generation
Ctrl+D       Exit Gemini CLI
Alt+M        Toggle markdown rendering
Alt+V        Paste image from clipboard
Up/Down      Navigate command history
Tab          Autocomplete file paths and commands

Authentication methods determine how Gemini CLI connects to Google's Gemini models. Two paths are supported.

AUTHENTICATION

1. Google Account (OAuth)
   - Run `gemini` and follow the browser OAuth flow
   - Free tier with generous rate limits
   - Best for personal use and exploration

2. API Key
   - Get a key from aistudio.google.com
   - Set GEMINI_API_KEY environment variable
   - Best for automation, CI/CD, and higher throughput
   - Supports pay-as-you-go for heavy usage