Slash commands

Slash commands are what you type at the start of a message to drive the agent itself — sessions, modes, background work, deployment. Type / and the menu opens; type more to filter, ↑/↓ to select, Tab to complete, Enter to run, Esc to close. The menu also matches a later word of hyphenated names, so /setup finds /terminal-setup and /dir finds /add-dir.

/help (or ? on an empty prompt) lists everything your installed version has; the tables below are the complete built-in set. For the shell side of nex, see the CLI reference.

Aliases

Alias	Same as
/new, /reset	/clear	Start fresh.
/quit	/exit	Leave.
/search	/find	Search everything.
/routines	/schedule	Cloud schedules.

Session & conversation

Command	What it does
/clear (aliases /new, /reset)	Clear the screen and start a new conversation.
/compact	Summarize the conversation to save tokens.
/copy	Copy the last reply to the clipboard.
/exit (alias /quit)	Exit Nexrall Code.
/export [file]	Write this conversation to a Markdown file.
/fork [name]	Branch the conversation into a new session, keeping the original.
/rename [title]	Rename this session.
/resume [id|URL]	Resume the last or a specific saved session — by id, by name, or by pasting a pull/merge request URL the session mentioned.
/rewind [id]	List file checkpoints, or roll back to one.
/save	Save the current session now.
/sessions	List saved sessions.
/status	Version, model, mode, directory and session at a glance.
/btw <question>	Ask a side question — answered without touching the conversation.

Model, cost & context

Command	What it does
/balance	Show what is paying (wallet or team) and what is left.
/logout	Sign out on this machine: revokes the saved refresh token server-side, then deletes the local config. A token supplied through NEXRALL_TOKEN is untouched — unset it too. Sign in again with nex auth from any terminal; this session picks it up without a restart.
/context	Show context window usage.
/cost	Where the tokens and money went (per model, per agent, per turn).
/effort [level]	Set the thinking effort level.
/model [name]	Switch the model.
/stats	Usage across sessions: activity by day, tokens and models (from the local event logs).
/skill-doctor	Which loaded skills went unused in recent sessions and what each costs in context (unused first, with the catalogue-entry token estimate at ≈4 chars/token; reads the local event logs — no logs reports usage as unknown, not unused).
/recap	Where this session left off: last prompt, turns, tools, changed files (rebuilt from the local event log — also printed automatically when you resume).
/cd <dir>	Move this session to a different working directory. The new folder is trust-gated (a choice box lists what can configure the agent there); relative paths resolve against the current directory and ~ expands to home. From then on the session lives in the new folder — --continue finds it there, not in the old one. MCP servers and the project brief stay as loaded.
/suggestions [on|off]	Prompt suggestions: after each turn a short background request proposes your next prompt as grey ghost text on the empty input row. Tab (or →) places it, start typing to dismiss it. Each suggestion is one small background request billed like any turn — turn it off here, with the promptSuggestions setting, or NEXRALL_PROMPT_SUGGESTIONS=0.
/team [use <id|personal>]	List your teams, or switch what usage is billed to.

Modes & permissions

Command	What it does
/mode [ask|edit|plan|auto]	Set agent mode: ask, edit, plan or auto.
/permissions	What the agent may do here (and what each sub-agent type may touch).
/plan	Switch to plan mode (read-only).
/trust	Manage folder trust.
/yolo	Auto-approve all permissions.

Input & context files

Command	What it does
/add <filepath>	Add a file to the conversation context.
/add-dir [path]	Let the agent work in another directory this session.
/exclude <glob>	Hide paths from @ mentions and attachments (e.g. node_modules).
/image <filepath> [caption]	Attach an image or PDF and send it.
/pin <path>	Keep a file in front of the model across /compact.
/unexclude <glob|all>	Stop hiding excluded paths.
/unpin <path|all>	Stop keeping a pinned file in context.

Project work

Command	What it does
/build	Build this project.
/deploy [target] [--env X|--yes|--dry-run]	Ship it: environment, dirty files and verified checks are shown before it asks.
/diff [file]	Show uncommitted changes — the full diff, or one file.
/failures	The chain of failing checks: what broke, what was edited, what passed (Ctrl+F rescues).
/fix	Send the last failing check back to the model (same as Ctrl+F).
/git [args]	Run any git command (bare /git = git status).
/health [--all]	Project health before you start: git state, checks this project declares, TODO and credential files.
/init	Generate nexrall.md for this project.
/lint	Lint this project.
/open [<path>[:line]]	Open a file in your editor.
/review [file|--all]	Uncommitted changes grouped by risk, with the reason for each — local, never calls a model.
/run [name]	Run this project's own command (npm scripts, make targets…).
/test	Run this project's tests.
/verify [--all]	What this session actually ran — typecheck, tests, lint, build, diff.

Background & agents

Command	What it does
/agents [<name|id>|types]	Sub-agent tree of this session: who spawned whom, who runs in parallel (/agents types lists the types).
/batch <instruction>	Split a mechanical change across parallel agents, each in its own git worktree.
/bashes [show|kill <id>]	List, show or stop commands moved to the background (Ctrl+B).
/graph	Shared tasks as a dependency graph: what can start now, what is blocked.
/handoff [<agent-type>] [instructions]	Package this session for another agent (goal, files, failure, criteria) — or delegate the task.
/merge	After parallel agents: shared-write files and worktree branches git cannot merge cleanly.
/peers [send <name> <msg>|broadcast <msg>]	List, message or broadcast to other sessions.
/subtask <task>	Hand a task to a background agent that starts from THIS conversation; its report arrives when it is done.
/tasks [stop <id>|all]	Background tasks: running sub-agents and !commands, and how to stop them.
/worktree [list]	Show this session's worktree isolation status.

Automation & sharing

Command	What it does
/accept [criteria|clear]	Declare what must pass before the work counts as done.
/channel [open [auto|ask]|close|accept|discard|auto|ask]	Let CI and webhooks push messages into this live session — you accept them (/channel open). Details
/goal [condition|clear]	Keep working until a condition is met — a judge checks after every turn.
/loop [5m prompt|list|stop]	Re-run a prompt on an interval while idle (min 1m, expires in 7 days).
/schedule [add <interval> <task>|pause|resume|run|rm|runs] (alias /routines)	Cloud schedules — recurring tasks that run on Nexrall's servers (this laptop can be closed).
/share [off|drive on|off]	Watch this session from the browser — read-only until /share drive on.

Extensions

Command	What it does
/commands	List custom skills and commands.
/hooks	List the hooks that will run (project, --settings, plugins).
/import [--user] [--apply]	Bring your Claude Code setup (.claude/, CLAUDE.md) into nex — dry run first.
/mcp [<name>|enable|disable|connect|disconnect|auth]	List, inspect and toggle MCP servers.
/memory [global] [archived] [clear]	View persistent memory; # text saves a fact (see below).
/plugins	List installed plugins.
/reload	Re-read skills, plugins and the project brief (nexrall.md).
/skills	List skills.

Interface & settings

Command	What it does
/config [key[=value]]	See and change settings (liveDiff, mouse, notifications) in one place.
/doctor	Check your setup (Node, sign-in, network, MCP…) and how to fix problems.
/expand [n]	Show the full output of a folded tool call (Ctrl+O).
/find [text] (alias /search)	Search files, conversation, tools, git, errors, agents.
/fullscreen	Toggle drawing the session on the alternate screen (re-wraps on resize, page with PgUp/PgDn).
/keybindings [<action> <key>|reset]	Show the shortcut map, or rebind one action.
/livediff [on|off]	Preview the lines each edit changes, as it happens.
/mouse	Wheel scrolling in fullscreen — on/off, remembered in settings.
/network	Hosts this session reached, and what pointed at each.
/notify [auto|bell|osc|off]	Terminal notifications for background jobs, long turns and waiting prompts.
/output-style [name]	Change how the agent answers (concise, explanatory, learning, or your own).
/terminal-setup	Make Shift+Enter insert a newline in this terminal.
/theme [name|reset]	Change the colours — built-ins (dark, light, auto, high contrast…) or a named theme file.
/vim	Vim-style editing in the input box — on/off, remembered in settings.

Help & updates

Command	What it does
/help	Show commands and keyboard shortcuts (also ? on an empty prompt).
/update	Update nex to the latest version.

Beyond the built-ins

The / menu is longer than this page — commands come from four more places:

• Skills — every skill is callable as /name; a skill with disable-model-invocation: true is only callable that way (for side-effecting workflows). See Sub-agents & skills.
• Custom commands — Markdown files in .nexrall/commands/ (project or user scope) become commands; /commands lists what is loaded.
• Plugins — an installed plugin contributes its commands, e.g. the official code-review plugin adds /code-review plus the reviewer agent. See Plugins.
• MCP prompts — each prompt from a connected server appears as /<server>:<prompt>. See Prompts as slash commands.

Some everyday input is not a slash command: @ mentions a file, !command runs a shell command whose output joins the conversation, # text saves a project memory fact (#global text saves it everywhere), and a plain ? on an empty prompt prints the key map. All of that lives in Interactive mode.