The TUI has pager-local slash commands, plus a smaller set provided by xai-grok-shell. User-invocable skills also appear as slash commands.

In the TUI, Shift+Tab cycles session modes. For the full key reference, see Keyboard Shortcuts.

Plan mode is planning first: only the session plan file can be edited until you approve. That file-edit gate is independent of the permission mode (ask, auto, or always-approve). Enter with /plan [description] or Shift+Tab, and reopen a plan with /view-plan. See Plan Mode.

Auto

Auto uses a classifier to auto-approve safe tools; dangerous ones may still prompt. Toggle with /auto or Shift+Tab when the feature is enabled. Full mode table: Permissions.

Always-approve

Always-approve skips permission prompts for tool calls (deny rules and hooks still apply). Toggle with /always-approve or Shift+Tab, or start with grok --always-approve. Modes, allow/deny rules, and how they relate to the sandbox are under Permissions.

Core TUI commands

The command palette groups session, context, model, and tool actions.

Use /context to check current context usage.

Command	What it does
`/quit` (alias `/exit`)	Quit the application
`/help`	Browse commands and keyboard shortcuts
`/home`	Return to the welcome screen
`/new` (alias `/clear`)	Start a new session
`/resume`	Resume a previous session
`/sessions`	Switch, rename, or close active sessions
`/fork`	Branch the current session into a peer agent
`/rename <title>` (alias `/title`)	Rename the current session
`/share`	Share the current session via URL
`/session-info`	Show session info
`/context`	View context usage
`/compact [context]`	Compact conversation history
`/rewind`	Rewind to a previous turn
`/export`	Export the conversation to a file or clipboard
`/copy [N]`	Copy the last (or Nth-latest) response to the clipboard
`/find`	Search the conversation scrollback
`/transcript`	View the full transcript in your pager (`$PAGER`)
`/model <name>` (alias `/m`)	Switch the active model
`/effort`	Set reasoning effort for the current model
`/always-approve`	Toggle always-approve mode
`/auto`	Toggle auto mode (classifier; when feature enabled)
`/plan [description]`	Enter plan mode
`/view-plan`	View the current plan
`/btw <question>`	Ask a side question without interrupting
`/loop [interval] <prompt>`	Run a prompt on a recurring interval — see Background Tasks
`/imagine <prompt>`	Generate an image from a text description
`/imagine-video <prompt>`	Generate a video from a text description
`/tasks`	List background tasks, subagents, and scheduled tasks
`/create-workflow [description]`	Author and save a new workflow
`/workflow <name> [args]`	Launch a saved workflow, or `pause` / `resume` / `stop` / `save` a run
`/workflows`	Open the live workflow run dashboard (fullscreen)
`/deep-research <query>`	Run a background research workflow
`/queue`	List the prompts queued behind the running turn
`/dashboard`	Open the Agent Dashboard
`/settings` (alias `/config`)	Open the settings modal
`/theme [name]` (alias `/t`)	Switch the color theme
`/compact-mode`	Toggle denser UI layout
`/multiline` (alias `/ml`)	Toggle multiline input
`/vim-mode`	Toggle vim-style scrollback keybindings
`/timestamps <mode>`	Set which messages show the clock time: `off`, `minimal`, or `on`
`/density <level>`	Set how much of each tool call the transcript shows: `Detailed`, `Diff focus`, `Balanced`, `Grouped`, or `Compact`
`/terminal-setup`	Check terminal and clipboard setup
`/config-agents` (alias `/agents`)	Manage agent definitions
`/personas`	Manage personas
`/remember <note>`	Save a memory note
`/import-claude`	Open the Claude settings import modal
`/feedback [text]`	Send feedback about the current session
`/release-notes` (alias `/changelog`)	View release notes for the current version
`/usage`	View credit usage or manage billing
`/privacy`	Show or toggle privacy and data-retention status
`/login`, `/logout`	Sign in, or sign out of the current account
`/hooks`	Open the unified extensions modal at the Hooks tab
`/plugins`	Open the unified extensions modal at the Plugins tab
`/marketplace`	Open the unified extensions modal at the Marketplace tab
`/skills`	Open the unified extensions modal at the Skills tab
`/mcps`	Open the unified extensions modal at the MCP tab

/hooks, /plugins, /marketplace, /skills, and /mcps all open the same extensions modal — they just pre-select a tab. A few commands appear only when their feature is available (for example /imagine and /loop).

Shell-provided commands

Command	What it does
`/goal <objective>`	Work toward an objective until it is verified complete — see Goals
`/flush`	Flush conversation memory to disk now
`/memory` (alias `/mem`)	Browse, view, and manage your memories
`/dream`	Run memory consolidation

Skills as commands

Any user-invocable skill can also appear as a slash command, for example /<skill-name>.

If names collide, use the qualified form, such as /local:commit.

Goals

/goal gives Grok an objective and lets it work toward it across many rounds, without you prompting each step. Goals work best for long-running, hard tasks with measurable success criteria.

How it works

After every round, Grok checks its progress and decides whether to keep working, stop because it needs something from you, or treat the goal as done.
When the goal looks done, a panel of reviewer subagents checks the work against the evidence. The goal is marked complete only when most of them agree.
If the reviewers find gaps, Grok keeps working on them. If it gets stuck on the same gaps, it pauses and shows you what's missing.

Writing a good goal

Say what "done" looks like, so the reviewers have something concrete to check.

/goal [objective]. Keep going until [verifiable end state].

Weak: no way to tell when it's done.

/goal Migrate the auth module to the new API

Strong: clear finish line and limits.

/goal Migrate the auth module to the new API, verified by the auth test suite passing, without changing public endpoints.

Commands

Command	What it does
`/goal <objective>`	Start a goal. Add `--budget <tokens>` to cap the tokens it may spend.
`/goal` or `/goal status`	Show the goal, its status, tokens used, and time elapsed.
`/goal pause`	Pause the goal.
`/goal resume`	Continue a paused goal.
`/goal clear`	Remove the goal.

Workflows

Workflows orchestrate a bounded set of subagents in the background: fan out work, verify, then return a result. They are on by default.

Create one with /create-workflow and describe what it should do. Grok asks about fan-out, verification, and scope, then authors, smoke-checks, and saves the file as .grok/workflows/<name>.rhai in the project, or ~/.grok/workflows/<name>.rhai for every project.

/create-workflow Review the current branch for bugs, then verify each finding

Launch a saved workflow with /workflow <name> (optional JSON args) or /<name>. /workflows is the live run dashboard in the fullscreen TUI; it lists active and retained runs, not saved files.

/workflow review-changes {"target":"origin/main...HEAD"}
/workflow pause review-changes
/workflow resume review-changes
/workflow stop review-changes-2

/deep-research <query> starts a built-in research workflow. Disable the feature with [workflows] enabled = false in ~/.grok/config.toml, or GROK_WORKFLOWS=0.

Last updated:October 8, 2026