Claude Code Cheat Sheet

Model	ID	Price / MTok	Role
Fable 5.1 NEW	`claude-fable-5-1`	$10 / $50	Most capable, never the default
Opus 5.5 NEW	`claude-opus-5-5`	$4 / $20	Default on every plan
Sonnet 5.5 NEW	`claude-sonnet-5-5`	$2 / $10	`sonnet` alias
Haiku 5.5 NEW	`claude-haiku-5-5`	$0.10 / $0.50	`haiku` alias, background tasks
Opus 5 / Sonnet 5	`claude-opus-5`	—	Fallback targets
Fable 5, Opus 4.8/4.7…	by ID	—	Still selectable

/ Slash commands (complete list)

Session & conversation

/clear [name] / /reset / /newNew conversation (names the old one) /compact [instructions]Summarize to free context /autocompact [size|auto]Auto-compact threshold (per model) NEW /resume [session] / /continueResume (incl. bg sessions) /rename [name]Rename the session /branch [name]Branch the conversation here /fork [prompt]Copy into a background session (own worktree) CHANGED /rewind / /checkpoint / /undoRewind code and/or conversation /btw [question]Side question outside context /recapOne-line session summary NEW /export [file]Export the conversation as text /copy [N]Copy the Nth response /cd <path>Change directory (config reloaded) /add-dir <path>Add a working directory /goal [condition|clear]Iterate until a condition holds /plan [description]Enter plan mode /exit / /quitExit (detaches a bg session)

Context, memory & diagnostics

/context [all]Context usage grid /memoryCLAUDE.md + auto memory /initGenerate a CLAUDE.md (CLAUDE_CODE_NEW_INIT=1) /doctor [prompt-audit] / /checkupFull checkup that proposes fixes CHANGED /skill-doctorContext cost and usage of each skill NEW /debug [description]Session debug logs /heapdumpJS heap snapshot NEW /insightsHTML usage report (+ auto mode advice) /statusVersion, model, account, connectivity /usage / /cost / /statsCosts, plan limits, cache, loops CHANGED /release-notesInteractive changelog

Model & settings

/model [model]Switch model (+ effort ←/→) /effort [level|ultracode on|off]Effort level / ultracode CHANGED /fast [on|off]Fast mode (Opus) /advisor [model|off]Enable the advisor NEW /config [key=value]Settings (or inline) CHANGED /output-style [style]Default, Proactive, Concise, Explanatory, Learning NEW /permissions / /allowed-toolsAllow/ask/deny rules + Auto mode tab CHANGED /sandboxToggle the Bash sandbox /hooksHooks grouped by event /keybindingsShortcuts file /statuslineConfigure the status line /themeThemes (auto, colorblind, custom) /tui default|fullscreenPick the renderer /focusFocus view (last prompt + summary) /colorPrompt bar color /scroll-speedWheel speed (fullscreen) /terminal-setupShift+Enter in VS Code, Cursor, Zed… /voice [hold|tap|off]Voice dictation /auto-mode-setupDraft autoMode.environment NEW /fewer-permission-promptsAllowlist from your transcripts NEW /update-configEdit settings.json in plain language NEW

Agents & parallelism

/agentsReminder: ask Claude or edit .claude/agents/ CHANGED /subtask <task>Forked subagent inheriting context NEW /background / /bg [prompt]Detach the session as a background agent NEW /stopStop the attached background session NEW /tasks / /bashesBackground work /list-agents / /peersReachable sessions and agents NEW /workflowsWatch / resume / save workflows /batch <task>Large change split into 5–30 units /loop [interval] [prompt] / /proactiveRepeat (or self-paced) /schedule / /routinesCloud routines

Review, quality & security

/code-review [effort|ultra] [--fix|--comment]Bug review in a background agent CHANGED /reviewAlias of /code-review CHANGED /simplify [target]4 agents: reuse, simplification /ultrareview [PR|branch]Multi-agent cloud review /security-reviewVulnerabilities on the branch /verifyRun the app and watch the result NEW /runLaunch and drive the app NEW /run-skill-generatorTeach /run how to build NEW /diffLive diff panel (fullscreen) CHANGED /autofix-pr [prompt]Cloud session that fixes the PR

Cloud, remote & integrations

/remote-control / /rcControl from claude.ai or the app /teleport / /tpPull a cloud session down /remote-envDefault cloud environment /web-setupConnect GitHub via gh /desktop / /appContinue in Claude Desktop /mobile / /ios / /androidMobile app QR code /chromeClaude in Chrome /ideIDE integrations /mcp [reconnect all|enable|disable]MCP servers CHANGED /install-github-appGitHub App (+ optional Actions) /install-slack-appSlack app /setup-bedrock / /setup-vertexCloud provider wizards NEW /import codex|gemini|cursorImport another agent's config NEW

Plugins, skills & artifacts

/plugin [install|list|enable…]Manage plugins and marketplaces /reload-plugins [--force]Reload without restart /skillsList (sort by tokens with t) /reload-skillsRe-scan skills /plugin-authoringReference to write a mod NEW /workflow-authoringWorkflow script reference NEW /artifactsList / attach / open NEW /artifact-capabilities / /artifact-diagrammingRuntime capabilities, SVG diagrams NEW /design [brief] / /design-sync / /design-loginClaude Design artboards NEW /slides [brief]Claude Slides deck NEW /datavizData visualization guide NEW /deep-research <question>Cited research workflow NEW /claude-api [migrate|…]Claude API / Managed Agents skill NEW /claude-in-chrome <task>Task in the browser NEW

Account & misc

/login / /logoutSign in (Console account supported) /upgradeUpgrade plan /usage-creditsUsage credits NEW /rate-limit-optionsOptions at a usage limit NEW /privacy-settingsPrivacy (Pro/Max) /feedback / /bug / /shareReport (Claude-drafted drafts) CHANGED /helpHelp /powerupInteractive lessons /team-onboardingTeam onboarding guide /passesShare a free week /radioClaude FM lo-fi radio NEW /stickersOrder stickers

Removed

/ultraplanRemoved (Aug): use plan mode or cloud REMOVED /vimRemoved: /config → Editor mode REMOVED /pr-commentsRemoved: just ask Claude REMOVED /extra-usageReplaced by /usage-credits REMOVED

Tip: about 120 built-in commands and skills (aliases share a line). Plugin skills are called /plugin:skill or just /skill; up to 5 can be stacked: /skill-a /skill-b do X.

💻 CLI: subcommands

Launch

claudeInteractive session claude "query"With an initial prompt claude -p "query"Non-interactive (SDK) cat f | claude -p "…"Piped content claude -c / -r <id|nom>Continue / resume

Background sessions & agent view

claude agents [--cwd|--json]Agent view: every session claude --bg "task"Start a background agent claude --bg --exec "cmd"Attachable PTY shell job claude attach <name>Attach (partial name OK) CHANGED claude logs <name>Recent output NEW claude stop / claude killStop a session claude respawn [--all]Restart with the conversation NEW claude rm <id> [--discard-unpushed|--force-remove-worktree]Remove from the list NEW claude daemon status|logs|run|stop [--keep-workers|--any]Session supervisor

Configuration & tooling

claude mcp add|add-json|add-from-claude-desktopAdd an MCP server claude mcp list|get|removeList / inspect / remove claude mcp serveExpose Claude Code as an MCP server claude mcp reset-project-choicesReset .mcp.json approvals claude mcp login|logout <srv> [--no-browser]MCP OAuth from the shell NEW claude plugin install|uninstall|enable|disable|updateLifecycle (-s, --json, -y, --marketplace, --keep-data, --prune) CHANGED claude plugin list [--available] [--data-size]Installed / available plugins claude plugin init [--with skills,hooks…]Scaffold a plugin NEW claude plugin details|configure|prune|tagDetails, options, dependency cleanup, version tag NEW claude plugin eval / claude plugin eval initEval a plugin vs baseline NEW claude plugin validate|testValidate / test claude plugin marketplace add / marketplace list / marketplace remove / marketplace updateManage marketplaces claude auto-mode defaults|config|resetClassifier rules CHANGED claude import <source> [--dry-run|--yes]Import Codex / Gemini / Cursor NEW claude doctorDiagnostics without a session claude purge [path] [--dry-run|-i|--interactive|--all|-y]Delete a project's local state NEW claude ultrareview <target> [--json|--post|--no-post|--timeout]Non-interactive ultrareview

Account, updates, infra

claude auth login|logout|statusAuthentication (--sso, --email, --console, --text) claude setup-tokenLong-lived OAuth token (CI) claude update / claude install [stable|latest|x.y.z]Update / install a version claude remote-controlRemote Control server claude self-hosted-runner [setup]Self-hosted runner NEW claude gateway --config gateway.yamlClaude apps gateway (admins) NEW

Install

curl -fsSL https://claude.ai/install.sh | bash      # macOS / Linux / WSL
irm https://claude.ai/install.ps1 | iex             # Windows PowerShell
brew install --cask claude-code                     # Homebrew (no auto-update)
winget install Anthropic.ClaudeCode                 # WinGet
sudo apt install claude-code                        # Anthropic apt repo

⚑ CLI: flags

Session

-c / --continueLatest conversation in this folder -r / --resumeResume by ID / name -n / --nameName the session --fork-sessionNew ID on resume --session-id <uuid>Fixed ID --from-pr <n|url>Sessions linked to a PR -w / --worktree [name|PR|MR]Isolated worktree (+ --tmux) CHANGED --bg / --backgroundStart as a background agent --add-dirExtra directories

Model & reasoning

--modelAlias or full ID --effortlow → max, or ultracode --fallback-model a,bFallback chain (max 3) --advisor <model>Server-side advisor NEW --autocompact <size>Auto-compact window NEW --betasBeta headers (API key)

Permissions & tools

--permission-modemanual, acceptEdits, plan, auto, dontAsk, bypassPermissions CHANGED --dangerously-skip-permissions= bypassPermissions --allow-dangerously-skip-permissionsAdds bypass to the Shift+Tab cycle --allowedTools / --allowed-toolsAllow rules --disallowedTools / --disallowed-toolsDeny rules (bare name = tool removed) --tools "Bash,Edit,Read"Built-in tool allowlist --permission-prompt-toolMCP tool answering prompts --permission-promptsWho answers in print mode NEW --restrictedNo command-running tools (eval harness) NEW --safe-modeAll customizations off --bareMinimal mode, nothing auto-discovered CHANGED

System prompt & config

--system-prompt / --system-prompt-fileReplace the system prompt --append-system-prompt / --append-system-prompt-fileAppend to the system prompt --append-subagent-system-prompt / --append-subagent-system-prompt-fileAppend to subagents NEW --system-prompt-snapshot offRebuild on every request NEW --exclude-dynamic-system-prompt-sectionsPer-user context out of system prompt --settings <json|file>Override settings --setting-sources user,project,localSources to load --agent / --agents <json>Main agent / ad-hoc subagents --mcp-config / --strict-mcp-configMCP from JSON / exclusively --plugin-dir / --plugin-urlLoad a plugin for the session CHANGED --disable-slash-commandsNo skills or commands

Print mode (-p) & SDK

-p / --printNon-interactive mode --output-format text|json|stream-jsonOutput format --input-format stream-jsonStreamed input --json-schemaValidated JSON output --max-turns / --max-budget-usdLimits (budget enforced on subagents) CHANGED --include-partial-messagesPartial events --include-hook-eventsHook events --forward-subagent-textSubagent text in the stream NEW --replay-user-messagesRe-emit user messages --prompt-suggestionsNext-prompt suggestion NEW --no-session-persistenceSave nothing --init / --init-only / --maintenanceSetup hooks

Cloud, remote, UI

--cloud "task" / --remoteCloud session (--remote deprecated) CHANGED --environment ccpool_… [--ref]Self-hosted environment NEW --teleportPull a cloud session --rc / --remote-controlRemotely controllable session --desktopOpen Claude Desktop here NEW --chrome / --no-chromeClaude in Chrome --ideAuto IDE connection --channelsListen to MCP channels --dangerously-load-development-channelsNon-allowlisted channels (dev) --teammate-modein-process, auto, tmux, iterm2 --ax-screen-readerScreen reader output NEW --debug[=filter] / --debug-file / --verboseDebug -v / --version / -h / --helpVersion / help --enable-auto-modeRemoved: --permission-mode auto REMOVED

🛡 Permissions, auto mode & sandbox

Manual→acceptEdits→plan→auto

autoA classifier judges each action. Default everywhere since v2.1.284 CHANGED manual (ex-default)Reads free, everything else asks CHANGED acceptEditsEdits and file commands accepted planRead-only exploration dontAskDenies anything not allowed bypassPermissionsNo prompts (not from project settings) CHANGED

Rule syntax

Bash(npm run *)Command pattern (Bash, Monitor) PowerShell(Get-ChildItem *)PowerShell command Read(~/secrets/**)Path (Read, Grep, Glob, LSP) Edit(/src/**)Path (Edit, Write, NotebookEdit) WebFetch(domain:x.com)Domain Skill(deploy *) / Agent(Explore)Skill / subagent type Agent(model:opus)By input parameter (* wildcard) NEW mcp__serveur__outilMCP tool

Auto mode

Server-side classifier by default, free on plans (CLAUDE_CODE_AUTO_MODE_SERVER)
Blocks git reset --hard, terraform destroy, cloud metadata access…
Asks before the first read outside working directories
autoMode.environment, hard_deny, classifyAllShell
Denial reason named (e.g. [Data Exfiltration])

Bash sandbox

macOS, Linux, WSL2 (not native Windows). Covers Bash, PowerShell, Monitor
sandbox.network.strictAllowlist, allowedDomains, tlsTerminate
sandbox.credentials: secret masking (mode: "mask", SigV4)
sandbox.filesystem.disabled: network only

🪝 Hooks

33 events

SessionStartStart / resume (startup, resume, clear, compact, fork) Setup--init, --init-only, --maintenance InstructionsLoadedCLAUDE.md or rule loaded UserPromptSubmitPrompt submitted UserPromptExpansionCommand expanding into a prompt NEW MessageDisplayDisplayed text (redact / reformat) PreToolUseBefore a tool (block, modify, defer) PermissionRequestPermission decision PermissionDeniedAuto mode denial (retry) PostToolUseAfter a successful tool PostToolUseFailureAfter a tool failure PostToolBatchAfter a batch of parallel calls NEW NotificationNotification (permission, idle…) SubagentStart / SubagentStopSubagent lifecycle TaskCreated / TaskCompletedTasks (TaskCreate) Stop / StopFailureTurn end / end on API error TeammateIdleAgent team teammate idle ConfigChangeConfig file changed CwdChanged / DirectoryAddedDirectory changed / added NEW FileChangedWatched file changed WorktreeCreate / WorktreeRemoveWorktrees (replaces git) PreCompact / PostCompactAround compaction PreModelSwitch / PostModelSwitchModel switch (blockable) NEW Elicitation / ElicitationResultMCP server input request SessionEndSession end

5 handler types

commandShell script (exec form args[], async, asyncRewake) httpPOST to a URL (headers, allowedEnvVars) mcp_toolCall an MCP tool (${tool_input.x}) promptA model judges ($ARGUMENTS) agentAn agent verifies with tools

Common fields

ifRule-syntax filter: Bash(git *) timeout / statusMessage / onceTimeout, spinner message, run once onFailure: "block"A failing hook blocks the action NEW shell: bash|powershellShell used

JSON output

continue · stopReason · suppressOutput · systemMessage · decision · reason · additionalContext · permissionDecision · permissionDecisionReason · updatedInput · updatedPermissions · updatedToolOutput · displayContent · continueOnBlock · retry · reloadSkills · sessionTitle · terminalSequence

Available variables

$CLAUDE_PROJECT_DIR · $CLAUDE_EFFORT · $CLAUDE_MODEL · ${CLAUDE_PLUGIN_ROOT} · ${CLAUDE_PLUGIN_DATA} · $CLAUDE_ENV_FILE (SessionStart : persist variables) · $CLAUDE_CODE_REMOTE · CLAUDECODE=1 in shells Claude starts

Tip: exit code 2 = block (stderr sent to Claude). In JSON: decision, additionalContext, hookSpecificOutput. Hooks live in settings.json, in skill or agent frontmatter, or in a plugin.

📝 Memory, skills & subagents

Instructions (CLAUDE.md)

Managed/etc/claude-code/CLAUDE.md (and OS equivalents) ~/.claude/CLAUDE.mdPersonal preferences, all projects ./CLAUDE.md / ./.claude/CLAUDE.mdProject, shared with the team ./AGENTS.mdRead when there is no CLAUDE.md NEW ./CLAUDE.local.mdPersonal per project (gitignore it) .claude/rules/*.mdRules scoped with paths: @path/fileImport inside a CLAUDE.md

Auto memory

Notes Claude writes in ~/.claude/projects/<project>/memory/, shared across worktrees, first 200 lines loaded
Subagent memory: memory: user|project|local

Skills: SKILL.md frontmatter

name · description · when_to_use · argument-hint · arguments · disable-model-invocation · user-invocable · allowed-tools · disallowed-tools · model · effort · context: fork · agent · background · hooks · paths · shell · metadata · license · compatibility

Substitutions: $ARGUMENTS, $name, inline shell !`cmd`. Skills nested in sub-folder .claude/skills load when you work there.

Subagents: agents/*.md frontmatter

name · description · tools · disallowedTools · model · permissionMode · maxTurns · skills · mcpServers · hooks · memory · background · omitClaudeMd NEW · effort · isolation: worktree · color · initialPrompt · autoCompactWindow NEW · experimental.cacheTtl

Built-in: general-purpose, Explore (inherits the main model, capped at Opus), Plan, fork.

Output styles

Default · Proactive · Concise NEW · Explanatory · Learning · custom in output-styles/*.md

🧩 Plugins, mods & MCP

What a plugin can ship

Skills · commands · agents · hooks · MCP servers · LSP servers · executables (bin/) · default settings · themes · output styles · channels · monitors · mods

/plugin install p@marketplaceInstall (refreshes the marketplace) /plugin marketplace add <src>GitHub, GitLab, URL, zip, npm, local command CHANGED claude-plugins-officialOfficial Anthropic marketplace security-guidanceSecurity review of each edit (rules: .claude/claude-security-guidance.md) claude-securityMulti-agent scan → CLAUDE-SECURITY-*/ NEW code intelligence (LSP)Type errors after each edit

Mods (October 2026)

A mod is a JS/TS plugin running inside Claude Code: panes, a band above the prompt, buttons, commands without a Claude turn, tool-call interception, model calls. CLI and Desktop Code tab. NEW

/plugin-authoringClaude writes the mod for you NEW cc-plugin-you-should-know@builtinBuilt-in mod: a side agent flags issues NEW $.ui.* / $.model.complete / $.agent.*Mods API NEW

A mod runs with your permissions: files, network, secrets, approvals. Install only from trusted sources.

MCP

claude mcp add --transport http|stdioAdd (local, project, user scopes) .mcp.jsonShared project servers /mcp reconnect allReconnect every failed server NEW alwaysLoad / tool searchTools loaded up front or deferred MCP_PROTOCOL_NEGOTIATION=legacyOpt out of protocol 2026-07-28 NEW

MCP calls over 2 min move to the background; descriptions capped at 4,096 characters; URL elicitation.

🔀 Parallelism & automation

Approach	When
Subagents	Side task that would flood context. Background by default, depth 3, 20 concurrent
Fork / /subtask	Subagent inheriting the whole conversation and cache
Agent view	`claude agents`: dispatch and monitor background sessions
Agent teams	Lead + teammates + shared tasks (`CLAUDE_CODE_EXPERIMENTAL_AGENT_TEAMS=1`)
Dynamic workflows	JS script orchestrating tens to hundreds of agents
Projects	One conversation launching parallel cloud threads (Pro/Max beta)
Cross-session	`SendMessage` / `ListAgents` between your sessions, `@name`
Worktrees	`-w`, `isolation: worktree`, `.worktreeinclude`

Automation

/goal <condition>Keep going until the condition (30-min check-ins) /loop 5m /testRepeat inside the session CronCreate / ScheduleWakeupSession-scoped schedules /schedule / RoutinesCloud agents: schedule, GitHub, API ChannelsAn MCP server pushes CI, chat, alerts MonitorFollow a process line by line (30-min cap) CHANGED claude-cli://open?repo=…&q=…Deep links into a session

🖥 Platforms & integrations

CLImacOS, Linux, Windows (PowerShell without Git Bash), WSL DesktopmacOS, Windows, Linux beta: built-in browser, iOS Simulator, background computer use, pop-out panes, auto-continue CHANGED Web / cloudclaude.ai/code: cloud sessions, routines, --cloud / --teleport Self-hostedCloud sessions on your infra (Team/Enterprise beta) NEW MobileiOS / Android: start a session on your machine (device card) CHANGED Remote ControlDrive a local session from web or phone VS CodeFocus view, agent map, bookmarks, session groups, timestamps CHANGED JetBrainsIntelliJ, PyCharm, WebStorm… ChromeClaude in Chrome (GA): tabs, forms, console CHANGED Computer usemacOS CLI: open apps, click, see the screen SlackClaude Tag (channels) + legacy Claude in Slack app CHANGED CI/CDGitHub Actions (@claude), GitLab CI/CD, GHES Code ReviewAutomatic PR review (~$15–25 per review) Agent SDKPython & TypeScript, claude -p

Providers

Anthropic API · Claude Platform on AWS · Amazon Bedrock (+ Mantle) · Google Cloud Agent Platform (formerly Vertex) · Microsoft Foundry · LLM gateways · self-hosted Claude apps gateway (SSO, policies, spend limits).

🔧 Built-in tools

Read / Write / EditFiles (allow_large on Read) Bash / PowerShellShell (embedded find/grep: bfs + ugrep) Glob / GrepAbsent by default outside Windows CHANGED NotebookEdit / LSPNotebooks, code intelligence WebFetch / WebSearchWeb (100 searches/hour refill) CHANGED Agent / SendMessage / ListAgentsSubagents and messaging SubagentHandbackA subagent's final report NEW WorkflowRun a dynamic workflow Skill / ToolSearchSkills, deferred tools AskUserQuestionClarifying multiple choice EnterPlanMode / ExitPlanModePlan mode EnterWorktree / ExitWorktreeWorktrees MonitorBackground process, line by line CronCreate / CronList / CronDeleteIn-session scheduling ScheduleWakeup / RemoteTriggerSelf-paced /loop, routines TaskCreate / Get / List / Update / StopTasks (off on recent models) CHANGED ArtifactPublish a page on claude.ai NEW PushNotification / SendUserFileNotify, send a file NEW SendFeedback / ReportFindingsFeedback draft, review findings NEW ListMcpResourcesTool / ReadMcpResourceToolMCP resources WaitForMcpServersWait for an MCP server NEW ShareOnboardingGuide / EndConversationOnboarding, end session (abuse) TaskOutput / TodoWriteDeprecated REMOVED

Tip: CLAUDE_CODE_ENABLE_TODO_TOOLS=1 brings task tools back; CLAUDE_CODE_DISABLE_WEB_FETCH turns WebFetch off.

📁 File structure map

project/
├── CLAUDE.md / AGENTS.md        # team instructions
├── CLAUDE.local.md              # personal, gitignored
├── .mcp.json                    # shared MCP servers
├── .worktreeinclude             # files copied into worktrees
└── .claude/
    ├── settings.json            # permissions, hooks, env
    ├── settings.local.json      # personal overrides
    ├── rules/*.md               # scoped rules (paths:)
    ├── skills/<nom>/SKILL.md
    ├── commands/*.md
    ├── agents/*.md
    ├── workflows/*.js           # become /commands
    ├── output-styles/*.md
    ├── agent-memory/<agent>/
    └── worktrees/

~/.claude/
├── CLAUDE.md · settings.json · keybindings.json
├── skills/ agents/ commands/ workflows/ output-styles/
├── themes/*.json                # custom themes
├── projects/<p>/memory/       # auto memory
├── plugins/                     # marketplaces + installed
└── daemon.log                   # background supervisor
~/.claude.json                   # state, OAuth, personal MCP

⚙ Configuration

Precedence (strongest first)

Managed (server, MDM, managed-settings.json) → CLI flags / --settings → .claude/settings.local.json → .claude/settings.json → ~/.claude/settings.json. Lists (permissions…) merge instead of overriding.

Settings that matter

model / effortLevel / fallbackModelModel, effort, fallback chain permissions.defaultModeStarting mode (auto by default) outputStyle / languageResponse style, language promptCacheTtl: "1h"1-hour cache (API key / cloud) NEW modelPicker / modelSettingsCustom picker, per-model settings NEW respondToBashCommandsReply to ! commands NEW autoContinueAtUsageLimitResume after a limit reset NEW attribution: falseNo commit / PR attribution CHANGED maxProseWidth / spellcheckProse width, spellcheck NEW syncClaudeAiSkills / syncClaudeAiPluginsSync from claude.ai NEW statusLine / subagentStatusLineStatus lines

Status line: JSON input

model · workspace · cost · context_window · rate_limits (five_hour, seven_day, spend_limit) · prompt_cache NEW · effort · thinking · fast_mode · output_style · vim · worktree · pr · session_name · transcript_path

Artifacts

Interactive pages published to claude.ai from the session, private by default, shareable (org or public link), updated in place, able to call your MCP connectors. All paid plans. /artifacts to find them. NEW

🏢 Admin & enterprise

Model control

availableModels (+ Match: "exact")Allowlist (exact versions) CHANGED deniedModelsBlock specific models NEW enforceAvailableModelsApply to the Default model too modelOverrides / modelPricingPer-version IDs, contracted rates CHANGED maxEffortLevelEffort cap NEW allowedProvidersProviders allowed per machine NEW

Versions & updates

requiredMinimumVersion / requiredMaximumVersionEnforced version bounds autoUpdatesChannelstable or latest channel

Policies

allowManagedPermissionRulesOnlyOnly managed rules count allowManagedHooksOnly / disableAllHooksLock down hooks managedMcpServers / allowedMcpServersMCP servers provided / allowed CHANGED strictKnownMarketplaces / allowedMarketplacesAllowed marketplaces disableAutoMode / disableWorkflowsTurn off auto mode, workflows forceLoginMethod / forceLoginOrgUUIDForce sign-in method policyHelper / processWrapperDynamic policy, required launcher NEW

Observability

OpenTelemetry (CLAUDE_CODE_ENABLE_TELEMETRY, OTEL_*), analytics dashboard, /usage per skill, agent, plugin and MCP. Project settings can no longer turn on telemetry export.

⚡ Quick reference

Getting started

claudeInteractive session

/initCreate CLAUDE.md

/import cursorBring your config

/doctorHealth check

/powerupLearn by doing

Context

/contextWhere context goes

/compact focus XCompact keeping X

/skill-doctorCostly skills

/btwQuestion without clutter

/clearStart fresh

Models

/model fableFable 5.1, big jobs

opus (default)Opus 5.5, medium

/effort xhighDeep reasoning

/effort ultracode onAuto workflows

/advisor opusA second opinion

Parallel

claude agentsDashboard

claude --bg "…"Delegate in background

/subtaskFork with context

/forkCopy the session

-w feat --tmuxWorktree + tmux

Quality

/code-review highBug hunt

/simplifyApplied cleanup

/verifyVerify for real

/ultrareviewCloud review

/security-reviewBranch security

Autonomy

/goal tests passIterate to green

/loop 5m /checkWatch

/scheduleCloud routine

/autofix-prSelf-fixing PR

/remote-controlFollow from phone

October 2026 defaults: Opus 5.5 is the default on Pro, Max, Team, Enterprise and the API · medium effort on the 5.5 family · auto mode by default · background subagents · fork mode on · Fable never the default (/model fable) · Glob/Grep replaced by embedded find/grep outside Windows.

📚 Appendix: full reference

To be truly exhaustive: the 244 keys of settings.json and the 391 environment variables documented as of October 9, 2026 (v2.1.296). Click to expand. Details for each: settings-reference and env-vars.

Every settings.json key (244)

Model and responses 23

advisorModelalwaysThinkingEnabledavailableModelsavailableModelsMatchdeniedModelseffortLevelenforceAvailableModelsfallbackModelfastModefastModePerSessionOptInlanguagemaxEffortLevelmodelmodelOverridesmodelPickermodelPricingmodelSettingsoutputStylepromptCacheTtlshowThinkingSummariessubagentPromptCacheTtlswitchModelsOnFlagultracode

Permission settings 15

allowManagedPermissionRulesOnlyautoModeautoMode.classifyAllShelldisableAutoModepermissionsuseAutoModeDuringPlanpermissions.allowpermissions.askpermissions.denypermissions.additionalDirectoriespermissions.blockReadsOutsideWorkingDirectoriespermissions.defaultModepermissions.disableBypassPermissionsModeskipAutoPermissionPromptskipDangerousModePermissionPrompt

Sandbox settings 38

sandboxsandbox.enabledsandbox.failIfUnavailablesandbox.autoAllowBashIfSandboxedsandbox.excludedCommandssandbox.allowUnsandboxedCommandssandbox.filesystemsandbox.filesystem.allowWritesandbox.filesystem.denyWritesandbox.filesystem.denyReadsandbox.filesystem.allowReadsandbox.filesystem.allowManagedReadPathsOnlysandbox.filesystem.disabledsandbox.ignoreViolationssandbox.enableWeakerNestedSandboxsandbox.enableWeakerNetworkIsolationsandbox.allowAppleEventssandbox.ripgrepsandbox.bwrapPathsandbox.socatPathsandbox.credentialssandbox.credentials.filessandbox.credentials.envVarssandbox.credentials.allowPlaintextInjectsandbox.credentials.awsPairssandbox.credentials.sigv4sandbox.networksandbox.network.allowUnixSocketssandbox.network.allowAllUnixSocketssandbox.network.allowLocalBindingsandbox.network.allowMachLookupsandbox.network.allowedDomainssandbox.network.deniedDomainssandbox.network.strictAllowlistsandbox.network.allowManagedDomainsOnlysandbox.network.httpProxyPortsandbox.network.socksProxyPortsandbox.network.tlsTerminate

Memory and context 13

autoCompactEnabledautoCompactWindowautoMemoryDirectoryautoMemoryEnabledbashOutputMaxCharsclaudeMdclaudeMdExcludesenvfileCheckpointingEnabledplansDirectoryskillListingBudgetFractionskillListingMaxDescCharstaskOutputMaxChars

Interface and terminal 39

askUserQuestionTimeoutautoContinueAtUsageLimitautoScrollEnabledaxScreenReaderbashEditDiffEnabledcompanyAnnouncementsdefaultShelldialogExpiryeditorModeemojiCompletionEnabledfileSuggestionfooterLinksRegexeskeybindingFlavormaxProseWidthprefersReducedMotionpromptSuggestionEnabledrespectGitignorerespondToBashCommandsshowClearContextOnPlanAcceptshowTurnDurationspellcheckspinnerTipsEnabledspinnerTipsOverridespinnerVerbsstatusLinesubagentStatusLinesyntaxHighlightingDisabledterminalProgressBarEnabledterminalTitleFromRenamethemetimeFormattimeZonetuiverboseviewModevimInsertModeRemapsvoicevoiceEnabledwheelScrollAccelerationEnabled

Git and attribution 7

attributionincludeCoAuthoredByincludeGitInstructionsprUrlTemplateattribution.commitattribution.prattribution.sessionUrl

Hooks and automation 9

allowedHttpHookUrlsallowManagedHooksOnlydisableAllHooksdisableWorkflowsenableWorkflowshookshttpHookAllowedEnvVarsworkflowKeywordTriggerEnabledworkflowSizeGuideline

Plugins and skills 22

disableBundledSkillsdisableSkillShellExecutionskillOverridessyncClaudeAiSkillssyncClaudeAiPluginsallowedChannelPluginsblockedMarketplaceschannelsEnableddisableCommandPluginSourcespluginSuggestionMarketplacespluginTrustMessagestrictKnownMarketplacesstrictPluginOnlyCustomizationstrictPluginOnlyCustomization.skillsstrictPluginOnlyCustomization.agentsstrictPluginOnlyCustomization.hooksstrictPluginOnlyCustomization.mcpenabledPluginsextraKnownMarketplacespluginConfigsprependPluginsappendPlugins

MCP 10

allowAllClaudeAiMcpsallowClaudeInChromeWithManagedMcpallowedMcpServersallowManagedMcpServersOnlydeniedMcpServersdisableClaudeAiConnectorsdisabledMcpjsonServersenableAllProjectMcpServersenabledMcpjsonServersmanagedMcpServers

Agents, sessions, and worktrees 12

agentcrossSessionInbounddisableAgentViewisolatePeerMachinesprocessWrapperteammateModeworktreeworktree.baseRefworktree.symlinkDirectoriesworktree.sparsePathsworktree.bgIsolationworktree.location

Remote, desktop, and notifications 13

agentPushNotifEnabledawaySummaryEnableddisableArtifactdisableDeepLinkRegistrationdisableDesktopLocalSessionsdisableRemoteControlenableArtifactinputNeededNotifEnabledpreferredNotifChannelremote.defaultEnvironmentIdremoteControlAtStartupsshConfigssshHostAllowlist

Authentication and providers 10

allowedProvidersapiKeyHelperawsAuthRefreshawsCredentialExportforceLoginMethodforceLoginGatewayUrlforceLoginOrgUUIDgatewayInternalNetworksgcpAuthRefreshotelHeadersHelper

Updates and versioning 4

autoUpdatesChannelminimumVersionrequiredMaximumVersionrequiredMinimumVersion

Tools 3

browserExternalPageToolsdisableBrowserExternalNavigationdisableMobileSimulatorTools

Privacy and telemetry 5

cleanupPeriodDaysdesktopSessionCleanupPeriodDaysfeedbackDraftsfeedbackSurveyRateskipWebFetchPreflight

Enterprise and managed settings 9

disableSideloadFlagsforceRemoteSettingsRefreshmanagedSourcesBehaviorparentSettingsBehaviorpolicyHelperpolicyHelper.pathpolicyHelper.timeoutMspolicyHelper.refreshIntervalMswslInheritsWindowsSettings

Global config settings 12

autoConnectIdeautoInstallIdeExtensionclaudeInChromeDefaultEnabledcopyFullResponsecopyOnSelectdefaultToAgentsViewdiffToolexternalEditorContextleftArrowOpensAgentspermissionExplainerEnabledprStatusFooterEnabledteammateDefaultModel

Every environment variable (391)

ANTHROPIC_* 46

ANTHROPIC_API_KEYANTHROPIC_AUTH_TOKENANTHROPIC_AWS_API_KEYANTHROPIC_AWS_BASE_URLANTHROPIC_AWS_WORKSPACE_IDANTHROPIC_BASE_URLANTHROPIC_BEDROCK_BASE_URLANTHROPIC_BEDROCK_MANTLE_BASE_URLANTHROPIC_BEDROCK_REGION_PREFIXANTHROPIC_BEDROCK_SERVICE_TIERANTHROPIC_BETASANTHROPIC_CUSTOM_HEADERSANTHROPIC_CUSTOM_MODEL_OPTIONANTHROPIC_CUSTOM_MODEL_OPTION_DESCRIPTIONANTHROPIC_CUSTOM_MODEL_OPTION_NAMEANTHROPIC_CUSTOM_MODEL_OPTION_SUPPORTED_CAPABILITIESANTHROPIC_DEFAULT_FABLE_MODELANTHROPIC_DEFAULT_FABLE_MODEL_DESCRIPTIONANTHROPIC_DEFAULT_FABLE_MODEL_NAMEANTHROPIC_DEFAULT_FABLE_MODEL_SUPPORTED_CAPABILITIESANTHROPIC_DEFAULT_HAIKU_MODELANTHROPIC_DEFAULT_HAIKU_MODEL_DESCRIPTIONANTHROPIC_DEFAULT_HAIKU_MODEL_NAMEANTHROPIC_DEFAULT_HAIKU_MODEL_SUPPORTED_CAPABILITIESANTHROPIC_DEFAULT_MODELANTHROPIC_DEFAULT_OPUS_MODELANTHROPIC_DEFAULT_OPUS_MODEL_DESCRIPTIONANTHROPIC_DEFAULT_OPUS_MODEL_NAMEANTHROPIC_DEFAULT_OPUS_MODEL_SUPPORTED_CAPABILITIESANTHROPIC_DEFAULT_SONNET_MODELANTHROPIC_DEFAULT_SONNET_MODEL_DESCRIPTIONANTHROPIC_DEFAULT_SONNET_MODEL_NAMEANTHROPIC_DEFAULT_SONNET_MODEL_SUPPORTED_CAPABILITIESANTHROPIC_FEDERATION_RULE_IDANTHROPIC_FOUNDRY_API_KEYANTHROPIC_FOUNDRY_AUTH_TOKENANTHROPIC_FOUNDRY_BASE_URLANTHROPIC_FOUNDRY_RESOURCEANTHROPIC_MODELANTHROPIC_ORGANIZATION_IDANTHROPIC_PROFILEANTHROPIC_SMALL_FAST_MODELANTHROPIC_SMALL_FAST_MODEL_AWS_REGIONANTHROPIC_VERTEX_BASE_URLANTHROPIC_VERTEX_PROJECT_IDANTHROPIC_WORKSPACE_ID

CLAUDE_CODE_* 220

CLAUDE_CODE_ACCESSIBILITYCLAUDE_CODE_ADDITIONAL_DIRECTORIES_CLAUDE_MDCLAUDE_CODE_ALT_SCREEN_FULL_REPAINTCLAUDE_CODE_ALWAYS_ENABLE_EFFORTCLAUDE_CODE_API_KEY_HELPER_TTL_MSCLAUDE_CODE_ARTIFACT_AUTO_OPENCLAUDE_CODE_ARTIFACT_COMMENTSCLAUDE_CODE_ARTIFACT_COMMENTS_AUTOREACTCLAUDE_CODE_ATTRIBUTION_HEADERCLAUDE_CODE_AUTO_BACKGROUND_WORKER_CHECKIN_SECONDSCLAUDE_CODE_AUTO_COMPACT_WINDOWCLAUDE_CODE_AUTO_CONNECT_IDECLAUDE_CODE_AUTO_MODE_SERVERCLAUDE_CODE_AWS_CHAIN_RESOLVE_TIMEOUT_MSCLAUDE_CODE_BASH_EDIT_DIFFCLAUDE_CODE_BG_TASKS_REPORT_RUNNINGCLAUDE_CODE_BRIDGE_SESSION_IDCLAUDE_CODE_BS_AS_CTRL_BACKSPACECLAUDE_CODE_CERT_STORECLAUDE_CODE_CHILD_SESSIONCLAUDE_CODE_CLIENT_CERTCLAUDE_CODE_CLIENT_KEYCLAUDE_CODE_CLIENT_KEY_PASSPHRASECLAUDE_CODE_CONNECT_TIMEOUT_MSCLAUDE_CODE_DEBUG_LOGS_DIRCLAUDE_CODE_DEBUG_LOG_LEVELCLAUDE_CODE_DISABLE_1M_CONTEXTCLAUDE_CODE_DISABLE_ADAPTIVE_THINKINGCLAUDE_CODE_DISABLE_ADMIN_ENV_UNIONCLAUDE_CODE_DISABLE_ADVISOR_TOOLCLAUDE_CODE_DISABLE_AGENT_VIEWCLAUDE_CODE_DISABLE_ALTERNATE_SCREENCLAUDE_CODE_DISABLE_ARTIFACTCLAUDE_CODE_DISABLE_ATTACHMENTSCLAUDE_CODE_DISABLE_AUTH_REFRESH_LOCKCLAUDE_CODE_DISABLE_AUTO_MEMORYCLAUDE_CODE_DISABLE_BACKGROUND_TASKSCLAUDE_CODE_DISABLE_BEDROCK_CONTENT_TYPE_DEFAULTCLAUDE_CODE_DISABLE_BEDROCK_CONTENT_TYPE_GUARDCLAUDE_CODE_DISABLE_BG_EXIT_HANDOFFCLAUDE_CODE_DISABLE_BG_SHELL_PRESSURE_REAPCLAUDE_CODE_DISABLE_BUNDLED_SKILLSCLAUDE_CODE_DISABLE_CFC_PROMPTCLAUDE_CODE_DISABLE_CLAUDE_MDSCLAUDE_CODE_DISABLE_CRONCLAUDE_CODE_DISABLE_DANGEROUS_RM_TIMEOUTCLAUDE_CODE_DISABLE_EXPERIMENTAL_BETASCLAUDE_CODE_DISABLE_EXPLORE_PLAN_AGENTSCLAUDE_CODE_DISABLE_FAST_MODECLAUDE_CODE_DISABLE_FEEDBACK_SURVEYCLAUDE_CODE_DISABLE_FILE_CHECKPOINTINGCLAUDE_CODE_DISABLE_GIT_INSTRUCTIONSCLAUDE_CODE_DISABLE_INLINE_SHELL_RM_PROMPTCLAUDE_CODE_DISABLE_LEGACY_MODEL_REMAPCLAUDE_CODE_DISABLE_MODEL_ACCESS_FALLBACKCLAUDE_CODE_DISABLE_MOUSECLAUDE_CODE_DISABLE_MOUSE_CLICKSCLAUDE_CODE_DISABLE_MTLS_RELOAD_ON_STALE_CONNECTIONCLAUDE_CODE_DISABLE_NONESSENTIAL_TRAFFICCLAUDE_CODE_DISABLE_NONSTREAMING_FALLBACKCLAUDE_CODE_DISABLE_NOTIFICATION_PRESENCE_CHECKCLAUDE_CODE_DISABLE_OFFICIAL_MARKETPLACE_AUTOINSTALLCLAUDE_CODE_DISABLE_PERMISSION_PROMPT_NOTIFY_HOOKSCLAUDE_CODE_DISABLE_POLICY_SKILLSCLAUDE_CODE_DISABLE_POWERSHELL_CMD_RM_DENYCLAUDE_CODE_DISABLE_REFUSAL_FALLBACKCLAUDE_CODE_DISABLE_STRUCTURED_OUTPUTSCLAUDE_CODE_DISABLE_SUBSTITUTION_RM_PROMPTCLAUDE_CODE_DISABLE_TERMINAL_TITLECLAUDE_CODE_DISABLE_THINKINGCLAUDE_CODE_DISABLE_UNKNOWN_MODEL_WINDOW_ENFORCEMENTCLAUDE_CODE_DISABLE_VIRTUAL_SCROLLCLAUDE_CODE_DISABLE_WEB_FETCHCLAUDE_CODE_DISABLE_WINDOWS_SHELL_LAUNCHERCLAUDE_CODE_DISABLE_WORKFLOWSCLAUDE_CODE_EFFORT_LEVELCLAUDE_CODE_EMIT_SESSION_STATE_EVENTSCLAUDE_CODE_ENABLE_AUTO_MODECLAUDE_CODE_ENABLE_AWAY_SUMMARYCLAUDE_CODE_ENABLE_BACKGROUND_PLUGIN_REFRESHCLAUDE_CODE_ENABLE_CFCCLAUDE_CODE_ENABLE_FEEDBACK_SURVEY_FOR_OTELCLAUDE_CODE_ENABLE_FINE_GRAINED_TOOL_STREAMINGCLAUDE_CODE_ENABLE_GATEWAY_MODEL_DISCOVERYCLAUDE_CODE_ENABLE_OPUS_4_7_FAST_MODECLAUDE_CODE_ENABLE_PROMPT_SUGGESTIONCLAUDE_CODE_ENABLE_TASKSCLAUDE_CODE_ENABLE_TELEMETRYCLAUDE_CODE_ENABLE_TODO_TOOLSCLAUDE_CODE_EXIT_AFTER_STOP_DELAYCLAUDE_CODE_EXPERIMENTAL_AGENT_TEAMSCLAUDE_CODE_EXTRA_BODYCLAUDE_CODE_FILE_READ_MAX_OUTPUT_TOKENSCLAUDE_CODE_FORCE_SESSION_PERSISTENCECLAUDE_CODE_FORCE_STRIKETHROUGHCLAUDE_CODE_FORCE_SYNC_OUTPUTCLAUDE_CODE_FORCE_TERMINAL_IMAGESCLAUDE_CODE_FORK_SUBAGENTCLAUDE_CODE_FORWARD_SUBAGENT_TEXTCLAUDE_CODE_GATEWAY_HINT_HEADERSCLAUDE_CODE_GATEWAY_MODEL_DISCOVERY_TIMEOUT_MSCLAUDE_CODE_GIT_BASH_PATHCLAUDE_CODE_GLOB_HIDDENCLAUDE_CODE_GLOB_NO_IGNORECLAUDE_CODE_GLOB_TIMEOUT_SECONDSCLAUDE_CODE_GOAL_CHECKIN_MINUTESCLAUDE_CODE_GZIP_REQUEST_BODIESCLAUDE_CODE_HIDE_CWDCLAUDE_CODE_IDE_HOST_OVERRIDECLAUDE_CODE_IDE_SKIP_AUTO_INSTALLCLAUDE_CODE_IDE_SKIP_VALID_CHECKCLAUDE_CODE_MAX_CONCURRENT_SUBAGENTSCLAUDE_CODE_MAX_CONTEXT_TOKENSCLAUDE_CODE_MAX_MCP_DESCRIPTION_LENGTHCLAUDE_CODE_MAX_OUTPUT_TOKENSCLAUDE_CODE_MAX_RETRIESCLAUDE_CODE_MAX_SUBAGENTS_PER_SESSIONCLAUDE_CODE_MAX_SUBAGENT_SPAWN_DEPTHCLAUDE_CODE_MAX_TOOL_USE_CONCURRENCYCLAUDE_CODE_MAX_TURNSCLAUDE_CODE_MAX_WEB_SEARCHES_PER_SESSIONCLAUDE_CODE_MCP_ALLOWLIST_ENVCLAUDE_CODE_MCP_AUTO_BACKGROUND_MSCLAUDE_CODE_MCP_STARTUP_WAIT_MSCLAUDE_CODE_MCP_TOOL_IDLE_TIMEOUTCLAUDE_CODE_MESSAGING_SOCKETCLAUDE_CODE_MESSAGING_TOKENCLAUDE_CODE_NATIVE_CURSORCLAUDE_CODE_NEW_INITCLAUDE_CODE_NONBLOCKING_STDOUTCLAUDE_CODE_NONSTREAMING_TIMEOUT_RETRIESCLAUDE_CODE_NO_FLICKERCLAUDE_CODE_OAUTH_REFRESH_TOKENCLAUDE_CODE_OAUTH_SCOPESCLAUDE_CODE_OAUTH_TOKENCLAUDE_CODE_OPUS_4_6_FAST_MODE_OVERRIDECLAUDE_CODE_OTEL_CONTENT_MAX_LENGTHCLAUDE_CODE_OTEL_DIAG_STDERRCLAUDE_CODE_OTEL_FLUSH_TIMEOUT_MSCLAUDE_CODE_OTEL_HEADERS_HELPER_DEBOUNCE_MSCLAUDE_CODE_OTEL_SHUTDOWN_TIMEOUT_MSCLAUDE_CODE_OVERLOADED_RETRY_BASE_DELAY_MSCLAUDE_CODE_PACKAGE_MANAGER_AUTO_UPDATECLAUDE_CODE_PERFORCE_MODECLAUDE_CODE_PLUGIN_CACHE_DIRCLAUDE_CODE_PLUGIN_DIRSCLAUDE_CODE_PLUGIN_DIR_WATCHCLAUDE_CODE_PLUGIN_GIT_TIMEOUT_MSCLAUDE_CODE_PLUGIN_KEEP_MARKETPLACE_ON_FAILURECLAUDE_CODE_PLUGIN_PREFER_HTTPSCLAUDE_CODE_PLUGIN_SEED_DIRCLAUDE_CODE_POWERSHELL_RESPECT_EXECUTION_POLICYCLAUDE_CODE_PRINT_BG_WAIT_CEILING_MSCLAUDE_CODE_PROCESS_WRAPPERCLAUDE_CODE_PROJECT_DIR_NAMECLAUDE_CODE_PROMPT_CACHE_TTLCLAUDE_CODE_PROPAGATE_TRACEPARENTCLAUDE_CODE_PROVIDER_MANAGED_BY_HOSTCLAUDE_CODE_PROXY_RESOLVES_HOSTSCLAUDE_CODE_REMOTECLAUDE_CODE_REMOTE_SESSION_IDCLAUDE_CODE_RESTRICTEDCLAUDE_CODE_RESUME_INTERRUPTED_TURNCLAUDE_CODE_RESUME_INTERRUPTED_TURN_MAX_AGE_MSCLAUDE_CODE_RESUME_PROMPTCLAUDE_CODE_RETRY_WATCHDOGCLAUDE_CODE_RETRY_WATCHDOG_MAX_WAIT_MSCLAUDE_CODE_SAFE_MODECLAUDE_CODE_SCRIPT_CAPSCLAUDE_CODE_SCROLL_SPEEDCLAUDE_CODE_SEND_FEEDBACKCLAUDE_CODE_SESSIONEND_HOOKS_TIMEOUT_MSCLAUDE_CODE_SESSION_IDCLAUDE_CODE_SHELLCLAUDE_CODE_SHELL_PREFIXCLAUDE_CODE_SIMPLECLAUDE_CODE_SIMPLE_SYSTEM_PROMPTCLAUDE_CODE_SKIP_ANTHROPIC_AWS_AUTHCLAUDE_CODE_SKIP_AWS_CRED_CACHECLAUDE_CODE_SKIP_BEDROCK_AUTHCLAUDE_CODE_SKIP_FAST_MODE_NETWORK_ERRORSCLAUDE_CODE_SKIP_FAST_MODE_ORG_CHECKCLAUDE_CODE_SKIP_FOUNDRY_AUTHCLAUDE_CODE_SKIP_MANTLE_AUTHCLAUDE_CODE_SKIP_MODEL_ACCESS_MEMORYCLAUDE_CODE_SKIP_PROMPT_HISTORYCLAUDE_CODE_SKIP_VERTEX_AUTHCLAUDE_CODE_STARTUP_FAILURE_RESULTSCLAUDE_CODE_STOP_HOOK_BLOCK_CAPCLAUDE_CODE_SUBAGENT_MODELCLAUDE_CODE_SUBAGENT_MODEL_FORCECLAUDE_CODE_SUBAGENT_PROMPT_CACHE_TTLCLAUDE_CODE_SUBPROCESS_ENV_SCRUBCLAUDE_CODE_SYNC_PLUGIN_INSTALLCLAUDE_CODE_SYNC_PLUGIN_INSTALL_TIMEOUT_MSCLAUDE_CODE_SYNC_SKILLSCLAUDE_CODE_SYNC_SKILLS_INSTALL_TIMEOUT_MSCLAUDE_CODE_SYNC_SKILLS_WAIT_TIMEOUT_MSCLAUDE_CODE_SYNTAX_HIGHLIGHTCLAUDE_CODE_TASK_LIST_IDCLAUDE_CODE_TEAM_TEARDOWN_PARK_TIMEOUT_MSCLAUDE_CODE_TMPDIRCLAUDE_CODE_TMUX_TRUECOLORCLAUDE_CODE_TOOL_MEMORY_CGROUP_EXCLUDECLAUDE_CODE_TOOL_MEMORY_LIMITCLAUDE_CODE_TRANSCRIPT_LOCAL_GCCLAUDE_CODE_USER_DIALOG_TIMEOUT_MSCLAUDE_CODE_USE_ANTHROPIC_AWSCLAUDE_CODE_USE_BEDROCKCLAUDE_CODE_USE_FOUNDRYCLAUDE_CODE_USE_MANTLECLAUDE_CODE_USE_NATIVE_FILE_SEARCHCLAUDE_CODE_USE_POWERSHELL_TOOLCLAUDE_CODE_USE_VERTEXCLAUDE_CODE_WEBFETCH_CACHE_TTL_MSCLAUDE_CODE_WEBFETCH_DEADLINE_MSCLAUDE_CODE_WEB_SEARCH_REFILLS_PER_HOURCLAUDE_CODE_WORKER_CHECKIN_SCHEDULECLAUDE_CODE_WORKFLOW_MAX_CONCURRENT_AGENTSCLAUDE_CODE_WORKFLOW_PREFIX_STAGGER_MS

CLAUDE_* (other) 26

CLAUDE_AFK_COUNTDOWN_MSCLAUDE_AFK_TIMEOUT_MSCLAUDE_AGENT_SDK_DISABLE_BUILTIN_AGENTSCLAUDE_AGENT_SDK_MCP_NO_PREFIXCLAUDE_ASYNC_AGENT_STALL_TIMEOUT_MSCLAUDE_AUTOCOMPACT_PCT_OVERRIDECLAUDE_AUTO_BACKGROUND_TASKSCLAUDE_AX_PREPARK_MSCLAUDE_AX_SCREEN_READERCLAUDE_AX_STARTUP_QUIET_MSCLAUDE_BASH_MAINTAIN_PROJECT_WORKING_DIRCLAUDE_BYTE_STREAM_IDLE_TIMEOUT_MSCLAUDE_CLIENT_PRESENCE_FILECLAUDE_CONFIG_DIRCLAUDE_DISABLE_ADOPTCLAUDE_EFFORTCLAUDE_ENABLE_BYTE_WATCHDOGCLAUDE_ENABLE_BYTE_WATCHDOG_BEDROCKCLAUDE_ENABLE_STREAM_WATCHDOGCLAUDE_ENV_FILECLAUDE_JOB_DIRCLAUDE_PIDCLAUDE_REMOTE_CONTROL_SESSION_NAME_PREFIXCLAUDE_STREAM_FIRST_BYTE_TIMEOUT_MSCLAUDE_STREAM_IDLE_TIMEOUT_MSCLAUDE_SUBAGENT_BG_SHELL_MAX_MS

OTEL_* 13

OTEL_ATTRIBUTE_VALUE_LENGTH_LIMITOTEL_LOG_ASSISTANT_RESPONSESOTEL_LOG_MANAGED_SETTINGSOTEL_LOG_RAW_API_BODIESOTEL_LOG_TOOL_CONTENTOTEL_LOG_TOOL_DETAILSOTEL_LOG_USER_PROMPTSOTEL_METRICS_INCLUDE_ACCOUNT_UUIDOTEL_METRICS_INCLUDE_ENTRYPOINTOTEL_METRICS_INCLUDE_REPOSITORYOTEL_METRICS_INCLUDE_RESOURCE_ATTRIBUTESOTEL_METRICS_INCLUDE_SESSION_IDOTEL_METRICS_INCLUDE_VERSION

DISABLE_* / ENABLE_* 27

DISABLE_AUTOUPDATERDISABLE_AUTO_COMPACTDISABLE_COMPACTDISABLE_COST_WARNINGSDISABLE_DOCTOR_COMMANDDISABLE_ERROR_REPORTINGDISABLE_EXTRA_USAGE_COMMANDDISABLE_FEEDBACK_COMMANDDISABLE_GROWTHBOOKDISABLE_INSTALLATION_CHECKSDISABLE_INSTALL_GITHUB_APP_COMMANDDISABLE_INTERLEAVED_THINKINGDISABLE_LOGIN_COMMANDDISABLE_LOGOUT_COMMANDDISABLE_PROMPT_CACHINGDISABLE_PROMPT_CACHING_FABLEDISABLE_PROMPT_CACHING_HAIKUDISABLE_PROMPT_CACHING_OPUSDISABLE_PROMPT_CACHING_SONNETDISABLE_TELEMETRYDISABLE_UPDATESDISABLE_UPGRADE_COMMANDENABLE_BETA_TRACING_DETAILEDENABLE_CLAUDEAI_MCP_SERVERSENABLE_PROMPT_CACHING_1HENABLE_PROMPT_CACHING_1H_BEDROCKENABLE_TOOL_SEARCH

MCP_* / MAX_* 17

MAX_MCP_OUTPUT_TOKENSMAX_STRUCTURED_OUTPUT_RETRIESMAX_THINKING_TOKENSMCP_CLIENT_SECRETMCP_CONNECTION_NONBLOCKINGMCP_CONNECT_TIMEOUT_MSMCP_DISCOVERY_CACHEMCP_DISCOVERY_CACHE_MAX_STALE_SMCP_DISCOVERY_CACHE_STRIKESMCP_DISCOVERY_CACHE_TTL_SMCP_OAUTH_CALLBACK_PORTMCP_PROTOCOL_NEGOTIATIONMCP_REMOTE_SERVER_CONNECTION_BATCH_SIZEMCP_SDK_GENERATIONMCP_SERVER_CONNECTION_BATCH_SIZEMCP_TIMEOUTMCP_TOOL_TIMEOUT

Other 42

API_FORCE_IDLE_TIMEOUTAPI_TIMEOUT_MSAWS_BEARER_TOKEN_BEDROCKBASH_DEFAULT_TIMEOUT_MSBASH_MAX_OUTPUT_LENGTHBASH_MAX_TIMEOUT_MSBETA_TRACING_ENDPOINTCCR_FORCE_BUNDLECLAUDECODEDEBUGDO_NOT_TRACKFALLBACK_FOR_ALL_PRIMARY_MODELSFORCE_AUTOUPDATE_PLUGINSFORCE_HYPERLINKFORCE_PROMPT_CACHING_5MHTTPS_PROXYHTTP_PROXYIS_DEMONO_PROXYSLASH_COMMAND_TOOL_CHAR_BUDGETTASK_MAX_OUTPUT_LENGTHUSE_BUILTIN_RIPGREPVERTEX_REGION_CLAUDE_3_5_HAIKUVERTEX_REGION_CLAUDE_3_5_SONNETVERTEX_REGION_CLAUDE_3_7_SONNETVERTEX_REGION_CLAUDE_4_0_OPUSVERTEX_REGION_CLAUDE_4_0_SONNETVERTEX_REGION_CLAUDE_4_1_OPUSVERTEX_REGION_CLAUDE_4_5_OPUSVERTEX_REGION_CLAUDE_4_5_SONNETVERTEX_REGION_CLAUDE_4_6_OPUSVERTEX_REGION_CLAUDE_4_6_SONNETVERTEX_REGION_CLAUDE_4_7_OPUSVERTEX_REGION_CLAUDE_4_8_OPUSVERTEX_REGION_CLAUDE_5_5_OPUSVERTEX_REGION_CLAUDE_5_5_SONNETVERTEX_REGION_CLAUDE_5_OPUSVERTEX_REGION_CLAUDE_5_SONNETVERTEX_REGION_CLAUDE_FABLE_5VERTEX_REGION_CLAUDE_FABLE_5_1VERTEX_REGION_CLAUDE_HAIKU_4_5VERTEX_REGION_CLAUDE_HAIKU_5_5

Every keybindings.json action (137)

Format: context:action, bound to a key in ~/.claude/keybindings.json.

app 11

app:interruptapp:exitapp:redrawapp:toggleTodosapp:toggleTranscriptapp:toggleReplTabapp:cycleDiffBaseapp:diffFileListUpapp:diffFileListDownapp:toggleDiffNoiseFilterapp:toggleDiffPreSession

history 3

history:searchhistory:previoushistory:next

chat 16

chat:cancelchat:clearInputchat:clearScreenchat:killAgentschat:cycleModechat:modelPickerchat:fastModechat:thinkingTogglechat:submitchat:queueSubmitchat:sendNowchat:newlinechat:undochat:externalEditorchat:stashchat:imagePaste

autocomplete 4

autocomplete:acceptautocomplete:dismissautocomplete:previousautocomplete:next

confirm 8

confirm:yesconfirm:noconfirm:previousconfirm:nextconfirm:nextFieldconfirm:previousFieldconfirm:toggleconfirm:cycleMode

permission 1

permission:toggleDebug

transcript 2

transcript:toggleShowAlltranscript:exit

historySearch 5

historySearch:nexthistorySearch:accepthistorySearch:cancelhistorySearch:executehistorySearch:cycleScope

task 1

task:background

theme 1

theme:toggleSyntaxHighlighting

help 1

help:dismiss

tabs 2

tabs:nexttabs:previous

attachments 4

attachments:nextattachments:previousattachments:removeattachments:exit

footer 8

footer:nextfooter:previousfooter:upfooter:downfooter:openSelectedfooter:clearSelectionfooter:closefooter:dismiss

diff 6

diff:dismissdiff:previousSourcediff:nextSourcediff:previousFilediff:nextFilediff:back

scroll 10

scroll:pageUpscroll:pageDownscroll:fullPageUpscroll:fullPageDownscroll:topscroll:bottomscroll:lineUpscroll:lineDownscroll:halfPageUpscroll:halfPageDown

modelPicker 3

modelPicker:decreaseEffortmodelPicker:increaseEffortmodelPicker:thisSessionOnly

effortSlider 4

effortSlider:decreaseEfforteffortSlider:increaseEfforteffortSlider:toggleUltracodeeffortSlider:thisSessionOnly

select 8

select:nextselect:previousselect:pageUpselect:pageDownselect:firstselect:lastselect:acceptselect:cancel

plugin 3

plugin:toggleplugin:installplugin:favorite

abovePrompt 8

abovePrompt:toggleabovePrompt:focusabovePrompt:nextabovePrompt:previousabovePrompt:pressabovePrompt:leaveabovePrompt:highlightNextabovePrompt:highlightPrevious

pane 11

pane:scrollUppane:scrollDownpane:pageUppane:pageDownpane:toppane:bottompane:growpane:shrinkpane:closepane:nextpane:previous

settings 2

settings:searchsettings:retry

agents 6

agents:switchViewagents:togglePinagents:findagents:renameagents:previousGroupagents:nextGroup

voice 1

voice:pushToTalk

selection 8

selection:copyselection:clearselection:extendLeftselection:extendRightselection:extendUpselection:extendDownselection:extendLineStartselection:extendLineEnd