OmniscioGuidesClaude Code commands cheat sheet
Claude Code commands cheat sheet
Last checked against Anthropic's documentation on November 16, 2026. By the Omniscio team. Updated .
The short version
- Which command, when — a first session, a long task, before you ship.
- Every built-in command, in twelve tables.
- The CLI flags worth typing, and the keys worth memorising.
- Your own commands — and running these from Omniscio instead of a terminal.
Which command, when
Most commands belong to one moment in a session, and Anthropic's own account of a typical session is the best order to meet them in. Start here, then jump to the full list when you need a name.
A first session in a repo
/initwrites a starterCLAUDE.md— the file Claude reads at the start of every session — then/memoryrefines it./mcpsets up any servers the project needs, and/permissionssets your approval rules.
During a task
/planswitches into plan mode before a large change, so you approve the approach first./modeland/effortchange which model answers and how much reasoning it applies.- When the conversation gets long,
/contextshows what is filling the window and/compactsummarizes it to free space. /usageshows spend and plan limits;/btwasks a side question that never joins the conversation.
Running work in parallel
/taskslists the session's background work, including subagents that have finished./backgrounddetaches the whole session so it keeps running while your terminal is free.
Before you ship
/diffshows what changed./code-reviewchecks the current diff for correctness bugs and can apply the fixes with--fix; pass a PR number to review a pull request instead./security-reviewlooks at the same changes for security problems.
Between sessions
/clearstarts a fresh conversation while keeping project memory./resumereturns to an earlier conversation,/branchbranches the current one, and/forkcopies it into a new background session.
When something is wrong
/doctorruns a setup checkup on the installation and configuration, and/debugdiagnoses a runtime problem./rewindrolls code and conversation back to a checkpoint./feedbackreports a bug, with a consent screen before anything is sent.- To stop a runaway answer first, press
Escto stop the current response or tool call, orCtrl+Cto interrupt a running operation.
Every built-in command
Every entry in Anthropic's command reference, in the groups you will reach for, with the three they have retired in their own table at the end. <arg> is required and [arg] is optional. Two kinds are marked where they appear: a skill is a prompt handed to Claude, and a workflow fans work out across many subagents in the background.
A command sits in the group for what you are doing with it, not for how it is built.
Sessions, context and memory
| Command | What it does |
|---|---|
/clear [name] | Start a new conversation with empty context. An optional name labels the previous one in the /resume picker. |
/compact [instructions] | Free up context by summarizing the conversation so far, with optional focus instructions. |
/autocompact [auto|<tokens>] | Set how full the context window gets before Claude Code compacts automatically, such as 500k, or auto for the window tuned to your model. |
/context [all] | Show current context usage — the tokens filling the window — as a colored grid, with suggestions for what is heavy. |
/memory | Edit CLAUDE.md files, turn auto memory on or off, and view auto-memory entries. |
/rewind | Rewind this conversation, the code, or both to an earlier point in it, or summarize from a chosen message. Aliases: /checkpoint, /undo. |
/resume [session] | Resume an earlier conversation — an earlier session — by ID or name, or open the picker. Background sessions appear marked bg. |
/rename [name] | Rename the current session. With no name, one is generated from the conversation. |
/branch [name] | Branch the conversation at this point and switch into the branch, so you can try a different direction while the original is kept. |
/fork [prompt] | Copy the conversation into a new background session, which keeps running while you work here. A prompt starts the copy working immediately. |
/btw [question] | Ask a side question about the session without adding it to the conversation. |
/recap | Generate a one-line summary of the session on demand. |
/plan [description] | Enter plan mode directly from the prompt, optionally starting on a task you describe. |
/goal [condition|clear] | Set a goal: Claude keeps working across turns until the condition is met. With no argument, shows the current goal. |
/export [filename] | Export the conversation as plain text, to a file or through a dialog for the clipboard. |
/copy [N] | Copy the last response to the clipboard. /copy 2 copies the second-to-last. |
/exit | Exit the CLI. In an attached background session it detaches and the session keeps running. Alias: /quit. |
/restart | Restart Claude Code and resume this session, which applies an installed update. Needs Pro, Max or Ultra and a native install on macOS or Linux, not WSL. Alias: /update. |
Model, effort, voice and display
| Command | What it does |
|---|---|
/model [model] | Switch the model and save it as your default for new sessions. Arrow keys adjust the effort level where the model supports it. |
/effort [level|auto|status] | Set the effort level from low to xhigh, or max, or auto; status prints the current one, and ultracode turns ultracode on for the session. |
/fast [on|off] | Toggle fast mode: a faster setting for the models that support it, at a higher cost per token. |
/advisor [model|off] | Turn the advisor tool on or off: a second model consulted for guidance at key moments. Accepts fable, opus, sonnet or a model ID. |
/output-style [style] | List output styles, or switch to one such as concise. |
/color [color|default] | Set the prompt bar color for this session. |
/theme | Change the color theme. The auto option follows your terminal's light or dark background. |
/tui [default|fullscreen] | Set the terminal renderer and relaunch into it with the conversation intact. fullscreen is the full-screen view that redraws without flicker. |
/focus | Toggle focus view: only your last prompt, a one-line tool-call summary and the final response. Fullscreen rendering only. |
/scroll-speed | Adjust mouse-wheel scroll speed interactively. Fullscreen rendering only, and not in the JetBrains IDE terminal. |
/statusline | Configure the status line. With no argument it auto-configures from your shell prompt. |
/voice [hold|tap|off] | Toggle voice dictation, or enable it in a specific mode. Needs a Claude.ai account. |
Project setup and working directories
| Command | What it does |
|---|---|
/init | Initialize the project with a starter CLAUDE.md. |
/add-dir <path> | Add a working directory for file access. Type a partial path to see directory suggestions, and press Tab to accept one. |
/cd <path> | Move the session to a new working directory, keeping the conversation. |
/import [codex|gemini|cursor] | Bring configuration from OpenAI Codex, Google Gemini CLI or Cursor into Claude Code: instruction files, MCP servers, commands, subagents and skills. |
/setup-bedrock | Configure Amazon Bedrock authentication, region and model pins. Hidden until CLAUDE_CODE_USE_BEDROCK=1 is set. |
/setup-vertex | Configure Google Cloud's Agent Platform authentication, project, region and model pins. Hidden until CLAUDE_CODE_USE_VERTEX=1 is set. |
/web-setup | Connect your GitHub account for cloud sessions, using your local gh credentials. |
/remote-env | Choose the default cloud environment for cloud sessions you start from the CLI. |
Permissions, settings and the terminal
| Command | What it does |
|---|---|
/permissions | Manage allow, ask and deny rules for tool permissions, view them by scope, and manage working directories. |
/hooks | View hook configurations. A hook is a command you make Claude Code run automatically at a set moment, such as before a tool call. |
/sandbox | Toggle sandbox mode, which sets an operating-system boundary around the shell commands Claude runs. macOS, Linux and WSL2. |
/config [key=value …] | Open the Settings interface, or set a value directly, for example /config theme=dark. |
/status | Open Settings on the Status tab: version, model, account and connectivity. |
/privacy-settings | View and update your privacy settings. Pro, Max and Ultra subscribers. |
/update-config [request] | Skill. Describe a settings change — allow a command, set an environment variable, add a hook — and Claude edits the matching settings.json. |
/auto-mode-setup | Draft autoMode.environment entries from your project and recent sessions, then review and save them. |
/fewer-permission-prompts | Skill. Scan your transcripts for common read-only Bash and MCP calls, then add a prioritized allowlist to the project settings. |
/terminal-setup | Install a Shift+Enter keybinding for newlines in VS Code, Cursor, Devin Desktop, Alacritty or Zed. In Apple Terminal it enables Option+Enter and turns off the bell. |
/keybindings | Open your keyboard shortcuts file. |
Review, verify and ship
| Command | What it does |
|---|---|
/diff | Review the changes in your working tree, including the edits Claude has made so far. |
/code-review [level] [--fix] [--comment] [pr#|branch|path] | Skill. Review the current diff, or a PR, branch or path you pass, for correctness bugs. --fix applies the findings. |
/review | Alias of /code-review. |
/security-review | Analyze the changes on the current branch for security vulnerabilities, reviewing the diff against origin's default branch. Needs an origin remote. |
/simplify [target] | Skill. Review the changed code for cleanup and apply the fixes: four review agents run in parallel over reuse, simplification, efficiency and whether the change is at the right level of abstraction. It does not look for correctness bugs; /code-review does. |
/verify | Skill. Confirm a change does what it should by building and running the app and observing the result, rather than relying on tests. |
/run | Skill. Launch and drive your project's app to see a change working. |
/run-skill-generator | Skill. Teach /run and /verify how to build, launch and drive your app from a clean environment. |
/autofix-pr [prompt] | Spawn a cloud session that watches the current branch's PR and pushes fixes when CI fails or reviewers comment. Needs the gh CLI and access to cloud sessions. |
/ultrareview [PR or branch] | Run a deep, multi-agent code review in a cloud sandbox. Three free runs on Pro, Max and Ultra, then usage credits. |
Background work, agents and the cloud
| Command | What it does |
|---|---|
/background [prompt] | Detach the current session to run as a background agent and free this terminal. Monitor it with claude agents. |
/tasks | View and manage background work in this session, including subagents that have finished. Alias: /bashes. |
/subtask <task> | Spawn a forked subagent — a helper that inherits the whole conversation — to work on the task while you keep working. |
/stop | Stop the background session you are attached to. The transcript and any worktree — a separate copy of the repository — are kept. |
/agents | Print a reminder to ask Claude to create or manage subagents, or to edit .claude/agents/ directly. |
/list-agents | List the subagents, agent team teammates and other sessions Claude can message, with the name to use for each. Alias: /peers. |
/schedule [description] | Create, update, list or run cloud scheduled tasks, formerly routines. |
/workflows | Open the workflow progress view to watch, pause, resume or save running and completed workflows. |
/teleport | Pull a cloud session into this terminal. Needs a claude.ai subscription. Alias: /tp. |
/desktop | Continue the current session in the Claude Code Desktop app. macOS or x64 Windows, with a Claude subscription. Alias: /app. |
/remote-control | Make this session available for Remote Control from claude.ai. |
Plugins, skills and MCP
| Command | What it does |
|---|---|
/plugin [subcommand] | Manage plugins. With no argument the menu opens; pass list, install, enable or disable to act directly. |
/reload-plugins [--force] | Reload all active plugins to apply pending changes without restarting, and report load errors. |
/skills | List available skills. Type to filter, press t to sort by token count, and Space or Enter to cycle a skill's visibility. |
/reload-skills | Re-scan skill and command directories so skills added or changed on disk become available without restarting. |
/skill-doctor | Show what each of your skills costs in context and how often it is used, so you can find skills to turn off. |
/mcp [reconnect|enable|disable …] | Manage MCP server connections and their OAuth authentication. MCP is the protocol Claude Code uses to plug in outside tools. With no argument the interactive list opens. |
/artifacts | List the artifacts you own or that are shared with you, then attach one to the session, open it, or copy its link. |
Accounts, plans and billing
| Command | What it does |
|---|---|
/login | Sign in to your Anthropic account. |
/logout | Sign out of your Anthropic account. |
/usage | Show session cost, plan usage limits and activity stats. |
/stats | Alias for /usage. Opens on the Stats tab. |
/cost | Alias for /usage. |
/usage-credits | Configure usage credits, or request them from your admin, when you hit a limit. |
/rate-limit-options | Show ways to keep working when a claude.ai usage limit blocks a request. Needs a claude.ai subscription. |
/upgrade | Open the upgrade page to switch to a higher plan tier. |
Help, feedback, editors and integrations
| Command | What it does |
|---|---|
/help | Show help and the available commands. |
/doctor [prompt-audit] | Skill. Run a setup checkup that diagnoses installation, settings, extension and CLAUDE.md problems, and proposes fixes. |
/debug [description] | Skill. Enable debug logging for the session and troubleshoot by reading the session debug log. |
/heapdump | Write a JavaScript heap snapshot and a memory breakdown to diagnose high memory use. Hidden from the menu; type it in full. |
/bug [report] | Report a bug or share your conversation, after choosing how much history to include on a consent screen. |
/feedback [report] | Send product feedback. Opens the same dialog as /bug, with the same consent step. |
/release-notes | View the changelog in an interactive version picker. |
/insights | Generate an HTML report analyzing your recent sessions on this machine. Not available in cloud sessions. |
/team-onboarding | Generate a team onboarding guide from your usage history over the past 30 days. |
/ide | Manage IDE integrations and show their status. |
/chrome | Configure Claude in Chrome settings. |
/claude-in-chrome [task] | Skill. Have Claude carry out a task in your browser, such as testing a page or reading console logs. |
/install-github-app | Install the Claude GitHub App for a repository, with an optional step for GitHub Actions workflows and secrets. |
/install-slack-app | Install the Claude Slack app through an OAuth flow in the browser. |
Fun and freebies
| Command | What it does |
|---|---|
/passes | Share a free week of Claude Code with friends. Only visible if your account is eligible. |
/mobile | Show a QR code to download the Claude mobile app. Aliases: /ios, /android. |
/powerup | Discover features through quick interactive lessons with animated demos. |
/radio | Open Claude FM lo-fi radio in your browser, or print the stream URL when no browser is available. |
/stickers | Check Claude Code sticker availability. |
Bundled skills and workflows
These also ship with Claude Code, but work like skills: each loads a set of instructions Claude then follows. A workflow is the larger kind, splitting the job across many subagents.
Several of them depend on artifacts, which are not available on Amazon Bedrock, Google Cloud's Agent Platform, Microsoft Foundry or Claude Platform on AWS, so /design, /slides and /design-sync do nothing there; /design-sync also needs claude.ai, which the CLI does not contact on those platforms or through a Claude apps gateway.
| Command | What it does |
|---|---|
/artifact-capabilities | Skill. The reference for runtime capabilities a published artifact can use, such as calling your connectors. |
/artifact-diagramming | Skill. Diagramming guidance: when a diagram helps, and how to write inline SVG that stays legible in light and dark themes. |
/batch <instruction> | Skill. Orchestrate large-scale changes in parallel: research the codebase, split the work into 5 to 30 units, show a plan, then spawn one background subagent per unit. |
/claude-api [subcommand] | Skill. Load Claude API and Managed Agents reference material for your project's language. |
/dataviz [request] | Skill. Design guidance for charts, graphs and dashboards, including a palette checked for colorblind safety and contrast. |
/deep-research <question> | Workflow. Fan out web searches on a question, fetch and cross-check sources, and synthesize a cited report. |
/design [brief] | Skill. Draft UI mockups, screen flows, landing pages or posters as artboards on one canvas. |
/design-login | Authorize design-system access for /design-sync with your claude.ai account. This signs in to the standalone Claude Design site, which closes by December 14, 2026. |
/design consent, /design revoke | Grant Claude access to Claude Design, or withdraw it. When a reply from Claude asks you to run /design consent, type it at the prompt yourself. |
/design-sync [hint] | Skill. Convert your repo's React design system and upload it to Claude Design, so the designs it produces use your real components. It uploads to the standalone claude.ai/design site, which Anthropic says closes by December 14, 2026. |
/loop [interval] [prompt] | Skill. Run a prompt repeatedly while the session stays open. Omit the interval and Claude paces itself between iterations. |
/plugin-authoring | Skill. The reference Claude works from to write a mod — a change to Claude's own interface, such as a theme, panel or status line. |
/slides [brief] | Skill. Make a new presentation as a Claude Slides artifact filled from your brief. |
/workflow-authoring | Skill. The reference for writing dynamic workflow scripts: the script API, resume behavior and quality patterns. |
Three commands that are gone
Anthropic's list still carries these to say what replaced them, so an old blog post you find will name them. They are the three that do not run.
| Command | What to use instead |
|---|---|
/pr-comments | Removed in v2.1.91. Ask Claude directly to view the pull request comments. |
/vim | Removed in v2.1.92. Toggle Vim and Normal editing modes with /config, then Editor mode. |
/ultraplan | Removed. Use plan mode, which /plan opens. |
The CLI flags worth typing
Flags go on the claude command that starts a session, so they set something before the session does anything. Anthropic's help text does not list every flag, so a flag missing from claude --help may still work. These are the ones that change how a session starts.
Start, resume, or hand off
| Flag | What it does |
|---|---|
--continue, -c | Load the most recent conversation in this directory, including a background session that has finished. |
--resume, -r | Resume a session by ID or name, or open the picker. You can also pass the path to a session's .jsonl transcript. |
--fork-session | With --resume or --continue, create a new session ID instead of reusing the original. |
--session-id | Use a specific session ID for the conversation. It must be a valid UUID. |
--name, -n | Name the session, shown in /resume and the terminal title. You can resume it later with claude --resume <name>. |
--worktree, -w | Start in an isolated git worktree at <repo>/.claude/worktrees/<name>, generating a name if you give none. |
--tmux | Create a tmux session for the worktree. Requires --worktree, and uses iTerm2 native panes where available. |
--bg, --background | Start as a background agent and return immediately, printing the session ID and management commands. |
--teleport | Resume a cloud session in your local terminal. |
Model, effort and context
| Flag | What it does |
|---|---|
--model | Set the model for this session with an alias such as sonnet, opus, haiku or fable, or a full model name. |
--effort | Set the effort level for this session: low, medium, high, xhigh, max or ultracode. Available levels depend on the model. |
--fallback-model | Fall back to the named model, or a comma-separated list tried in order, when the primary model is overloaded or unavailable. |
--autocompact | Set the auto-compact window for this session without changing your saved settings. |
--advisor <model> | Enable the advisor tool for this session with a model alias or a full model ID. |
Permissions and tools
| Flag | What it does |
|---|---|
--permission-mode | Begin in a named permission mode: default, acceptEdits, plan, auto, dontAsk or bypassPermissions. |
--dangerously-skip-permissions | Skip permission prompts. Equivalent to --permission-mode bypassPermissions. |
--allow-dangerously-skip-permissions | Add bypassPermissions to the Shift+Tab cycle without starting in it. |
--allowedTools, --allowed-tools | Tools that run without prompting, written as permission rules such as "Bash(git log *)". |
--disallowedTools, --disallowed-tools | Deny rules. A bare tool name removes it from Claude's context; "mcp__*" removes every MCP tool. |
--tools | Restrict which built-in tools Claude can use: "" for none, "default" for the default set, or names such as "Bash,Edit,Read". |
What loads, and what is left out
| Flag | What it does |
|---|---|
--add-dir | Add extra working directories Claude may read and edit. Most .claude/ configuration there is not discovered. |
--settings | Point at a settings JSON file or an inline JSON string. Its values override the same keys in your settings files for this session. |
--setting-sources | Choose which settings sources load: user, project, local, comma-separated. |
--mcp-config | Load MCP servers from JSON files or strings, space-separated. |
--strict-mcp-config | Skip the MCP servers configured elsewhere and load only the ones passed with --mcp-config. |
--plugin-dir | Load a plugin from a directory or .zip for this session only. Repeat the flag for more. |
--bare | Minimal mode: skip auto-discovery of hooks, skills, custom commands, subagents, plugins, MCP servers, auto memory and CLAUDE.md so scripted calls start faster. |
--safe-mode | Start with all customizations disabled to troubleshoot a broken configuration. Authentication, model selection, tools and permissions work normally. |
--disable-slash-commands | Disable all skills and commands for this session. |
Scripted and unattended runs
| Flag | What it does |
|---|---|
--print, -p | Print the response and exit, without an interactive session. |
--output-format | Set the print-mode output: text, json or stream-json. |
--input-format | Set the print-mode input: text or stream-json. |
--include-partial-messages | Include partial streaming events. Requires --print and --output-format stream-json. |
--json-schema | Get validated JSON output matching a JSON Schema when the run finishes. Print mode only. |
--max-turns | Limit the number of agentic turns. Print mode only, no limit by default. |
--max-budget-usd | Stop the run once estimated API spend reaches this amount. Print mode only, and spend can pass the cap, so leave headroom. |
--verbose | Full turn-by-turn output. |
--no-session-persistence | Do not save the session to disk, so it cannot be resumed. Print mode only. |
Agents, editors and diagnostics
| Flag | What it does |
|---|---|
--agent | Run this session as a named agent, overriding the agent setting. |
--agents | Define custom subagents for this session as JSON. With --print it can be a path to a JSON file instead. |
--ide | Connect to the IDE at startup when exactly one valid IDE is available. |
--chrome, --no-chrome | Turn Chrome integration on or off for this session. |
--debug | Turn debug mode on, optionally filtered by category such as --debug='mcp,startup'. The filter binds only in the = form. |
--debug-file <path> | Write debug logs to a specific file. Turns debug mode on. |
--version, -v | Print the version number. |
The full list, including the cloud, plugin, prompt-editing and managed-machine flags, is in Anthropic's CLI reference.
The keys worth memorising
Shortcuts vary by platform and terminal, and most of them can be rebound — a few reserved keys, Ctrl+C and Ctrl+D among seven, cannot. These are the ones worth memorising; the keyboard shortcuts guide has the full list by context and the file that rebinds them.
Getting a prompt in
| Key | What it does |
|---|---|
/ | Open the command menu, or filter it as you type letters after the slash. |
! | Shell mode: run a command directly, add its output to the session, and have Claude respond to it. |
@ | Mention a file path, with autocomplete. |
? | On an empty prompt, toggle the shortcut help panel. |
\ then Enter | Newline without sending. Works in every terminal. |
Ctrl+J | Newline without sending, with no setup. |
Ctrl+V | Paste an image from the clipboard. On Windows the key is Alt+V. |
Stopping and queueing
| Key | What it does |
|---|---|
Esc | Stop the current response or tool call so you can redirect, keeping the work done so far. |
Esc Esc | With text in the prompt, clear it and keep the draft in history. On an empty prompt, open the rewind menu. |
Ctrl+C | Interrupt a running operation. With nothing running, the first press clears the prompt and a second exits. |
Ctrl+D | Exit: the first press shows a hint, a second within 800ms exits. |
Ctrl+Enter / Ctrl+X Ctrl+S | Send your queued messages right away. |
Changing and viewing the session
| Key | What it does |
|---|---|
Shift+Tab | Cycle permission modes. |
Option+P / Alt+P | Switch model without clearing your prompt. |
Option+T / Alt+T | Toggle extended thinking. It has no effect on Opus 5.5, Sonnet 5.5, Haiku 5.5 or the Fable models, which always use it. |
Ctrl+B | Background a Bash command or an agent. In tmux, press it twice. |
Ctrl+S | Stash the prompt. On an empty prompt, restore the stashed text. |
Ctrl+R | Reverse-search your command history. |
Ctrl+O | Toggle the transcript viewer, which shows detailed tool usage per message. |
Ctrl+T | Toggle Claude's to-do checklist. This is not the background-task view, which is /tasks. |
Ctrl+G | Open your prompt in your default text editor; Ctrl+X Ctrl+E does the same. |
On macOS, most Option-key shortcuts need Option set as Meta in your terminal, though Option+T works without it. /terminal-setup sets up the terminals it supports: in Apple Terminal it enables Option+Enter and turns off the bell.
Commands you write yourself
The list above is what ships. You can add your own command as a Markdown file in .claude/commands/, or as a skill folder, and it then appears in the same / menu beside the built-in ones.
A command that never shows up is usually a path or a settings-block problem rather than a broken Claude Code: the custom slash commands guide takes the causes in the order to check them.
The two pages are made to be used together: this one is the name you forgot, that one is the fix when a name you wrote is missing.
Where this page stops
- Every keystroke, by context. This page carries the keys worth memorising. The full table, the
keybindings.jsonfile that rebinds them and why a rebind can do nothing: the keyboard shortcuts guide. - Your own commands. Writing them, naming them, and the reasons one does not appear: the custom slash commands guide.
- A command that is missing from your menu. Anthropic ties availability to platform, plan and environment; the command not found guide covers the install and
PATHcases. - What each permission mode allows.
Shift+Taband--permission-modeare one row each here: the permission modes guide.
Running these from Omniscio
Omniscio is a desktop app that runs Claude Code sessions side by side. You do not have to give up these commands to get the window. Seven of them, named below, run for real in the underlying Claude Code session when you type them at the start of a composer message, instead of being sent to the model as text.
- What runs.
/model,/clear,/cost,/contextand/statusare dispatched, and so are/effortand/mcp. You get the live model switch, the real token figures and the real context state, not a guess. - Claude Code sessions only. These are Claude Code's own commands, so in a session on another engine the text is delivered to the model normally.
- The cost figure for other models. On a non-Claude model, Claude Code's dollar amounts would be a guess at Claude's prices, so Omniscio prices the session from that model's own published rate and leaves Claude Code's dollar figures out. Token counts and durations are still real.
- Nothing to switch on. It is already on for Claude Code sessions. The native slash commands page in the library has the detail.
It is free to use, with up to 8 projects on the Free plan; Pro is $20/month or $192/year and lifts that limit. Download it for Windows, macOS or Linux, or see the features page.
Sources
Each behaviour here rests on one of these, read on the date beside it.
- Commands — the full command table read row by row, the note on how commands are recognized, the workflow walkthrough and the note that availability varies. Read November 16, 2026.
- CLI reference — the flags, what each one changes, and the note that
--helpdoes not list them all. Read November 16, 2026. - Interactive mode — the key tables, the quick keys, the macOS Option note. Read November 16, 2026.
- Customize keyboard shortcuts — which keys are reserved and cannot be rebound. Read November 16, 2026.
- Sandboxing, for what the sandbox boundary covers and which platforms it runs on; Fast mode, for what fast mode changes. Read November 16, 2026.
- Skills, for what a bundled skill is and how to add your own commands. Read November 16, 2026.
- Omniscio's native slash commands page, read November 16, 2026, and
src/shared/paid-offering.tsin this repo for the Free plan's project limit. - One capture here on November 16, 2026:
claude --versionprinting2.1.325 (Claude Code).
This layer moves with every release: Anthropic adds commands, retires a few and changes a flag's default, so the version at the top describes the tables.
How we tested this
Every command, flag and key on this page was read from Anthropic's own documentation on November 16, 2026 — the commands page, the CLI reference and interactive mode — against the release npm reported that day, 2.1.325, which claude --version also printed.
Nothing was run to write the page: no session was reconfigured and no flag was exercised, so these are Anthropic's account of each row rather than a run of our own. The Omniscio rows come from the library page named above and from this repository's own source.
Common questions
How do I see every command Claude Code has?
Type / on its own and the command menu opens; type / plus letters to filter it. /help shows help and the available commands. Not every command appears for every user: Anthropic says availability depends on your platform, plan and environment, so a command this page lists may be absent from your menu.
Why does a slash command in the middle of my message do nothing?
A command is only recognized at the start of your message. Text that follows the command name becomes its arguments. So /init at the start of a message runs the command, while a sentence that happens to contain /init is ordinary text.
What happens if I send a command while Claude is still answering?
Claude Code queues it and runs it after the current turn finishes. A few commands run immediately without interrupting the response, such as /status, /tasks and /usage.
How do I run a shell command from inside Claude Code?
Start the message with an exclamation mark. Shell mode runs the command directly, adds its output to the session, and Claude responds to that output. It is a keystroke, not a slash command, and the same trick works in any terminal.
What is the difference between a command, a skill and a workflow?
Most of Claude Code's commands are built in: their behavior is coded into the CLI. Two kinds of entry in Anthropic's list are different. A bundled skill is a prompt handed to Claude, the same way a skill you write yourself works. A bundled workflow fans the work out across many subagents and runs in the background.
Keep the commands, lose the terminal juggling
Omniscio runs Claude Code sessions side by side in one window. The commands it dispatches run for real when you type them in a Claude Code session's composer, and the other agents it runs sit beside them in the same window. Any model, any harness — Claude Code, Codex, OpenCode, Devin, Pi, and Hermes all supported.
For the full key list, see the keyboard shortcuts guide; for your own commands, the custom slash commands guide.