Every Claude Code setting on one page

Pick which model answers when Claude asks the advisor tool.

Example: {"advisorModel": "opus"}

Source: docsAny settings file

Start every session as a named subagent with its prompt, tools, and model.

Example: {"agent": "code-reviewer"}

Source: docsSince v2.0.59Any settings file

Let Claude send a push notification to your phone when it decides to.

Example: {"agentPushNotifEnabled": true}

Source: docsAny settings file

Load the claude.ai connectors Claude Code fetches itself alongside a deployed managed-mcp.json.

Example: {"allowAllClaudeAiMcps": true}

Source: docsManaged settings only

Let the built-in Claude in Chrome server run alongside a deployed managed-mcp.json.

Example: {"allowClaudeInChromeWithManagedMcp": true}

Source: docsSince v2.1.282Managed settings only

Replace the default allowlist of channel plugins that can push messages.

Source: docsSince v2.1.84Managed settings only

Limit which URLs HTTP hooks can target.

Example: {"allowedHttpHookUrls": ["https://hooks.example.com/*", "http://localhost:*"]}

Source: docsAny settings file

Allowlist which MCP servers users can add.

Source: docsAny settings file

Limit which API providers a machine may use.

Example: {"allowedProviders": ["anthropic", "bedrock"]}

Source: docsSince v2.1.285Managed settings only

Run only the hooks your organization deploys.

Example: {"allowManagedHooksOnly": true}

Source: docsManaged settings only

Make the managed MCP allowlist the only one that applies.

Source: docsManaged settings only

Make managed settings the only settings source of permission rules.

Example: {"allowManagedPermissionRulesOnly": true}

Source: docsManaged settings only

Turn extended thinking off for every session.

Example: {"alwaysThinkingEnabled": false}

Source: docsAny settings file

Generate the API credential with your own command.

Example: {"apiKeyHelper": "/bin/generate_temp_api_key.sh"}

Source: docsAny settings file

Run your organization's mods after every mod a user installs.

Source: docsUser or managed settings

Let an unanswered question auto-continue after idle time.

Example: {"askUserQuestionTimeout": "5m"}

Source: docsUser or managed settings

Customize the attribution Claude Code adds to commits and pull requests.

Source: docsSince v2.0.62Any settings file

Change or hide the trailer Claude Code adds to commits.

Source: docsAny settings file

Change or hide the attribution line in pull request descriptions.

Example: {"attribution": {"pr": ""}}

Source: docsAny settings file

Omit the claude.ai session link from cloud and Remote Control commits.

Example: {"attribution": {"sessionUrl": false}}

Source: docsSince v2.1.183Any settings file

Turn automatic compaction off or on.

Example: {"autoCompactEnabled": false}

Source: docsAny settings file

Set how full the context gets before Claude Code compacts.

Example: {"autoCompactWindow": 500000}

Source: docsAny settings file

Connect to a running VS Code or JetBrains IDE automatically from an external terminal.

Example: {"autoConnectIde": true}

Source: docsGlobal config, ~/.claude.json

Wait in the open session and continue the task automatically after a claude.ai usage limit resets.

Example: {"autoContinueAtUsageLimit": false}

Source: docsSince v2.1.234User or managed settings

Let /ide install the IDE extension from an external terminal.

Example: {"autoInstallIdeExtension": false}

Source: docsGlobal config, ~/.claude.json

Store auto memory in a directory you choose.

Example: {"autoMemoryDirectory": "~/my-memory-dir"}

Source: docsSince v2.1.74Any settings file

Turn auto memory off or on.

Example: {"autoMemoryEnabled": false}

Source: docsAny settings file

Add your own allow and deny rules to the auto mode classifier.

Example: {"autoMode": {"soft_deny": ["$defaults", "Never run terraform apply"]}}

Source: docsUser or managed settings

Send every shell command through the auto mode classifier, even ones a narrow allow rule matches.

Example: {"autoMode": {"classifyAllShell": true}}

Source: docsSince v2.1.193User or managed settings

Follow new output to the bottom in fullscreen rendering.

Example: {"autoScrollEnabled": false}

Source: docsSince v2.1.110Any settings file

Follow the stable release channel instead of latest.

Example: {"autoUpdatesChannel": "stable"}

Source: docsAny settings file

Restrict which models people can pick.

Example: {"availableModels": ["sonnet", "haiku"]}

Source: docsAny settings file

Make each availableModels model ID entry permit only the version it names.

Source: docsSince v2.1.283Managed settings only

Turn off the session recap shown when you come back to the terminal.

Example: {"awaySummaryEnabled": false}

Source: docsAny settings file

Refresh expired Bedrock credentials in .aws with your own command.

Example: {"awsAuthRefresh": "aws sso login --profile myprofile"}

Source: docsAny settings file

Supply Bedrock credentials as JSON from your own command.

Example: {"awsCredentialExport": "/bin/generate_aws_grant.sh"}

Source: docsAny settings file

Render screen-reader friendly output.

Example: {"axScreenReader": true}

Source: docsUser or managed settings

Record the files that changed while a Bash command ran in every permission mode.

Example: {"bashEditDiffEnabled": true}

Source: docsSince v2.1.269User or managed settings

Set how much of a successful command's output Claude receives inline.

Example: {"bashOutputMaxChars": 100000}

Source: docsSince v2.1.261Any settings file

Block plugin marketplace sources for your organization.

Example: {"blockedMarketplaces": [{"source": "github", "repo": "untrusted/plugins"}]}

Source: docsManaged settings only

Keep Claude's tools off external pages in the desktop Browser pane.

Example: {"browserExternalPageTools": "disabled"}

Source: docsManaged settings only

Allow channels for your organization.

Example: {"channelsEnabled": true}

Source: docsManaged settings only

Turn on Chrome integration when a session starts, in the interactive CLI and the VS Code extension.

Example: {"claudeInChromeDefaultEnabled": true}

Source: docsGlobal config, ~/.claude.json

Inject organization-wide CLAUDE.md instructions from managed settings.

Source: docsManaged settings only

Skip specific CLAUDE.md files when memory loads.

Example: {"claudeMdExcludes": ["**/vendor/**/CLAUDE.md"]}

Source: docsAny settings file

Choose how many days Claude Code keeps transcripts before deleting them.

Example: {"cleanupPeriodDays": 20}

Source: docsAny settings file

Show your organization's announcements at startup.

Source: docsSince v2.0.32Any settings file

Make /copy copy the full response without showing the picker.

Example: {"copyFullResponse": true}

Source: docsGlobal config, ~/.claude.json

Turn off automatic copying of text you select with the mouse in fullscreen rendering and agent view.

Example: {"copyOnSelect": false}

Source: docsGlobal config, ~/.claude.json

Choose whether Claude Code delivers, flags, or refuses messages from your other sessions.

Example: {"crossSessionInbound": "hold"}

Source: docsAny settings file

Choose whether Bash or PowerShell runs the shell commands you type with the ! prefix.

Example: {"defaultShell": "powershell"}

Source: docsAny settings file

Open agent view instead of a new conversation when you run claude with no arguments.

Example: {"defaultToAgentsView": true}

Source: docsGlobal config, ~/.claude.json

Block specific MCP servers by URL, command, or name.

Example: {"deniedMcpServers": [{"serverName": "filesystem"}]}

Source: docsAny settings file

Block specific models, even ones availableModels permits.

Example: {"availableModels": ["opus", "sonnet"], "deniedModels": ["claude-opus-5-5"]}

Source: docsSince v2.1.283Managed settings only

Set an age limit in days for Claude Desktop and Cowork transcripts.

Example: {"desktopSessionCleanupPeriodDays": 90}

Source: docsSince v2.1.248User or managed settings

Set how long Claude Code waits for a forwarded dialog to be answered before it cancels it.

Example: {"dialogExpiry": "10m"}

Source: docsSince v2.1.224User or managed settings

Choose whether Claude's proposed file changes open in the VS Code or JetBrains diff viewer or stay in the terminal.

Example: {"diffTool": "terminal"}

Source: docsGlobal config, ~/.claude.json

Turn off background agents and agent view.

Example: {"disableAgentView": true}

Source: docsAny settings file

Turn off hooks, a custom status line, and a custom @ file suggestion command at once.

Example: {"disableAllHooks": true}

Source: docsAny settings file

Deprecated. enableArtifact replaces it.

Example: {"disableArtifact": true}

Source: docsAny settings file

Remove auto mode from the permission mode cycle.

Example: {"disableAutoMode": "disable"}

Source: docsAny settings file

Limit the desktop Browser pane to localhost for people and Claude.

Example: {"disableBrowserExternalNavigation": true}

Source: docsManaged settings only

Turn off the skills and workflows included with Claude Code.

Example: {"disableBundledSkills": true}

Source: docsSince v2.1.169Any settings file

Turn off claude.ai connectors so Claude Code doesn't fetch them.

Example: {"disableClaudeAiConnectors": true}

Source: docsAny settings file

Block plugins that install by running a marketplace-declared command.

Example: {"disableCommandPluginSources": true}

Source: docsSince v2.1.229Managed settings only

Stop Claude Code from registering the claude-cli:// handler.

Example: {"disableDeepLinkRegistration": "disable"}

Source: docsSince v2.1.83Any settings file

Turn off Desktop Code sessions that run on the device, leaving SSH to other hosts and cloud.

Example: {"disableDesktopLocalSessions": true}

Source: docsManaged settings only

Reject specific servers from a project's .mcp.json.

Example: {"disabledMcpjsonServers": ["filesystem"]}

Source: docsAny settings file

Block Claude's tools in the desktop iOS Simulator pane.

Example: {"disableMobileSimulatorTools": true}

Source: docsManaged settings only

Turn off Remote Control everywhere it can start.

Example: {"disableRemoteControl": true}

Source: docsAny settings file

Reject the CLI flags that sideload plugins, subagents, and MCP servers.

Example: {"disableSideloadFlags": true}

Source: docsManaged settings only

Stop skills and custom commands from running inline shell.

Example: {"disableSkillShellExecution": true}

Source: docsSince v2.1.91Any settings file

Turn dynamic workflows off for everyone.

Example: {"disableWorkflows": true}

Source: docsAny settings file

Use vim key bindings in the input prompt.

Example: {"editorMode": "vim"}

Source: docsAny settings file

Set a default effort level for models without a saved level of their own.

Example: {"effortLevel": "xhigh"}

Source: docsAny settings file

Turn off :shortcode: emoji suggestions and replacement in the prompt input.

Example: {"emojiCompletionEnabled": false}

Source: docsAny settings file

Approve every server in project .mcp.json files without a prompt.

Example: {"enableAllProjectMcpServers": true}

Source: docsAny settings file

Turn the Artifact tool off with a false in any file.

Example: {"enableArtifact": false}

Source: docsAny settings file

Approve specific servers from a project's .mcp.json.

Example: {"enabledMcpjsonServers": ["memory", "github"]}

Source: docsAny settings file

Turn individual plugins on or off per scope.

Source: docsAny settings file

Turn dynamic workflows on or off against your plan's default.

Example: {"enableWorkflows": true}

Source: docsAny settings file

Keep the /model Default choice inside your availableModels allowlist.

Example: {"availableModels": ["sonnet", "haiku"], "enforceAvailableModels": true}

Source: docsSince v2.1.175Any settings file

Show Claude's last response as comments when you press Ctrl+G to edit.

Example: {"externalEditorContext": true}

Source: docsGlobal config, ~/.claude.json

Register marketplaces for a repository or an organization.

Source: docsAny settings file

Name backup models for when the primary is overloaded.

Example: {"fallbackModel": ["claude-sonnet-5", "claude-haiku-4-5"]}

Source: docsSince v2.1.166Any settings file

Turn fast mode on for sessions where it's available.

Example: {"fastMode": true}

Source: docsAny settings file

Require people to turn fast mode on each session.

Example: {"fastModePerSessionOptIn": true}

Source: docsAny settings file

Control whether Claude queues feedback drafts for you to review.

Example: {"feedbackDrafts": "quiet"}

Source: docsUser or managed settings

Change how often the session quality survey appears.

Example: {"feedbackSurveyRate": 0.05}

Source: docsAny settings file

Turn off or on the file snapshots that /rewind restores.

Example: {"fileCheckpointingEnabled": false}

Source: docsAny settings file

Supply @ file autocomplete from your own command.

Source: docsSince v2.0.65Any settings file

Make issue or review IDs in output into clickable links below the input box.

Source: docsSince v2.1.176User or managed settings

Set the gateway URL the login screen connects to.

Example: {"forceLoginGatewayUrl": "https://claude-gateway.example.com"}

Source: docsUser or managed settings

Restrict login to claude.ai, Claude Console, or a cloud gateway.

Example: {"forceLoginMethod": "claudeai"}

Source: docsAny settings file

Pin claude.ai logins to your organization.

Source: docsAny settings file

Block startup until server-managed settings are freshly fetched.

Example: {"forceRemoteSettingsRefresh": true}

Source: docsSince v2.1.92Managed settings only

Let /login reach a cloud gateway on public IPv4 space your organization uses internally.

Example: {"gatewayInternalNetworks": ["203.0.113.0/24"]}

Source: docsSince v2.1.268Managed settings only

Refresh Google Cloud credentials with your own command.

Example: {"gcpAuthRefresh": "gcloud auth application-default login"}

Source: docsAny settings file

Limit which env vars HTTP hooks can put in headers.

Example: {"httpHookAllowedEnvVars": ["MY_TOKEN", "HOOK_SECRET"]}

Source: docsAny settings file

Stop Claude Code from compacting a long conversation that you've stepped away from.

Example: {"idleCompaction": false}

Source: docsAny settings file

Deprecated. attribution replaces it.

Example: {"includeCoAuthoredBy": false}

Source: docsAny settings file

Remove the built-in commit and PR instructions from Claude's context.

Example: {"includeGitInstructions": false}

Source: docsSince v2.1.69Any settings file

Get a push notification when Claude is waiting on you.

Example: {"inputNeededNotifEnabled": true}

Source: docsAny settings file

Ask you before Claude messages one of your sessions on another machine.

Example: {"isolatePeerMachines": true}

Source: docsAny settings file

Deprecated, and has no effect.

Source: docsSince v2.1.238Any settings file

Have Claude respond in a language other than English.

Example: {"language": "japanese"}

Source: docsSince v2.1.0Any settings file

Turn off the ← shortcut that backgrounds the session and opens agent view.

Example: {"leftArrowOpensAgents": false}

Source: docsGlobal config, ~/.claude.json

Provide remote MCP servers to every user alongside the ones they add.

Source: docsSince v2.1.259Managed settings only

Compose every managed source you deploy instead of using the highest-priority one alone.

Example: {"managedSourcesBehavior": "merge"}

Source: docsSince v2.1.242Managed settings only

Cap the effort level for every model or per model, on every provider.

Source: docsSince v2.1.267Any settings file

Cap how wide the prose in Claude's responses runs in a wide terminal.

Example: {"maxProseWidth": 80}

Source: docsSince v2.1.282Any settings file

Keep auto-updates from installing anything below a version.

Example: {"autoUpdatesChannel": "stable", "minimumVersion": "2.1.100"}

Source: docsAny settings file

Map model IDs to your provider's IDs, such as Bedrock ARNs.

Source: docsSince v2.1.73Any settings file

Choose which models the /model picker lists, in your own order and with your own labels.

Source: docsUser or managed settings

Report spend at your organization's contracted rates instead of list price.

Source: docsManaged settings only

Keep a saved effort level or auto-compact window per model, or cap one model's effort.

Example: {"modelSettings": {"claude-opus-5-5": {"effortLevel": "high"}}}

Source: docsAny settings file

Generate rotating OpenTelemetry headers with your own command.

Example: {"otelHeadersHelper": "/bin/generate_otel_headers.sh"}

Source: docsAny settings file

Apply or drop restrictions an SDK or IDE host passes when you deploy managed settings.

Example: {"parentSettingsBehavior": "merge"}

Source: docsManaged settings only

Set allow, ask, and deny rules and the default permission mode.

Source: docsAny settings file

Give Claude file access to directories outside the current one.

Example: {"permissions": {"additionalDirectories": ["../docs/"]}}

Source: docsAny settings file

Always prompt before listed tool uses.

Example: {"permissions": {"ask": ["Bash(git push *)"]}}

Source: docsAny settings file

Make the file tools refuse reads outside the working directories in every permission mode.

Example: {"permissions": {"blockReadsOutsideWorkingDirectories": true}}

Source: docsSince v2.1.257Any settings file

Make Write, Edit, and NotebookEdit refuse paths outside the working directories.

Example: {"permissions": {"blockWritesOutsideWorkingDirectories": true}}

Source: docsSince v2.1.314Any settings file

Deny tool calls that would wait on a permission prompt, so a session nobody is watching keeps working.

Example: {"permissions": {"defaultMode": "auto", "denyInsteadOfPrompting": true}}

Source: docsSince v2.1.319User or managed settings

Prevent anyone from entering bypassPermissions mode.

Example: {"permissions": {"disableBypassPermissionsMode": "disable"}}

Source: docsAny settings file

Choose where plan mode writes plan files.

Example: {"plansDirectory": "./plans"}

Source: docsSince v2.1.9Any settings file

Store the answers you gave a plugin's configuration dialog.

Source: docsUser or managed settings

Choose which marketplaces can surface plugin install suggestions in /plugin.

Example: {"pluginSuggestionMarketplaces": ["acme-corp-plugins"]}

Source: docsSince v2.1.152Managed settings only

Add your own text to the plugin trust warning.

Example: {"pluginTrustMessage": "All plugins from our marketplace are approved by IT"}

Source: docsManaged settings only

Run an executable that computes managed settings at startup.

Source: docsManaged settings only

Name the helper executable Claude Code runs.

Example: {"policyHelper": {"path": "/usr/local/bin/claude-policy"}}

Source: docsManaged settings only

Re-run the helper in the background on an interval.

Source: docsManaged settings only

Set how long Claude Code waits for the helper.

Example: {"policyHelper": {"path": "/usr/local/bin/claude-policy", "timeoutMs": 5000}}

Source: docsManaged settings only

Choose a terminal bell or desktop notification for task completion.

Example: {"preferredNotifChannel": "terminal_bell"}

Source: docsAny settings file

Reduce or turn off spinner, shimmer, and flash animations.

Example: {"prefersReducedMotion": true}

Source: docsAny settings file

Run your organization's mods before every mod a user installs.

Source: docsUser or managed settings

Run Claude Code's background processes through a corporate launcher on macOS and Linux.

Example: {"processWrapper": "/opt/corp/launcher --profile claude"}

Source: docsSince v2.1.210User or managed settings

Choose the prompt cache lifetime for the main conversation.

Example: {"promptCacheTtl": "1h", "subagentPromptCacheTtl": "5m"}

Source: docsAny settings file

Hide the grayed-out prompt suggestions in the input box.

Example: {"promptSuggestionEnabled": false}

Source: docsAny settings file

Turn off the prompt footer's PR review status badge and the pull request check behind it.

Example: {"prStatusFooterEnabled": false}

Source: docsGlobal config, ~/.claude.json

Point PR links at an internal code-review tool instead of github.com.

Example: {"prUrlTemplate": "https://reviews.example.com/{owner}/{repo}/pull/{number}"}

Source: docsSince v2.1.119Any settings file

Pick the default cloud environment for claude --cloud.

Example: {"remote": {"defaultEnvironmentId": "env_0123abcd"}}

Source: docsAny settings file

Connect Remote Control automatically when a session starts.

Example: {"remoteControlAtStartup": true}

Source: docsAny settings file

Refuse to start on a version newer than your organization allows.

Example: {"requiredMaximumVersion": "2.1.150"}

Source: docsSince v2.1.163Managed settings only

Refuse to start on a version older than your organization requires.

Example: {"requiredMinimumVersion": "2.1.150"}

Source: docsSince v2.1.163Managed settings only

Keep gitignored files out of the @ file picker.

Example: {"respectGitignore": false}

Source: docsAny settings file

Stop Claude from responding after a ! shell command runs.

Example: {"respondToBashCommands": false}

Source: docsAny settings file

Isolate Bash commands from your filesystem and network on macOS, Linux, and WSL2.

Source: docsAny settings file

Let sandboxed commands send Apple Events on macOS.

Example: {"sandbox": {"enabled": true, "allowAppleEvents": true}}

Source: docsSince v2.1.181User or managed settings

Let sandboxed commands create user namespaces on Linux and WSL2.

Example: {"sandbox": {"enabled": true, "allowNestedUserNamespaces": true}}

Source: docsSince v2.1.305User or managed settings

Let Claude retry a blocked command outside the sandbox, or forbid it.

Example: {"sandbox": {"enabled": true, "allowUnsandboxedCommands": false}}

Source: docsAny settings file

Run sandboxed commands without a permission prompt.

Example: {"sandbox": {"enabled": true, "autoAllowBashIfSandboxed": false}}

Source: docsAny settings file

Point the sandbox at a bubblewrap binary outside PATH.

Example: {"sandbox": {"enabled": true, "bwrapPath": "/opt/admin/bwrap"}}

Source: docsManaged settings only

Hide or mask credential files and variables inside the sandbox.

Source: docsSince v2.1.187Any settings file

Let masked credentials reach plain HTTP services on trusted test networks.

Example: {"sandbox": {"credentials": {"allowPlaintextInject": true}}}

Source: docsUser or managed settings

Link custom-named AWS key variables into one credential for re-signing.

Source: docsSince v2.1.224User or managed settings

Unset or mask an environment variable inside the sandbox.

Source: docsAny settings file

Block or mask reads of a credential file inside the sandbox.

Source: docsAny settings file

Choose whether streaming, presigned, or SigV4A AWS requests fail or pass through.

Example: {"sandbox": {"credentials": {"sigv4": {"streaming": "passthrough"}}}}

Source: docsSince v2.1.224User or managed settings

Turn on Bash sandboxing on macOS, Linux, and WSL2.

Example: {"sandbox": {"enabled": true}}

Source: docsAny settings file

Run the Linux sandbox inside an unprivileged container.

Example: {"sandbox": {"enabled": true, "enableWeakerNestedSandbox": true}}

Source: docsAny settings file

Let gh, gcloud, and terraform verify TLS behind a MITM proxy inside the sandbox on macOS.

Example: {"sandbox": {"enabled": true, "enableWeakerNetworkIsolation": true}}

Source: docsSince v2.1.69Any settings file

Name commands Claude Code can run outside the sandbox.

Example: {"sandbox": {"excludedCommands": ["docker *"]}}

Source: docsAny settings file

Refuse to start when the sandbox can't, instead of running unsandboxed.

Example: {"sandbox": {"enabled": true, "failIfUnavailable": true}}

Source: docsSince v2.1.83Any settings file

Control which paths sandboxed commands can read and write.

Source: docsAny settings file

Stop developers from re-opening read paths your organization blocked.

Source: docsManaged settings only

Re-open reading inside a region denyRead blocks.

Example: {"sandbox": {"filesystem": {"denyRead": ["~/"], "allowRead": ["."]}}}

Source: docsAny settings file

Add paths sandboxed commands can write to.

Example: {"sandbox": {"filesystem": {"allowWrite": ["/tmp/build", "~/.kube"]}}}

Source: docsAny settings file

Block sandboxed commands from reading specific paths.

Example: {"sandbox": {"filesystem": {"denyRead": ["~/.aws/credentials"]}}}

Source: docsAny settings file

Block sandboxed commands from writing to specific paths.

Example: {"sandbox": {"filesystem": {"denyWrite": ["/etc", "/usr/local/bin"]}}}

Source: docsAny settings file

Turn off filesystem isolation while keeping network isolation.

Source: docsSince v2.1.216User or managed settings

Silence violation reports for paths a command is expected to probe.

Example: {"sandbox": {"ignoreViolations": {"*": ["/etc/hosts"]}}}

Source: docsAny settings file

Control which hosts, ports, and sockets sandboxed commands reach.

Source: docsAny settings file

Let sandboxed commands connect to every Unix socket.

Example: {"sandbox": {"network": {"allowAllUnixSockets": true}}}

Source: docsAny settings file

Pre-allow domains so sandboxed commands don't prompt for them.

Source: docsAny settings file

Let sandboxed commands listen on network ports and connect to localhost on macOS.

Example: {"sandbox": {"network": {"allowLocalBinding": true}}}

Source: docsAny settings file

Let macOS sandboxed tools like the iOS Simulator or Playwright reach their XPC services.

Example: {"sandbox": {"network": {"allowMachLookup": ["com.apple.CoreSimulator.*"]}}}

Source: docsAny settings file

Lock the network allowlist to managed settings.

Source: docsManaged settings only

List Unix socket paths sandboxed commands can use on macOS.

Example: {"sandbox": {"network": {"allowUnixSockets": ["~/.ssh/agent-socket"]}}}

Source: docsAny settings file

Block domains for sandboxed commands, even inside an allowed wildcard.

Example: {"sandbox": {"network": {"deniedDomains": ["sensitive.cloud.example.com"]}}}

Source: docsSince v2.1.113Any settings file

Route sandbox HTTP traffic through your own proxy.

Example: {"sandbox": {"network": {"httpProxyPort": 8080}}}

Source: docsAny settings file

Route sandbox SOCKS traffic through your own proxy.

Example: {"sandbox": {"network": {"socksProxyPort": 8081}}}

Source: docsAny settings file

Deny hosts outside the allowlist instead of prompting.

Example: {"sandbox": {"network": {"strictAllowlist": true}}}

Source: docsSince v2.1.219User or managed settings

Have the sandbox proxy terminate TLS so it can read HTTPS requests.

Example: {"sandbox": {"network": {"tlsTerminate": {}}}}

Source: docsUser or managed settings

Use your own ripgrep binary inside the sandbox.

Example: {"sandbox": {"ripgrep": {"command": "/usr/local/bin/rg"}}}

Source: docsUser or managed settings

Point the sandbox proxy at a socat binary outside PATH.

Example: {"sandbox": {"enabled": true, "socatPath": "/opt/admin/socat"}}

Source: docsManaged settings only

Show a "clear context" option on the plan accept screen.

Example: {"showClearContextOnPlanAccept": true}

Source: docsAny settings file

See summaries of Claude's thinking instead of a collapsed stub.

Example: {"showThinkingSummaries": true}

Source: docsAny settings file

Hide the "Cooked for" duration after each response.

Example: {"showTurnDuration": false}

Source: docsSince v2.1.7Any settings file

Reserve more or less context for the skill listing.

Example: {"skillListingBudgetFraction": 0.02}

Source: docsAny settings file

Cap each skill's description length in the skill listing.

Example: {"skillListingMaxDescChars": 2048}

Source: docsAny settings file

Hide or collapse a skill without editing its SKILL.md.

Example: {"skillOverrides": {"legacy-context": "name-only", "deploy": "off"}}

Source: docsAny settings file

Skip the one-time notice shown when you first enter auto mode yourself.

Example: {"skipAutoPermissionPrompt": true}

Source: docsUser or managed settings

Skip the confirmation dialog before bypassPermissions mode.

Example: {"skipDangerousModePermissionPrompt": true}

Source: docsUser, local, or managed

Skip the WebFetch hostname check when Anthropic is unreachable.

Example: {"skipWebFetchPreflight": true}

Source: docsAny settings file

Underline misspelled words in the prompt input with a spell checker you install.

Example: {"spellcheck": {"enabled": true, "language": "en_GB"}}

Source: docsSince v2.1.235User or managed settings

Hide tips in the spinner while Claude works.

Example: {"spinnerTipsEnabled": false}

Source: docsSince v1.0.112Any settings file

Add your own tips to the spinner rotation, or replace the built-in tips.

Source: docsSince v2.1.45Any settings file

Add or replace the verbs shown while a turn runs.

Example: {"spinnerVerbs": {"mode": "append", "verbs": ["Pondering", "Crafting"]}}

Source: docsAny settings file

Add SSH connections to the Desktop environment selector.

Source: docsUser or managed settings

Limit which hosts Desktop SSH sessions can reach.

Example: {"sshHostAllowlist": ["*.devboxes.example.com", "bastion.example.com"]}

Source: docsManaged settings only

Add SSH connections whose host comes from a command to the Desktop environment selector.

Source: docsUser or managed settings

Allowlist the marketplace sources users can add and install from.

Source: docsManaged settings only

Block skills, agents, hooks, and MCP servers from user and project sources.

Example: {"strictPluginOnlyCustomization": ["skills", "hooks"]}

Source: docsManaged settings only

Block agents from user and project sources.

Example: {"strictPluginOnlyCustomization": ["agents"]}

Source: docsManaged settings only

Block hooks from user and project sources.

Example: {"strictPluginOnlyCustomization": ["hooks"]}

Source: docsManaged settings only

Block MCP servers from user and project sources.

Example: {"strictPluginOnlyCustomization": ["mcp"]}

Source: docsManaged settings only

Block skills from user and project sources.

Example: {"strictPluginOnlyCustomization": ["skills"]}

Source: docsManaged settings only

Choose the prompt cache lifetime for subagents and other requests outside the main conversation.

Example: {"subagentPromptCacheTtl": "1h"}

Source: docsAny settings file

Rewrite rows in the subagent task display with your own command.

Source: docsAny settings file

Switch models automatically or pause when a safety classifier flags a request.

Example: {"switchModelsOnFlag": false}

Source: docsAny settings file

Stop loading the plugins enabled on your claude.ai account and stop downloading new ones.

Example: {"syncClaudeAiPlugins": false}

Source: docsSince v2.1.273User, local, or managed

Stop loading the skills enabled on your claude.ai account and stop downloading new ones.

Example: {"syncClaudeAiSkills": false}

Source: docsUser, local, or managed

Turn off syntax highlighting in diffs and code blocks.

Example: {"syntaxHighlightingDisabled": true}

Source: docsAny settings file

Choose how agent team teammates display.

Example: {"teammateMode": "auto"}

Source: docsAny settings file

Hide the terminal progress bar in terminals that support it.

Example: {"terminalProgressBarEnabled": false}

Source: docsAny settings file

Stop /rename and --name from changing the terminal tab title.

Example: {"terminalTitleFromRename": false}

Source: docsAny settings file

Pick the interface color theme, built-in or custom.

Example: {"theme": "light-daltonized"}

Source: docsAny settings file

Show the times in the interface on a 12-hour or 24-hour clock, in UTC, or with a strftime pattern.

Example: {"timeFormat": "24-hour"}

Source: docsSince v2.1.257Any settings file

Show the times in the interface in a time zone other than your system's.

Example: {"timeZone": "Europe/Dublin"}

Source: docsSince v2.1.257Any settings file

Choose the fullscreen or classic terminal renderer.

Example: {"tui": "fullscreen"}

Source: docsAny settings file

Have Claude plan a workflow for each substantive task without being asked.

Example: {"ultracode": true}

Source: docsAny settings file

Let the auto mode classifier review shell commands in plan mode.

Example: {"useAutoModeDuringPlan": false}

Source: docsUser, local, or managed

Show full tool output instead of truncated summaries.

Example: {"verbose": true}

Source: docsAny settings file

Start every session in default, verbose, or focus view.

Example: {"viewMode": "focus"}

Source: docsAny settings file

Map a two-key INSERT-mode sequence such as jj to Escape.

Example: {"vimInsertModeRemaps": {"jj": "<Esc>"}}

Source: docsSince v2.1.208User or managed settings

Turn on voice dictation and pick hold or tap mode.

Example: {"voice": {"enabled": true, "mode": "tap"}}

Source: docsAny settings file

Turn on voice dictation with the older single-key form.

Example: {"voiceEnabled": true}

Source: docsAny settings file

Turn off mouse-wheel acceleration in fullscreen rendering.

Example: {"wheelScrollAccelerationEnabled": false}

Source: docsSince v2.1.174Any settings file

Let the word ultracode in a prompt start a workflow.

Example: {"workflowKeywordTriggerEnabled": false}

Source: docsAny settings file

Set the agent count Claude aims for in dynamic workflows.

Example: {"workflowSizeGuideline": "small"}

Source: docsSince v2.1.219Any settings file

Configure how Claude Code creates git worktrees.

Example: {"worktree": {"baseRef": "head", "symlinkDirectories": ["node_modules"]}}

Source: docsAny settings file

Branch new worktrees from the remote default branch or your local HEAD.

Example: {"worktree": {"baseRef": "head"}}

Source: docsSince v2.1.133Any settings file

Let background sessions edit the working copy without a worktree.

Example: {"worktree": {"bgIsolation": "none"}}

Source: docsAny settings file

Choose where Desktop SSH sessions create their worktrees on a remote machine.

Example: {"worktree": {"location": "~/worktrees"}}

Source: docsUser settings only

Check out only the directories you need in each worktree.

Example: {"worktree": {"sparsePaths": [".claude", "packages/my-app", "shared/utils"]}}

Source: docsSince v2.1.76Any settings file

Symlink large directories into each worktree instead of duplicating them.

Example: {"worktree": {"symlinkDirectories": ["node_modules", ".cache"]}}

Source: docsAny settings file

Have WSL read managed settings from the Windows policy chain.

Example: {"wslInheritsWindowsSettings": true}

Source: docsSince v2.1.282Managed settings only

Sets a custom Authorization header value. Claude Code adds Bearer in front.

Source: docs

Workspace API key for Claude Platform on AWS, generated in the AWS Console.

Source: docs

Override the Claude Platform on AWS endpoint URL.

Source: docs

Required for Claude Platform on AWS.

Source: docs

Override the Amazon Bedrock endpoint URL.

Source: docs

Override the Amazon Bedrock Mantle endpoint URL.

Source: docs

Sets the cross-region inference profile prefix Claude Code tries first (us, eu, apac, jp, au, or global).

Source: docsSince v2.1.224

Amazon Bedrock service tier (default, flex, or priority).

Source: docsSince v2.1.122

Comma-separated list of additional anthropic-beta header values to include in API requests.

Source: docs

Adds custom headers to API requests, in Name: Value format.

Source: docsSince v2.1.227

Model ID to add as a custom entry in the /model picker.

Source: docsSince v2.1.78

Display description for the custom model entry in the /model picker.

Source: docs

Display name for the custom model entry in the /model picker.

Source: docs

Lists the capabilities the custom model supports, such as effort,thinking.

Source: docs

Sets the model ID that the fable alias points to.

Source: docs

Display description for the pinned Fable model in the /model picker.

Source: docs

Display name for the pinned Fable model in the /model picker.

Source: docs

Lists the capabilities the pinned Fable model supports, such as effort,thinking.

Source: docs

Sets the model ID that the haiku alias points to. Also used for background tasks.

Source: docs

Display description for the pinned Haiku model in the /model picker.

Source: docs

Display name for the pinned Haiku model in the /model picker.

Source: docs

Lists the capabilities the pinned Haiku model supports, such as effort,thinking.

Source: docs

Model that new sessions start on by default.

Source: docsSince v2.1.236

Sets the model ID that the opus alias points to. opusplan uses it in Plan Mode.

Source: docs

Display description for the pinned Opus model in the /model picker.

Source: docs

Display name for the pinned Opus model in the /model picker.

Source: docs

Lists the capabilities the pinned Opus model supports, such as effort,thinking.

Source: docs

Sets the model ID that the sonnet alias points to. opusplan uses it outside Plan Mode.

Source: docs

Display description for the pinned Sonnet model in the /model picker.

Source: docs

Display name for the pinned Sonnet model in the /model picker.

Source: docs

Lists the capabilities the pinned Sonnet model supports, such as effort,thinking.

Source: docs

Federation rule ID for Workload Identity Federation.

Source: docs

API key for Microsoft Foundry authentication (see Microsoft Foundry).

Source: docs

Bearer token for Microsoft Foundry authentication, such as a Microsoft Entra access token.

Source: docsSince v2.1.203

Sets the full base URL of the Microsoft Foundry resource.

Source: docs

Sets the Microsoft Foundry resource name.

Source: docs

Organization ID for Workload Identity Federation.

Source: docs

Names the Anthropic profile to sign in with, such as one made by ant auth login.

Source: docs

Deprecated. Sets the Haiku-class model for background tasks.

Source: docs

Override AWS region for the Haiku-class model when using Amazon Bedrock or Amazon Bedrock Mantle.

Source: docs

Override Google Cloud's Agent Platform endpoint URL.

Source: docs

GCP project ID that Google Cloud's Agent Platform requests are addressed to.

Source: docs

Workspace ID for workload identity federation.

Source: docsSince v2.1.141

Changes the 5-minute idle timeout that stops a streaming reply when no bytes arrive.

Source: docs

Sets how long each API request waits for response headers, in milliseconds.

Source: docs

Amazon Bedrock API key for authentication (see Amazon Bedrock API keys).

Source: docs

Sets the most characters of command output Claude Code reads back into a result.

Source: docs

Sets the longest timeout the model can pick for a foreground command, in milliseconds.

Source: docs

Sets the OTLP/HTTP endpoint for detailed beta tracing.

Source: docs

Set to 1 to make claude --cloud upload your local repository instead of cloning it.

Source: docs

Set to 1 to turn off built-in subagent types such as Explore and Plan.

Source: docs

Set to 1 to skip the mcp__<server>__ prefix on tool names from SDK-created MCP servers.

Source: docs

Sets the stall timeout for subagents and workflow agents, in milliseconds.

Source: docs

Set to 1 to move long-running agent tasks to the background automatically.

Source: docs

Sets what percent (1 to 100) of the auto-compact window triggers auto-compaction.

Source: docs

In screen reader mode, sets how long Claude Code waits before writing a new or changed line.

Source: docsSince v2.1.233

Set to 1 to render screen-reader friendly output: flat text without decorative borders or animations.

Source: docsSince v2.1.181

In screen reader mode, sets how long Claude Code holds the first render after startup.

Source: docsSince v2.1.217

Returns to the original working directory after each Bash or PowerShell command.

Source: docs

Sets the timeout for the byte-level streaming idle watchdog, in milliseconds.

Source: docsSince v2.1.210

Points to a file an outside tool creates when you unlock your screen and deletes when you lock it.

Source: docsSince v2.1.181

Set to 1 to keep the native cursor visible and turn off the inverted-text cursor.

Source: docs

Set to 1 to load memory files from directories added with --add-dir.

Source: docs

Set to 1 to repaint the whole screen on every frame in fullscreen rendering.

Source: docs

Set to 1 to send the effort setting with every request, even for unknown model IDs.

Source: docs

Sets how often credentials from apiKeyHelper refresh, in milliseconds.

Source: docs

Set to 0 to stop Claude Code from opening the browser when an artifact is published.

Source: docs

Set to 0 to stop Claude from reading and replying to comments on an artifact.

Source: docsSince v2.1.221

Set to 0 to stop Claude from replying to comments on its own.

Source: docsSince v2.1.228

Set to 0 to stop Claude from making small edits to an artifact's file when publishing.

Source: docsSince v2.1.300

Set to 0 to leave the attribution block out of the start of the system prompt.

Source: docs

Sets how full the context gets before compaction, from 100000 to 1000000 tokens.

Source: docs

Overrides automatic IDE connection. false stops it, true forces a try.

Source: docs

Sets the model for the classifier requests Claude Code sends in auto mode.

Source: docsSince v2.1.302

Controls whether Claude Code asks the server to review auto mode actions. 0 uses its own classifier.

Source: docs

Sets how long Claude Code waits for the AWS default credential chain, in milliseconds.

Source: docsSince v2.1.207

0 turns off the diff of files a Bash command changed. 1 records it in every permission mode.

Source: docsSince v2.1.269

Set to 0 to report an idle status at each turn end in non-interactive sessions, even with background work running.

Source: docs

Set automatically in shell and hook processes while a Remote Control connection is active.

Source: docsSince v2.1.199

0 reads the 0x08 byte as Backspace. 1 reads it as Ctrl+Backspace.

Source: docs

Lists the CA certificate sources for TLS connections. The default is bundled,system.

Source: docs

Set to 1 in processes that Claude Code starts, such as shell tools and hooks.

Source: docsSince v2.1.172

Path to client certificate file for mTLS authentication.

Source: docs

Path to client private key file for mTLS authentication.

Source: docs

Passphrase for encrypted CLAUDE\_CODE\_CLIENT\_KEY (optional).

Source: docs

Sets how long claude mcp add and similar commands wait for the ~/.claude.json lock, in milliseconds.

Source: docsSince v2.1.318

Minimum log level written to the debug log file.

Source: docs

Override the debug log file path.

Source: docs

Set to 1 to turn off 1M context window support.

Source: docsSince v2.1.50

Set to 1 to use a fixed thinking budget (MAX_THINKING_TOKENS) instead of adaptive reasoning on Opus 4.6 and Sonnet 4.6.

Source: docs

Set to 1 to disable the advisor tool.

Source: docs

Set to 1 to turn off background agents and agent view.

Source: docs

Set to 1 to disable fullscreen rendering and use the classic main-screen renderer.

Source: docs

Set to 1 to turn off the Artifact tool.

Source: docs

Set to 1 to disable attachment processing.

Source: docs

Set to 1 to run the auth refresh command in each process instead of waiting for another process.

Source: docsSince v2.1.286

Set to 1 to disable auto memory.

Source: docs

Set to 1 to remove background runs, the run_in_background parameter, and the Ctrl+B shortcut.

Source: docsSince v2.1.4

Set to 1 to stop treating a Bedrock stream with no Content-Type as a binary event stream.

Source: docsSince v2.1.239

Set to 1 to skip the check that a Bedrock stream has the event-stream content type.

Source: docsSince v2.1.208

Set to 1 to stop a background session's work when its process restarts, instead of handing it over.

Source: docsSince v2.1.196

Set to 1 to stop Claude Code from ending background shell commands under memory pressure.

Source: docsSince v2.1.193

On macOS with the native installer, set to 1 to run background sessions from the installed binary.

Source: docsSince v2.1.295

Set to 1 to remove the bundled skills and workflows.

Source: docs

Set to 1 to leave the Chrome section out of the system prompt while keeping the Chrome tools.

Source: docsSince v2.1.257

Set to 1 to stop loading any CLAUDE.md memory files, including auto memory.

Source: docs

Set to 1 to disable session-scoped scheduled tasks.

Source: docsSince v2.1.72

Set to 1 to turn off the time limit and denials for critical-path removals in auto and bypassPermissions modes.

Source: docsSince v2.1.281

Set to 1 to remove pre-release beta headers and fields from API requests.

Source: docs

Set to 1 to disable the built-in Explore and Plan subagents.

Source: docsSince v2.1.198

Set to 1 to disable fast mode.

Source: docs

Set to 1 to turn off feature-flag fetching.

Source: docsSince v2.1.319

Set to 1 to disable the "How is Claude doing?" session quality surveys.

Source: docs

Set to 1 to disable file checkpointing.

Source: docs

Set to 1 to stop Claude Code from expanding @ mentions of files and directories.

Source: docsSince v2.1.299

Set to 1 to remove the built-in commit and PR instructions and the git status snapshot.

Source: docs

Set to 1 to stop reading scripts passed to a shell with -c for critical-path removals.

Source: docsSince v2.1.288

Set to 1 to stop remapping Opus 4 and Opus 4.1 to the current Opus version.

Source: docs

Set to 1 so Claude waits for an MCP task result in an interactive session.

Source: docs

Set to 1 to stop the switch to an older model when your account loses access mid-session.

Source: docsSince v2.1.285

Set to 1 to turn off the mounted drive prompt for paths such as /Volumes/USB.

Source: docsSince v2.1.304

Set to 1 to disable mouse tracking in fullscreen rendering.

Source: docs

Set to 1 to turn off click, drag, and hover in fullscreen rendering but keep wheel scrolling.

Source: docsSince v2.1.195

Set to 1 to stop re-reading the mTLS certificate and key after a connection-level error.

Source: docsSince v2.1.232

Set to any non-empty value to turn off nonessential network traffic, such as auto-updates and telemetry.

Source: docs

Set to 1 to turn off the non-streaming fallback when a streaming request fails mid-stream.

Source: docsSince v2.1.83

Set to 1 to send desktop notifications even while you are typing in the terminal.

Source: docsSince v2.1.193

Set to 1 to stop automatic registration of the official plugin marketplace.

Source: docs

Set to 1 to stop Notification hooks from running for unanswered permission requests sent to the Agent SDK.

Source: docsSince v2.1.233

Set to 1 to skip loading skills from the system-wide managed skills directory.

Source: docs

Set to 1 to turn off the check that denies rd, rmdir, del, and erase on system paths in PowerShell.

Source: docsSince v2.1.283

Set to 1 to turn off the automatic model switch when a safety classifier flags a request.

Source: docs

Set to 1 to stop sending the structured-output field and its beta header.

Source: docsSince v2.1.288

Set to 1 to turn off the critical-path check for rm on the output of a command substitution.

Source: docsSince v2.1.281

Set to 1 to stop automatic terminal title updates.

Source: docs

Set to 1 to omit the thinking parameter from API requests entirely.

Source: docs

Set to 1 so claude ultrareview skips the trust check in an untrusted folder, such as in CI.

Source: docsSince v2.1.309

Set to 1 to skip early auto-compaction when Claude Code does not know the model ID.

Source: docsSince v2.1.223

Set to 1 to turn off virtual scrolling in fullscreen rendering.

Source: docs

Set to 1 to turn off the WebFetch tool.

Source: docsSince v2.1.285

Set to 1 to start PowerShell commands on Windows directly, not through cmd.exe.

Source: docsSince v2.1.269

Set to 1 to disable workflows.

Source: docs

Set to 1 to drop a deferred tool call whose tool is missing when you resume with -p.

Source: docsSince v2.1.319

Set to 1 to add session_state_changed messages to the message stream.

Source: docs

Accepted for older releases. It has no effect.

Source: docs

Overrides session recap availability. 0 forces recaps off, 1 forces them on.

Source: docs

Set to 1 to refresh plugin state at turn ends in non-interactive mode after a background install.

Source: docs

1 starts a CLI session with Chrome integration on. 0 starts it off.

Source: docs

Set to 1 to send the session quality survey to your own OpenTelemetry collector.

Source: docsSince v2.1.136

Controls whether tool call inputs stream from the API as Claude writes them.

Source: docs

Set to 1 to fill the /model picker from your gateway's /v1/models endpoint.

Source: docs

Set to false to turn off prompt suggestions, the grayed-out predictions in the prompt box.

Source: docsSince v2.1.238

Set to 1 to make /share save its report and transcript to a local archive instead of sending it.

Source: docsSince v2.1.299

Selects which task-tracking tools Claude Code provides.

Source: docs

Set to 1 to enable OpenTelemetry data collection for metrics and logging.

Source: docs

Set to 1 to get the task-tracking tools on every model.

Source: docsSince v2.1.233

Set to 1 to run dangerous rm commands with no prompt in bypassPermissions mode.

Source: docsSince v2.1.311

Sets how long to wait after the query loop goes idle before exiting, in milliseconds.

Source: docsSince v2.0.35

Set to 1 to enable agent teams.

Source: docs

Merges a JSON object into the top level of model request bodies.

Source: docs

Changes the token limit for file reads. The default is 25,000 tokens.

Source: docsSince v2.1.0

Set to 1 to save transcripts and history even when launched inside another Claude Code session.

Source: docs

Set to 1 to force strikethrough for ~~text~~ when your terminal supports it but is not detected.

Source: docsSince v2.1.186

Set to 1 to force synchronized output when your terminal supports it but is not detected.

Source: docs

Set to 1 to draw mod Image elements as pictures in terminals that support kitty graphics.

Source: docs

Controls fork mode, which is on by default. 0 turns it off.

Source: docs

Set to 1 to emit subagent text and thinking in stream-json output.

Source: docsSince v2.1.211

Set to 1 to send gateway hint headers on a custom proxy or third-party provider.

Source: docsSince v2.1.273

Sets the timeout for gateway model discovery, in milliseconds. The default is 3000.

Source: docsSince v2.1.269

Windows only: path to the Git Bash executable (bash.exe).

Source: docs

Set to false to leave dotfiles out of Glob tool results.

Source: docs

Set to false to make the Glob tool respect .gitignore patterns.

Source: docs

Timeout in seconds for Glob and Grep tool searches.

Source: docs

Sets how many minutes background work can keep a goal waiting before Claude Code asks Claude to check.

Source: docsSince v2.1.234

Set to 0 to turn off gzip compression of request bodies sent to api.anthropic.com.

Source: docs

Set to 1 to hide the working directory in the startup logo.

Source: docsSince v2.1.119

Override the host address used to connect to the IDE extension.

Source: docs

Set to 1 to stop /ide from installing the IDE extension.

Source: docs

Set to 1 to skip checking IDE lockfile entries when connecting.

Source: docs

Sets the context size, in tokens, at which idle compaction applies. The default is 200000.

Source: docs

Sets how many subagents can run in one session. The default is 20.

Source: docsSince v2.1.217

Override the context window size Claude Code assumes for the active model.

Source: docs

Sets the most characters of MCP tool descriptions and server instructions sent to the model.

Source: docsSince v2.1.280

Set the maximum number of output tokens for most requests.

Source: docs

Override the number of times to retry failed API requests (default: 10).

Source: docs

Sets how many subagent layers can sit below the main conversation. The default is 3.

Source: docsSince v2.1.217

Cap the number of agentic turns when no explicit limit is passed.

Source: docs

Cap on WebSearch calls (default: 200).

Source: docsSince v2.1.212

Set to 1 to start stdio MCP servers with only a safe baseline environment plus their own env.

Source: docs

Sets how long an MCP tool call runs before moving to the background, in milliseconds. The default is 120000.

Source: docsSince v2.1.212

Sets how long a non-interactive first turn waits for MCP servers to connect, in milliseconds.

Source: docsSince v2.1.274

Idle timeout in milliseconds for MCP tool calls.

Source: docsSince v2.1.187

Set by Claude Code. It holds the path of the session's inbox socket.

Source: docsSince v2.1.224

Set by Claude Code. It holds a per-session token for the inbox socket.

Source: docsSince v2.1.228

Set to 1 to show the terminal's own cursor at the input caret.

Source: docs

Set to 1 to make /init run an interactive setup flow.

Source: docs

Set to 1 to turn on fullscreen rendering, a research preview that reduces flicker.

Source: docs

Set to 1 to write output through a non-blocking file descriptor so a stalled terminal can't freeze Claude Code.

Source: docsSince v2.1.261

Limit how many times Claude Code re-sends a non-streaming request that times out.

Source: docsSince v2.1.285

OAuth refresh token for claude.ai authentication.

Source: docs

Lists the OAuth scopes the refresh token was issued with, separated by spaces.

Source: docs

OAuth access token for claude.ai authentication.

Source: docs

Sets the maximum length of content-bearing OpenTelemetry attributes. The default is 61440.

Source: docsSince v2.1.214

Set to 1 to write OpenTelemetry exporter errors to stderr.

Source: docsSince v2.1.179

Sets the timeout for flushing pending OpenTelemetry spans, in milliseconds. The default is 5000.

Source: docs

Sets how often dynamic OpenTelemetry headers refresh, in milliseconds.

Source: docs

Timeout in milliseconds for the OpenTelemetry exporter to finish on shutdown (default: 2000).

Source: docs

Sets the starting delay for retries after a 529 overloaded error, in milliseconds. The default is 500.

Source: docsSince v2.1.292

Sets the longest delay for retries after a 529 overloaded error, in milliseconds. The default is 32000.

Source: docsSince v2.1.296

Set to 1 to let Claude Code run your package manager's upgrade command in the background.

Source: docsSince v2.1.129

Set to 1 to turn on Perforce-aware write protection.

Source: docsSince v2.1.98

Overrides the plugins root directory. The default is ~/.claude/plugins.

Source: docs

Controls whether Claude Code reloads a mod when its files change.

Source: docsSince v2.1.287

Lists plugin directories to load for the session, like --plugin-dir.

Source: docsSince v2.1.280

Sets the timeout for cloning or refreshing a plugin marketplace, in milliseconds. The default is 120000.

Source: docs

Set to 1 to keep the existing marketplace checkout when a refresh can't reach the remote.

Source: docsSince v2.1.90

Set to 1 to clone GitHub owner/repo sources over HTTPS instead of SSH.

Source: docsSince v2.1.141

Points to one or more read-only plugin seed directories.

Source: docs

Set to 1 to stop passing -ExecutionPolicy Bypass to PowerShell and respect the machine's policy.

Source: docs

Sets the longest idle wait for background work after the final turn with -p, in milliseconds.

Source: docsSince v2.1.182

Starts the processes Claude Code launches from its own binary through a corporate launcher.

Source: docsSince v2.1.208

With CLAUDE_CONFIG_DIR, sets the projects/ directory name for a session's transcripts and auto memory.

Source: docsSince v2.1.234

Sets the prompt cache TTL for the main conversation. Accepts 5m or 1h.

Source: docsSince v2.1.242

Set to 1 to pass W3C trace context when ANTHROPIC_BASE_URL points at a custom proxy.

Source: docs

Set by host apps that manage model provider routing for Claude Code.

Source: docs

Set to 1 to let the proxy do DNS resolution instead of the caller.

Source: docs

Set automatically to true when Claude Code is running as a cloud session.

Source: docs

Set automatically in cloud sessions to the current session's ID.

Source: docs

Set to 1 to start the session in restricted mode, the same as passing --restricted.

Source: docsSince v2.1.248

Set to 1 to resume automatically if the last session ended mid-turn.

Source: docs

Sets how old the last message can be for an interrupted turn to resume automatically, in milliseconds.

Source: docsSince v2.1.211

Overrides the message Claude Code sends when it continues an interrupted turn.

Source: docs

Set to 1 to keep interrupted tool calls, marked as cut off, when a turn continues.

Source: docsSince v2.1.309

Set to 1 for unattended sessions. It retries 429 and 529 errors without limit.

Source: docs

Sets the longest each request waits out 429 and 529 errors with the retry watchdog, in milliseconds.

Source: docsSince v2.1.295

Sets the shortest gap between checks for a lifted usage limit with the retry watchdog, in milliseconds.

Source: docsSince v2.1.305

Set to 1 to start in safe mode, without CLAUDE.md, skills, plugins, hooks, or MCP servers.

Source: docs

Sets a JSON object that limits how often named scripts can run per session.

Source: docs

Set the mouse wheel scroll multiplier in fullscreen rendering.

Source: docs

Set to 0 to turn off Claude-drafted feedback for a session.

Source: docs

Set automatically to the current session ID in shell, hook, and stdio MCP processes.

Source: docsSince v2.1.132

Overrides the time budget for SessionEnd hooks, in milliseconds.

Source: docs

Set the shell Claude Code uses to run Bash tool commands.

Source: docsSince v2.0.65

Sets a command prefix that wraps shell commands Claude Code starts.

Source: docsSince v1.0.61

Set to 1 to run with a minimal system prompt and only the Bash, file read, and file edit tools.

Source: docs

Picks between the full system prompt and a shorter one with abbreviated tool descriptions.

Source: docs

Skips client-side sign-in for Claude Platform on AWS, for gateways that sign requests.

Source: docs

Set to 1 to turn off the in-process credential cache and resolve AWS credentials on every request.

Source: docsSince v2.1.207

Skip AWS authentication for Amazon Bedrock (for example, when using an LLM gateway).

Source: docs

Set to 1 to treat a failed fast mode check as available.

Source: docs

Set to 1 to skip the client-side fast mode availability check.

Source: docs

Skips Azure sign-in for Microsoft Foundry, for a proxy that adds its own Authorization header.

Source: docs

Skip AWS authentication for Amazon Bedrock Mantle (for example, when using an LLM gateway).

Source: docs

Set to 1 to stop remembering which models your account can't use on Bedrock and Google Cloud.

Source: docsSince v2.1.285

Set to 1 to skip writing prompt history and session transcripts to disk.

Source: docs

Skip Google authentication for Google Cloud's Agent Platform (for example, when using an LLM gateway).

Source: docs

Set to 1 to write a result message when Claude Code refuses to start in stream-json mode.

Source: docsSince v2.1.274

Sets how many times in a row a Stop hook can block the turn from ending. The default is 8.

Source: docs

Sets the default model for subagents, teammates, and workflow agents with no model set.

Source: docs

Set to 1 to force one model onto subagents, teammates, and workflow agents.

Source: docsSince v2.1.257

Sets the prompt cache TTL for subagents and background work. Accepts 5m or 1h.

Source: docsSince v2.1.242

Set to 1 to remove credentials from the environments of processes Claude Code starts.

Source: docs

Set to 1 with -p to wait for plugin installs to finish before the first query.

Source: docs

Timeout in milliseconds for synchronous plugin installation.

Source: docs

Set to 1 with -p to download your claude.ai skills and wait for them before the first query.

Source: docs

Sets the timeout for the mid-session skills resync in Agent SDK apps, in milliseconds. The default is 30000.

Source: docs

Sets how long the first query waits for the skill list, in milliseconds. The default is 5000.

Source: docs

Set to false to turn off syntax highlighting in diff output.

Source: docs

Shares a task list across sessions that use the same ID.

Source: docs

Override the temp directory used for internal temp files.

Source: docsSince v2.1.5

Set to any non-empty value, such as 1, to allow 24-bit truecolor output inside tmux.

Source: docs

On Linux and WSL, lists the process kinds left out of the tool memory cap, such as mcp or lsp.

Source: docsSince v2.1.246

On Linux and WSL, sets a memory cap, such as 4G, for shell commands run by tools.

Source: docsSince v2.1.233

Set to 1 to limit how large the transcript of a long -p or Agent SDK session grows.

Source: docsSince v2.1.287

Use Claude Platform on AWS.

Source: docs

Use Amazon Bedrock.

Source: docs

Use Microsoft Foundry.

Source: docs

Use the Amazon Bedrock Mantle endpoint.

Source: docs

Controls the PowerShell tool. The default depends on your platform and account.

Source: docs

Use Google Cloud's Agent Platform.

Source: docs

Sets how long Claude Code waits before cancelling a dialog sent to a remote client, in milliseconds.

Source: docs

Sets the free disk space, in GiB, below which Claude Code warns Claude.

Source: docsSince v2.1.304

On Linux and WSL, sets the free memory, in GiB, below which Claude Code warns Claude.

Source: docsSince v2.1.304

Sets how fast a session's WebSearch limit refills, in calls per hour.

Source: docsSince v2.1.290

Sets how long WebFetch caches each fetched URL, in milliseconds.

Source: docsSince v2.1.233

Sets the longest WebFetch waits for a page to download, in milliseconds.

Source: docsSince v2.1.268

With CLAUDE_AUTO_BACKGROUND_TASKS, sets the waits between reminders to check on background subagents.

Source: docsSince v2.1.283

Sets how many agents one workflow run executes at once, from 1 to 256.

Source: docsSince v2.1.269

Sets the longest a workflow agent waits for a sibling's first response before sending its own, in milliseconds.

Source: docsSince v2.1.229

Set to 1 to show a line above the prompt for each recent check by the You should know mod.

Source: docsSince v2.1.287

Set automatically in shell and hook processes to the effort level: low, medium, high, xhigh, or max.

Source: docs

1 forces the byte-level streaming idle watchdog on. 0 forces it off.

Source: docs

Set to 1 to turn on the byte-level idle watchdog for Bedrock event-stream responses.

Source: docs

0 forces the event-level streaming idle watchdog off. 1 forces it on.

Source: docs

Points to a shell script that Claude Code runs before each Bash command.

Source: docs

Set by Claude Code in each background session to that session's ~/.claude/jobs/<id> directory.

Source: docs

Set by Claude Code to its own process ID in the processes it starts.

Source: docsSince v2.1.214

Sets the prefix for auto-generated Remote Control session names.

Source: docs

Sets the deadline for the first response byte of a streaming request, in milliseconds.

Source: docsSince v2.1.242

Sets how long the streaming idle watchdogs wait before closing a stalled connection, in milliseconds.

Source: docsSince v2.1.84

Set to 1 in processes that Claude Code starts, such as shell commands and hooks.

Source: docs

Set to 1 to enable debug mode, equivalent to launching with --debug.

Source: docs

Set to 1 to turn off automatic compaction and idle compaction.

Source: docs

Set to 1 to disable automatic background updates.

Source: docs

Set to 1 to turn off all compaction, including the /compact command.

Source: docs

Set to 1 to disable cost warning messages.

Source: docs

Set to 1 to hide the /doctor command and its /checkup alias.

Source: docs

Set to any non-empty value, such as 1, to opt out of error reporting.

Source: docs

Set to 1 to hide the /usage-credits command.

Source: docs

Set to 1 to disable the /feedback command and Claude-drafted feedback.

Source: docs

Set to 1 to turn off feature-flag fetching.

Source: docs

Set to 1 to hide the /install-github-app command.

Source: docs

Set to 1 to disable installation warnings.

Source: docs

Set to 1 to stop sending the interleaved-thinking beta header.

Source: docsSince v1.0.1

Set to 1 to hide the /login command.

Source: docs

Set to 1 to hide the /logout command.

Source: docs

Set to 1 to turn off prompt caching for all models.

Source: docs

Set to 1 to disable prompt caching for Fable models.

Source: docs

Set to 1 to disable prompt caching for the default Haiku model, wherever it runs.

Source: docs

Set to 1 to disable prompt caching for the default Opus model.

Source: docs

Set to 1 to disable prompt caching for the default Sonnet model.

Source: docs

Set to 1 to block all updates, including claude update and claude install.

Source: docsSince v2.1.118

Set to 1 to hide the /upgrade command.

Source: docs

Set to 1 to opt out of telemetry. It works like DISABLE_TELEMETRY.

Source: docs

Set to 1, with BETA_TRACING_ENDPOINT, to turn on detailed beta tracing.

Source: docs

Set to false to stop Claude Code from fetching claude.ai MCP servers.

Source: docs

Set to 1 to ask for a 1-hour prompt cache TTL instead of the default 5 minutes.

Source: docsSince v2.1.108

Deprecated. ENABLE_PROMPT_CACHING_1H replaces it.

Source: docs

Controls MCP tool search. false loads all tools up front.

Source: docs

Set to 1 to update plugins automatically even when DISABLE_AUTOUPDATER is set.

Source: docsSince v2.1.2

Set to 1 to turn on clickable links when your terminal supports them but is not detected.

Source: docs

Set to 1 to force the 5-minute prompt cache TTL.

Source: docs

Sets the HTTP proxy server.

Source: docs

Sets the HTTPS proxy server.

Source: docs

Set to any non-empty value to turn on demo mode, which hides your email and organization name.

Source: docsSince v2.1.0

Sets the most tokens allowed in an MCP tool response. The default is 25000.

Source: docs

Sets how many tries Claude Code allows when a reply fails the --json-schema check with -p.

Source: docs

OAuth client secret for MCP servers that require pre-configured credentials.

Source: docs

Sets how long blocking MCP startup waits for the connection batch, in milliseconds. The default is 5000.

Source: docs

Controls whether startup waits for MCP servers to connect. It does not wait by default.

Source: docs

Turns the MCP discovery cache on or off.

Source: docs

Sets the oldest a discovery-cache entry can be, in seconds. The default is 14400.

Source: docs

Sets how many failed refreshes in a row discard a discovery-cache entry. The default is 1.

Source: docsSince v2.1.238

Sets how long a discovery-cache entry is used without refreshing, in seconds. The default is 900.

Source: docs

Sets a fixed port for the OAuth redirect callback when adding an MCP server.

Source: docs

On the v2 MCP client only, controls whether Claude Code probes servers for the newer MCP protocol revision.

Source: docsSince v2.1.221

Sets how many remote MCP servers connect in parallel at startup. The default is 20.

Source: docs

Picks which MCP client runtime to use: v1 or v2.

Source: docsSince v2.1.218

Sets how many local MCP servers connect in parallel at startup. The default is 3.

Source: docs

Sets the timeout for MCP server startup, in milliseconds. The default is 30000.

Source: docs

Sets the timeout for MCP tool runs, in milliseconds.

Source: docs

Lists domains and IPs that skip the proxy.

Source: docs

Sets the standard OpenTelemetry limit on attribute value length.

Source: docsSince v2.1.214

Set to 1 to include the model's reply text in assistant_response log events.

Source: docsSince v2.1.193

Set to 1 to add redacted managed settings and a digest to managed_settings_resolved log events.

Source: docsSince v2.1.274

Emits Messages API request and response JSON as log events.

Source: docsSince v2.1.111

Set to 1 to include tool content in the tool.output OpenTelemetry span event.

Source: docs

Set to 1 to include tool input arguments.

Source: docs

Set to 1 to include user prompt text in OpenTelemetry traces and logs.

Source: docs

Set to false to leave the account UUID out of metric and event attributes.

Source: docs

Set to true to add the session entrypoint to metric and event attributes.

Source: docs

Set to true to tag metrics and events with vcs.* attributes for the repository.

Source: docsSince v2.1.269

Set to false to leave OTEL_RESOURCE_ATTRIBUTES keys off metric labels.

Source: docs

Set to false to leave the session ID out of metric and event attributes.

Source: docs

Set to true to add the Claude Code version to metric and event attributes.

Source: docs

Overrides the character budget for skill metadata shown to the Skill tool.

Source: docs

Set to 0 to use your installed rg instead of the one bundled with Claude Code.

Source: docs

Override region for Claude 3.5 Haiku when using Google Cloud's Agent Platform.

Source: docs

Override region for Claude 3.5 Sonnet when using Google Cloud's Agent Platform.

Source: docs

Override region for Claude 3.7 Sonnet when using Google Cloud's Agent Platform.

Source: docs

Override region for Claude Opus 4 when using Google Cloud's Agent Platform.

Source: docs

Override region for Claude Sonnet 4 when using Google Cloud's Agent Platform.

Source: docs

Override region for Claude Opus 4.1 when using Google Cloud's Agent Platform.

Source: docs

Override region for Claude Opus 4.5 when using Google Cloud's Agent Platform.

Source: docs

Override region for Claude Sonnet 4.5 when using Google Cloud's Agent Platform.

Source: docs

Override region for Claude Opus 4.6 when using Google Cloud's Agent Platform.

Source: docs

Override region for Claude Sonnet 4.6 when using Google Cloud's Agent Platform.

Source: docs

Override region for Claude Opus 4.7 when using Google Cloud's Agent Platform.

Source: docs

Override region for Claude Opus 4.8 when using Google Cloud's Agent Platform.

Source: docs

Override region for Claude Opus 5.5 when using Google Cloud's Agent Platform.

Source: docs

Override region for Claude Sonnet 5.5 when using Google Cloud's Agent Platform.

Source: docs

Override region for Claude Opus 5 when using Google Cloud's Agent Platform.

Source: docs

Override region for Claude Sonnet 5 when using Google Cloud's Agent Platform.

Source: docs

Override region for Claude Fable 5 when using Google Cloud's Agent Platform.

Source: docs

Override region for Claude Fable 5.1 when using Google Cloud's Agent Platform.

Source: docs

Override region for Claude Fable 5.2 when using Google Cloud's Agent Platform.

Source: docs

Override region for Claude Haiku 4.5 when using Google Cloud's Agent Platform.

Source: docs

Override region for Claude Haiku 5.5 when using Google Cloud's Agent Platform.

Source: docs

Fires with --init-only, or with --init or --maintenance in -p mode.

Source: docs

Fires when a typed command expands into a prompt. It can block the expansion.

Source: docs

Fires when a tool call needs a permission decision.

Source: docsSince v2.0.45

Fires when auto mode denies a tool call.

Source: docsSince v2.1.89

Fires after a tool call fails.

Source: docs

Fires after a batch of parallel tool calls finishes, before the next model call.

Source: docs

Fires while assistant message text is displayed.

Source: docsSince v2.1.152

Fires when a subagent is spawned.

Source: docsSince v2.0.43

Fires when a task is created with TaskCreate.

Source: docsSince v2.1.84

Fires when a task is marked as completed.

Source: docs

Fires when the turn ends because of an API error.

Source: docsSince v2.1.78

Fires when an agent team teammate is about to go idle.

Source: docs

Fires when a CLAUDE.md or .claude/rules/*.md file is loaded into context.

Source: docsSince v2.1.69

Fires when a configuration file changes during a session.

Source: docsSince v2.1.49

Fires when the working directory changes, such as after a cd command.

Source: docs

Fires when a working directory is added mid-session with /add-dir.

Source: docsSince v2.1.219

Fires when a watched file changes on disk.

Source: docs

Fires when a worktree is being created. It replaces the default git behavior.

Source: docs

Fires when a worktree that a WorktreeCreate hook made is being removed.

Source: docs

Fires before context compaction.

Source: docs

Fires after context compaction finishes.

Source: docsSince v2.1.76

Fires before Claude Code applies a requested model switch. It can block the switch.

Source: docs

Fires after the session's model changes.

Source: docs

Fires when an MCP server asks for user input during a tool call.

Source: docs

Fires after a user answers an MCP elicitation, before the answer goes back to the server.

Source: docs

Adds a working directory for file access in the current session.

Usage: /add-dir <path>

Source: docs

Turns the advisor tool on or off. The advisor asks a second model for guidance.

Usage: /advisor [model|off]

Source: docs

Prints a reminder to ask Claude to create or manage subagents.

Source: docs

Loads the reference for what a published artifact can do.

Source: docs

Loads diagramming guidance for artifacts.

Source: docs

Lists the artifacts you own or that are shared with you.

Source: docs

Drafts autoMode.environment entries from your project and recent sessions.

Source: docsSince v2.1.304

Sets how full the context gets before Claude Code compacts it.

Usage: /autocompact [auto|<tokens>]

Source: docs

Starts a cloud session that watches your branch's PR and pushes fixes.

Usage: /autofix-pr [prompt]

Source: docs

Moves the current session to a background agent and frees the terminal.

Usage: /background [prompt]

Source: docs

Runs large changes across a codebase in parallel.

Usage: /batch <instruction>

Source: docs

Branches the current conversation at this point.

Usage: /branch [name]

Source: docs

Asks a side question without adding it to the conversation.

Usage: /btw [question]

Source: docs

Reports a bug or shares your conversation.

Usage: /bug [report]

Source: docs

Moves the session to a new working directory and keeps the conversation.

Usage: /cd <path>

Source: docsSince v2.1.169

Opens the Claude in Chrome settings.

Source: docs

Loads Claude API and Managed Agents reference material for your project's language.

Usage: /claude-api [migrate|upgrade|api-version|managed-agents-onboard|managed-agents-secure|prompt-audit|cost-optimize|build-eval|hillclimb|preserved-thinking-migration]

Source: docsSince v2.1.69

Has Claude do a task in your browser through Claude in Chrome.

Usage: /claude-in-chrome [task]

Source: docs

Reviews the current diff, or a PR, branch, or path, for correctness bugs.

Usage: /code-review [low|medium|high|xhigh|max|ultra] [--fix] [--comment] [--max-findings n|all|default] [pr#|branch|path]

Source: docs

Sets the prompt bar color for the session.

Usage: /color [color|default]

Source: docsSince v2.1.75

Opens the Settings interface.

Usage: /config [key=value ...]

Source: docs

Copies the last response to the clipboard.

Usage: /copy [N]

Source: docsSince v2.1.59

Alias for /usage.

Source: docs

Gives design guidance for charts, graphs, and dashboards.

Usage: /dataviz [request]

Source: docsSince v2.1.198

Turns on debug logging and reads the debug log to find problems.

Usage: /debug [description]

Source: docsSince v2.1.30

Searches the web on a question, checks sources, and writes a cited report.

Usage: /deep-research <question>

Source: docs

Drafts UI mockups, screen flows, landing pages, or posters as a Claude Design artifact.

Usage: /design [brief]

Source: docs

Gives Claude access to Claude Design. /design revoke takes it back.

Source: docs

Signs in for design-system access used by /design-sync.

Source: docs

Converts your repo's React design system and uploads it to Claude Design.

Usage: /design-sync [hint]

Source: docs

Continues the current session in the Claude Code Desktop app.

Source: docs

Shows the changes in your working tree, including Claude's edits.

Source: docs

Runs a setup checkup and proposes fixes for installation, settings, and CLAUDE.md problems.

Usage: /doctor [prompt-audit [path]]

Source: docs

Sets the effort level, from low to max, or auto.

Usage: /effort [level|auto|status|ultracode [on|off]]

Source: docsSince v2.1.76

Exits the CLI.

Source: docs

Exports the current conversation as plain text.

Usage: /export [filename]

Source: docs

Turns fast mode on or off.

Usage: /fast [on|off]

Source: docs

Sends product feedback about Claude Code.

Usage: /feedback [report]

Source: docs

Scans your transcripts for common read-only calls and adds an allowlist to project settings.

Source: docs

Toggles the focus view, which shows only your last prompt, a tool summary, and the final response.

Source: docs

Copies the conversation into a new background session and lets you keep working here.

Usage: /fork [prompt]

Source: docs

Sets a goal. Claude keeps working across turns until the condition is met.

Usage: /goal [condition|clear]

Source: docsSince v2.1.139

Writes a JavaScript heap snapshot and memory breakdown to a file.

Source: docs

Shows help and the available commands.

Source: docs

Shows your hook configurations.

Source: docs

Manages IDE integrations and shows status.

Source: docs

Brings config from OpenAI Codex, Gemini CLI, or Cursor into Claude Code.

Usage: /import [codex|gemini|cursor] [--dry-run] [--yes]

Source: docs

Writes an HTML report on your recent sessions.

Source: docs

Installs the Claude GitHub App for a repository.

Source: docs

Installs the Claude Slack app.

Source: docs

Opens your keyboard shortcuts file.

Source: docs

Lists the subagents, teammates, and other sessions Claude can message.

Source: docs

Signs in to your Anthropic account.

Source: docs

Signs out of your Anthropic account.

Source: docs

Runs a prompt again and again while the session stays open.

Usage: /loop [interval] [prompt]

Source: docsSince v2.1.71

Manages MCP server connections and OAuth sign-in.

Usage: /mcp [reconnect (<server>|all)|enable|disable [<server>|all]]

Source: docs

Shows a QR code to get the Claude mobile app.

Source: docs

Lists output styles or switches to one.

Usage: /output-style [style]

Source: docs

Shares a free week of Claude Code with friends.

Source: docs

Enters plan mode from the prompt.

Usage: /plan [description]

Source: docsSince v2.1.0

Manages Claude Code plugins.

Usage: /plugin [subcommand]

Source: docs

Loads the reference Claude uses to write a mod.

Source: docs

Teaches Claude Code features through short interactive lessons.

Source: docsSince v2.1.90

Shows and updates your privacy settings.

Source: docs

Opens Claude FM lo-fi radio in your browser.

Source: docs

Shows ways to keep working when a usage limit blocks a request.

Source: docsSince v2.1.284

Writes a one-line summary of the current session.

Source: docs

Shows the changelog in a version picker.

Source: docs

Reloads all active plugins without a restart.

Usage: /reload-plugins [--force]

Source: docs

Re-scans skill and command folders so new or changed skills load without a restart.

Source: docsSince v2.1.152

Makes this session available for Remote Control from claude.ai.

Source: docs

Sets the default cloud environment for cloud sessions you start from the CLI.

Source: docs

Renames the current session.

Usage: /rename [name]

Source: docs

Restarts Claude Code and resumes this session.

Source: docs

Alias of /code-review.

Usage: /review [low|medium|high|xhigh|max|ultra] [--fix] [--comment] [--max-findings n|all|default] [pr#|branch|path]

Source: docs

Rewinds the conversation, the code, or both to an earlier point.

Source: docs

Launches and runs your project's app to see a change working.

Source: docs

Writes a per-project skill that teaches /run and /verify to launch your app.

Source: docs

Turns sandbox mode on or off.

Source: docs

Creates, updates, lists, or runs cloud scheduled tasks.

Usage: /schedule [description]

Source: docs

Adjusts mouse wheel scroll speed.

Source: docsSince v2.1.139

Checks the changes on your branch for security problems.

Source: docs

Sets up Amazon Bedrock sign-in, region, and model pins with a wizard.

Source: docs

Sets up Google Cloud sign-in, project, region, and model pins with a wizard.

Source: docs

Reviews changed code for cleanup chances and applies the fixes.

Usage: /simplify [target]

Source: docs

Shows what each skill costs in context and how often it is used.

Source: docsSince v2.1.261

Lists available skills.

Source: docs

Makes a presentation as a Claude Slides artifact from your brief.

Usage: /slides [brief]

Source: docs

Alias for /usage.

Source: docs

Opens the Status tab with version, model, account, and connection.

Source: docs