AI Assistant

Claude Code ships with over 100 slash commands organized across categories. Commands are invoked by typing / followed by the command name. Some commands accept arguments (shown in the argumentHint field).

Not all commands are available in every environment. Commands may be gated by auth type (claude.ai subscriber, Console API user), feature flags, platform, or internal-only restrictions. Use /help to see commands available in your current session.

CommandDescription
/helpShow help and available commands
/clearClear conversation history
/exitExit Claude Code
/configView and edit configuration
/statusShow current session status
/modelSet the AI model for Claude Code
/themeChange the terminal color theme
/colorSet agent color
/vimToggle vim keybinding mode
/versionShow current version
CommandDescription
/resumeResume a previous conversation
/sessionManage sessions
/teleportTeleport to a remote session
/rewindRewind conversation to a previous state
/shareShare the current conversation
/compactCompact conversation history to save tokens
/renameRename the current session
/contextView context window usage
/summaryGenerate a conversation summary
CommandDescription
/diffShow file diffs
/reviewRequest a code review
/copyCopy content to clipboard
/exportExport conversation or code
/branchCreate or switch git branches
/tagTag a point in the conversation
/commitCreate a git commit (internal)
/commit-push-prCommit, push, and create a PR (internal)
/pr_commentsView and respond to PR comments
/release-notesGenerate release notes
/security-reviewPerform a security review
CommandDescription
/planEnter plan mode for structured task planning
/tasksView and manage background tasks
/effortSet reasoning effort level
/fastToggle fast mode
/thinkbackReplay model thinking process
/thinkback-playPlay back thinking in real time
/passesConfigure multi-pass processing
/advisorUse advisor model for guidance
/ultraplanLaunch ultra planning mode (internal)
CommandDescription
/mcpManage MCP server connections
/skillsBrowse and manage skills
/pluginManage plugins
/hooksConfigure event hooks
/filesBrowse and manage tracked files
/agentsView and manage agent definitions
/reload-pluginsReload plugin configurations
/add-dirAdd a directory for CLAUDE.md loading
/memoryView and edit memory/CLAUDE.md files
/chromeToggle browser automation
CommandDescription
/permissionsView and manage tool permissions
/privacy-settingsConfigure privacy and data sharing
/keybindingsCustomize keyboard shortcuts
/output-styleSet output formatting style
/sandbox-toggleToggle sandbox mode
/rate-limit-optionsConfigure rate limit behavior
/statuslineCustomize status line display
CommandDescription
/voiceToggle voice input mode (feature-gated)
/bridgeManage bridge connections (feature-gated)
/ideConnect to IDE extensions
/desktopOpen in desktop app
/mobileConnect to mobile app
/btwSend a brief aside message
/feedbackSend feedback to the team
CommandDescription
/doctorRun system health diagnostics
/costShow session cost breakdown
/usageView API usage statistics
/statsShow session statistics
/envDisplay environment information (internal)
/insightsGenerate a report analyzing your sessions
/ctx_vizVisualize context window usage (internal)
/extra-usageView extended usage details
CommandDescription
/loginAuthenticate with the API
/logoutClear authentication credentials
/oauth-refreshRefresh OAuth tokens
/install-github-appInstall the GitHub App integration
/install-slack-appInstall the Slack App integration
CommandDescription
/initInitialize Claude Code in a project
/init-verifiersInitialize verification tools (internal)
/onboardingRun the onboarding flow (internal)
/terminal-setupConfigure terminal settings
/upgradeUpgrade Claude Code to the latest version
/stickersManage sticker packs

These commands are conditionally available based on build-time feature flags or runtime gates:

CommandFeature GateDescription
/voiceVOICE_MODEVoice input and transcription
/bridgeBRIDGE_MODEAlways-on bridge for remote control
/remote-controlDAEMON + BRIDGE_MODERemote control server
/briefKAIROS or KAIROS_BRIEFToggle brief output mode
/assistantKAIROSAssistant mode commands
/proactivePROACTIVE or KAIROSProactive suggestion mode
/force-snipHISTORY_SNIPForce a history snip boundary
/workflowsWORKFLOW_SCRIPTSManage workflow scripts
/webCCR_REMOTE_SETUPRemote setup configuration
/peersUDS_INBOXView peer agent connections
/forkFORK_SUBAGENTFork a sub-agent session
/buddyBUDDYCompanion/buddy features
/torchTORCHTorch debugging tool
/subscribe-prKAIROS_GITHUB_WEBHOOKSSubscribe to PR webhooks

These commands are restricted to Anthropic employees (USER_TYPE === 'ant'):

CommandDescription
/backfill-sessionsBackfill session data
/break-cacheSet a cache-breaking injection
/bughunterRun the bug hunter
/good-claudeSend positive feedback
/issueFile an internal issue
/mock-limitsSimulate rate limits
/bridge-kickForce-disconnect bridge session
/ant-traceView OpenTelemetry trace
/perf-issueReport a performance issue
/debug-tool-callDebug a specific tool call
/heapdumpCapture a heap dump
/autofix-prAuto-fix a PR
/agents-platformAgents platform management

Skill Commands

In addition to built-in commands, skills provide dynamic slash commands. Skills are loaded from:

  • .claude/skills/ directory in the project or home directory
  • Plugins that register skill commands
  • Bundled skills shipped with Claude Code
  • Dynamic skills discovered during file operations (matched by paths globs)

Skill commands use the prompt type and generate content that is sent to the model. They can specify execution context (inline or fork), register hooks, and declare file-path triggers.