Verified
New since your last visit
Unverified
Reference for the Antigravity CLI (agy) and its core extensibility surface. Official documentation at antigravity.google/docs.
TUI & Keys
Global controls
| Item | Description |
|---|---|
#Esc | Close active panels, halt streams, or clear an empty prompt |
#Ctrl+C | Terminate the CLI session
|
#Ctrl+D | Exit the session, only when the prompt box is empty |
#Ctrl+L | Refresh and clear the visual terminal buffer |
Prompt focus
| Item | Description |
|---|---|
#Enter | Submit your prompt or the active menu selection |
#Shift+Enter / Ctrl+J | Insert a newline without submitting |
#Ctrl+V | Paste media files or clipboard blocks into the prompt |
#Ctrl+O | Expand or collapse detailed tool reasoning output |
#Ctrl+R | Open the Artifact Review Panel |
#Ctrl+G | Launch your default $EDITOR to compose the prompt |
#Alt+J | Switch focus to the next subagent awaiting confirmation |
#Ctrl+K | Instantly approve the pending subagent action |
#Ctrl+A / Ctrl+E | Move the cursor to the start or end of the line |
#Ctrl+Z / Ctrl+Shift+Z | Undo or redo the last text edit in the prompt |
#F5 | Start or stop voice dictation |
#Alt+F / Alt+B | Move the cursor forward or back one word
|
#Ctrl+W / Alt+Backspace / Alt+D | Delete the word before or after the cursor
|
Navigation & scrolling
| Item | Description |
|---|---|
#↑ / ↓ | Scroll highlighted selections up or down by one item |
#PgUp / Shift+↑ | Scroll the active text viewport up by one page |
#PgDn / Shift+↓ | Scroll the active text viewport down by one page |
#← / → | Swap pages inside multipage panels like the Session Picker |
#Tab | Confirm the highlighted slash-command autofill option |
Slash Commands (Core)
Session lifecycle
| Item | Description |
|---|---|
#/help | Open the help panel of commands and shortcuts |
#/exit / /quit | Close the TUI session and restore your host shell |
#/clear / /new | Resets active conversation contexts, not just the display
|
#/rewind / /undo | Roll back conversation history to a previous message
|
#/fork / /branch | Clone the current conversation thread into a new parallel session |
#/rename <name> | Rename the current session thread |
#/resume / /switch / /conversation | Open the conversation picker overlay to load a previous thread |
Configuration & account
| Item | Description |
|---|---|
#/config / /settings | Open the interactive Settings Editor Overlay |
#/model | Choose your preferred reasoning model for the session
|
#/permissions | Open the interactive tool permissions manager panel |
#/planning | Enable multi-turn plan generation for complex tasks |
#/fast | Enable fast mode, bypassing reasoning plans for quick actions |
#/keybindings | Open the interactive Keyboard Shortcut Editor |
#/statusline | Open the Status Bar customization overlay |
#/title [on/off] | Toggle or set terminal window title updates |
#/credits | View remaining G1 credits and purchase links |
#/logout | Disconnect your profile and purge auth tokens from the keyring |
Slash Commands (Tools)
Tools & tasks
| Item | Description |
|---|---|
#/agents | Open the Agent Manager Panel for background subagents |
#/artifact | Open the Artifact Review Panel |
#/hooks | Browse active pre-flight/post-format script hooks
|
#/mcp | Open the Model Context Protocol server manager |
#/skills | Browse loaded local and global Agent Skills
|
#/tasks | Open the Task Manager Panel for background shell logs
|
Reasoning & collaboration
| Item | Description |
|---|---|
#/boost <task> | Run on-demand multi-agent deep reasoning loops |
#/teamwork-preview / /teamwork | Launch collaborative multi-agent teams for long-horizon work |
Utilities
| Item | Description |
|---|---|
#/add-dir <path> | Add a directory path to the active workspace |
#/btw <query> | Ask a side question without interrupting the main thread |
#/context | Open the context usage visualization panel |
#/copy | Copy the last agent response to the clipboard
|
#/diff | Open the Interactive Diff Viewer |
#/feedback | Open the feedback submission panel |
#/open <path> | Force the path open in your default system editor |
#/remote-control [on/off] | Toggle a session-scoped Remote Control link for the active terminal session
|
#/usage / /quota | Display model quota usage |
#/voice / /record | Dictate a prompt using your microphone |
CLI & Subcommands
Session flags
| Item | Description |
|---|---|
#agy | Launch an interactive Antigravity CLI coding session |
#-c / --continue | Continue the most recent conversation session |
#--conversation <id> | Resume a previous conversation by its unique ID |
#-p / --print / --prompt | Run prompt non-interactively and print response to stdout
|
#-i / --prompt-interactive | Execute initial prompt and continue interactive session |
#--model <name> | Specify model for current CLI session |
#--effort <level> | Set reasoning effort for session (low, medium, high) |
#--mode <mode> | Set agent execution mode (accept-edits, plan) |
#--remote-control | Start the session with a Remote Control connection for another device
|
Workspace & sandbox
| Item | Description |
|---|---|
#--project <id> | Project ID or project name for current CLI session |
#--new-project | Initialize a new project state for current session |
#--add-dir <path> | Add directory to workspace context (repeatable flag) |
#--agent <name> | Specify agent role or persona for current session |
#--sandbox | Run inside sandbox with terminal execution restrictions |
#--dangerously-skip-permissions | Auto-approve tool permission requests without prompting |
Piping & output
| Item | Description |
|---|---|
#--input-format <fmt> | Input format for print mode (text, stream-json)
|
#--output-format <fmt> | Output format for print mode (text, json, stream-json) |
#--json-schema <schema> | Enforce structured output using JSON schema string or file |
#--print-timeout <dur> | Timeout for a print-mode run; unset, the run continues until the response completes
|
#--disable-slash-commands | Disable slash command and skill expansion in print mode |
#--log-file <path> | Override CLI log file path |
Subcommands
| Item | Description |
|---|---|
#agy models | List available models and reasoning configurations |
#agy agents / agy agent | List available subagent roles and configurations |
#agy mcp <cmd> | Manage MCP servers (add, remove, list, enable, disable) |
#agy plugin <cmd> | Manage plugins (install, list, import, validate) |
#agy changelog | Show changelog and release notes for installed CLI |
#agy install | Configure environment paths and shell completions |
#agy update | Update CLI binary to latest available release |
#agy mic-serve | Serve host microphone input to CLI on another host |
#agy remote-control | Manage remote-control background daemon (start, status, stop) |
Settings (settings.json)
Appearance & editing
| Item | Description |
|---|---|
#colorScheme | TUI color theme, default "terminal"
|
#altScreenMode | Screen buffer mode: default, always, or never |
#editor | External editor target: auto, vim, emacs, or custom |
#editorMode | Prompt editing model: default (flat) or vim (modal) |
#vimInsertFirst | Start Vim mode in Insert; requires editorMode: "vim" |
#verbosity | Visual verbosity: high (full output) or low (minimal) |
#runningLightSpeed | Progress animation speed: fast, medium, slow, off |
Safety & permissions
| Item | Description |
|---|---|
#toolPermission | Global safety preset, default "request-review"
|
#artifactReviewPolicy | Code review policy: asks-for-review, agent-decides, always-proceed |
#allowNonWorkspaceAccess | Let file read/write tools navigate outside the workspace root |
#enableTerminalSandbox | Restrict local execution commands to OS containment rings |
Notifications & account
| Item | Description |
|---|---|
#notifications | Desktop and terminal bell chime on task completion |
#showTips | Show agentic tips above the prompt panel while generating |
#showFeedbackSurvey | Show periodic quality feedback surveys after tasks |
#useG1Credits | Use personal AI credits once plan quotas are exhausted |
#enableTelemetry | Allow metric collection and crash log streaming |
Environment Variables
Authentication
| Item | Description |
|---|---|
#GEMINI_API_KEY | Authenticate with a Gemini API key instead of Google sign-in
|
#GOOGLE_GEMINI_BASE_URL | Point the CLI at a different Gemini-compatible endpoint |
Rules & Skills
Project rules
| Item | Description |
|---|---|
#~/.gemini/GEMINI.md | Global rules applied across every workspace |
#.agents/rules/*.md | Workspace-level rule files, 12,000 characters each
|
#Manual / Always On / Model Decision / Glob | The four rule activation modes
|
Custom skills
| Item | Description |
|---|---|
#.agents/skills/<name>/SKILL.md | Workspace-level skill definition with YAML frontmatter
|
#~/.gemini/config/skills/ | Global machine-wide custom skills directory |
#progressive disclosure | Only a skill's name and description load until it's activated
|
Hooks & MCP
Lifecycle hooks
| Item | Description |
|---|---|
#hooks.json | Config file in .agents/ or ~/.gemini/config/ mapping hooks to events
|
#PreToolUse | Fires before a tool executes |
#PostToolUse | Fires after a tool finishes |
#PreInvocation | Fires before calling the model |
#PostInvocation | Fires immediately after the model responds |
#Stop | Fires when execution terminates |
Plugins
| Item | Description |
|---|---|
#plugins/<name>/plugin.json | Manifest bundling skills, rules, and MCP configurations |
#agy plugin list | List all installed, enabled, and imported plugins |
#agy plugin import gemini | Import plugin bundle from Gemini ecosystem |
#agy plugin import claude | Import compatible plugin bundle from Claude ecosystem |
Model Context Protocol
| Item | Description |
|---|---|
#mcp_config.json | Configuration registering external MCP tool servers |
#agy mcp add <name> <cmd> | Register an MCP server with executable command and flags |
#agy mcp list | Display configured MCP servers and connection status |
#agy mcp enable <name> | Enable a previously registered MCP server |
#agy mcp disable <name> | Disable an MCP server without removing its configuration |
No matches — clear search to see everything.