We got tired of not being able to quickly search through Chrome flags that are constantly changing. So we built this. 694 flags, searchable, updated daily.
Whether you're disabling GPU acceleration, running headless, or doing something weird with sandboxing — it's probably in here.
Last updated: February 2, 2026
--accept-lang----------------------------------------------------------------------------- Can't find the switch you are looking for? Try looking in: ash/constants/ash_switches.cc base/base_switches.cc etc. When commenting your switch, please use the same voice as surrounding comments. Imagine "This switch..." at the beginning of the phrase, and it'll all work out. ----------------------------------------------------------------------------- Specifies Accept-Language to send to servers and expose to JavaScript via the navigator.language DOM property. language[-country] where language is the 2 letter code from ISO-639.
--adaboost--add-xr-appcontainer-capsAdd additional capabilities to the AppContainer sandbox used for XR compositing.
--allarticles--allow-command-line-pluginsAllows plugins to be loaded in the command line for testing.
--allow-cross-origin-auth-promptAllows third-party content included on a page to prompt for a HTTP basic auth username/password pair.
--allow-file-access-from-filesBy default, file:// URIs cannot read other file:// URIs. This is an override for developers who need the old behavior for testing.
--allow-http-screen-captureAllow non-secure origins to use the screen capture API and the desktopCapture extension API.
--allow-insecure-localhostEnables TLS/SSL errors on localhost to be ignored (no interstitial, no blocking of requests).
--allow-ra-in-dev-modeAllows remote attestation (RA) in dev mode for testing purpose. Usually RA is disabled in dev mode because it will always fail. However, there are cases in testing where we do want to go through the permission flow even in dev mode. This can be enabled by this flag.
--allow-running-insecure-contentBy default, an https page cannot run JavaScript, CSS or plugins from http URLs. This provides an override to get the old insecure behavior.
--allow-sandbox-debuggingAllows debugging of sandboxed processes (see zygote_main_linux.cc).
--allow-silent-pushAllows Web Push notifications that do not show a notification.
--allow-third-party-modulesAllows third party modules to inject by disabling the BINARY_SIGNATURE mitigation policy on Win10+. Also has other effects in ELF.
--alsa-input-deviceThe Alsa device to use when opening an audio input stream.
--alsa-output-deviceThe Alsa device to use when opening an audio stream.
--alwaystrue--angle--animation-duration-scaleScale factor to apply to every animation duration. Must be >= 0.0. This will only apply to LinearAnimation and its subclasses.
--appSpecifies that the associated value should be launched in "application" mode.
--app-idSpecifies that the extension-app with the specified id should be launched according to its configuration.
--app-mode-auth-codeValue of GAIA auth code for --force-app-mode.
--app-mode-oauth-tokenValue of OAuth2 refresh token for --force-app-mode.
--app-run-on-os-login-modeThis is used along with kAppId to indicate an app was launched during OS login, and which mode the app was launched in.
--app-shimA process type (switches::kProcessType) that is used by App Shim processes. See chrome/app_shim/app_shim_main_delegate.mm.
--apps-gallery-download-urlOverrides the URL that the webstore APIs download extensions from. Note: the URL must contain one '%s' for the extension ID.
--apps-gallery-update-urlOverrides the update url used by webstore extensions.
--apps-gallery-urlOverrides the url that the browser treats as the webstore, granting it the webstore APIs and giving it some special protections.
--apps-keep-chrome-alive-in-testsPrevents Chrome from quitting when Chrome Apps are open.
--arcoreThe following are the runtimes that WebXr supports.
--as-browserFlag to launch tests in the browser process.
--ash-enable-software-mirroringscripts have been updated, the leading "ash" prefix should be removed. Enables software based mirroring.
--ash-enable-unified-desktopEnables unified desktop mode.
--ash-host-window-boundsSets a window size, optional position, optional scale factor and optional panel radii. "1024x768" creates a window of size 1024x768. "100+200-1024x768" positions the window at 100,200. "1024x768*2" sets the scale factor to 2 for a high DPI display. "1024x768~15|15|12|12" sets the radii of the panel corners as (upper_left=15px,upper_right=15px, lower_right=12px, upper_left=12px) "800,0+800-800x800" for two displays at 800x800 resolution. "800,0+800-800x800,0+1600-800x800" for three displays at 800x800 resolution.
--audio-buffer-sizeAllow users to specify a custom buffer size for debugging purpose.
--audio-codecs-from-edidAudio codecs supported by the HDMI sink is retrieved from the audio service process. EDID contains the Short Audio Descriptors, which list the audio decoders supported, and the information is presented as a bitmask of supported audio codecs.
--audio-process-high-priorityUse high priority for the audio process.
--auth-server-allowlistAllowlist for Negotiate Auth servers
--auth-spnego-account-typeAndroid authentication account type for SPNEGO authentication
--autoThe values the kTouchEventFeatureDetection switch may have, as in --touch-events=disabled. auto: enabled at startup when an attached touchscreen is present.
--auto-accept-browser-signin-for-testsAutomatically signs the user into Chrome when signing in to other Google services on the web. This makes it easier for automated browsers to sign in.
--auto-accept-this-tab-captureThese flags make Chrome auto-accept/reject requests to capture the current tab. It should only be used for tests.
--auto-open-devtools-for-tabsThis flag makes Chrome auto-open DevTools window for each tab. It is intended to be used by developers and automation to not require user interaction for opening DevTools.
--auto-reject-captureThis flag makes Chrome auto-reject requests capture a tab/window/screen.
--auto-reject-this-tab-capture--autofill-api-keySets the API key that will be used when calling Autofill API instead of using Chrome's baked key by default. You can use this to test new versions of the API that are not linked to the Chrome baked key yet.
--autofill-server-urlOverride the default autofill server URL with "scheme://host[:port]/prefix/".
--autoplay-policyCommand line flag name to set the autoplay policy.
--bgra--bootstrapValues for the kExtensionContentVerification flag. See ContentVerifierDelegate::Mode for more explanation.
--browser-startup-dialogCauses the browser process to display a dialog on launch.
--browser-subprocess-pathPath to the exe to run for the renderer and plugin subprocesses.
--browser-testBrowser test mode for the |kStartStackProfiler| switch. Limits the profile durations to be significantly less than the test timeout. On ChromeOS, forces the stack sampling profiler to run on all processes as well.
--browser-ui-tests-verify-pixelsCauses tests to attempt to verify pixel output.
--canvas-2d-layersEnable in-progress canvas 2d API methods BeginLayer and EndLayer.
--cardboard--cc-layer-tree-test-long-timeoutIncreases timeout for memory checkers.
--cc-layer-tree-test-no-timeoutPrevents the layer tree unit tests from timing out.
--change-stack-guard-on-forkAfter a zygote forks a new process, change the stack canary. This switch is useful so not all forked processes use the same canary (a secret value), which can be vulnerable to information leaks and brute force attacks. See https://crbug.com/1206626. This requires that all functions on the stack at the time content::RunZygote() is called be compiled without stack canaries. Valid values are "enable" or "disable".
--check-damage-earlyChecks damage early and aborts the frame if no damage, so that clients like Android WebView don't invalidate unnecessarily.
--check-for-update-intervalHow often (in seconds) to check for updates. Should only be used for testing purposes.
--cipher-suite-blacklistComma-separated list of SSL cipher suites to disable.
--classThe same as the --class argument in X applications. Overrides the WM_CLASS window property with the given value.
--clear-key-cdm-path-for-testingSpecifies the path to the Clear Key CDM for testing, which is necessary to support External Clear Key key system when library CDM is enabled. Note that External Clear Key key system support is also controlled by feature kExternalClearKeyForTesting.
--clear-token-serviceCommand line switches, sorted by name. Clears the token service before using it. This allows simulating the expiration of credentials during testing.
--code-sign-clone-cleanupA process type (switches::kProcessType) that cleans up the browser's temporary code sign clone.
--collect-dawn-info-eagerlyStart the GPU process for Dawn info collection immediately after the browser starts. The default is to delay for 120 seconds.
--compile-shader-always-succeedsAlways return success when compiling a shader. Linking will still fail.
--crash-on-hang-threadsComma-separated list of BrowserThreads that cause browser process to crash if the given browser thread is not responsive. UI/IO are the BrowserThreads that are supported. For example: --crash-on-hang-threads=UI:18,IO:18 --> Crash the browser if UI or IO is not responsive for 18 seconds while the other browser thread is responsive.
--crash-testCauses the browser process to crash on startup.
--creditsPrints licensing information (same content as found in about:credits) and quits.
--crosh-commandCustom crosh command.
--custom-devtools-frontendSpecifies the http:// endpoint which will be used to serve devtools://devtools/custom/<path> Or a file:// URL to specify a custom file path to load from for devtools://devtools/bundled/<path>
--d3d11--d3d11-nullSpecial switches for "NULL"/stub driver implementations.
--d3d11-warp--d3d11-warp-webgl--d3d11on12--d3d9--dawn--dawn-d3d11--dawn-d3d12--dawn-metal--dawn-opengles--dawn-swiftshader--dawn-vulkan--debug-packed-appsAdds debugging entries such as Inspect Element to context menus of packed apps.
--debug-printEnables support to debug printing subsystem.
--default--default-country-codeDefault country code to be used for search engine localization.
--device-management-urlSpecifies the URL at which to communicate with the device management backend to fetch configuration policies and perform other device tasks.
--device-scale-factorDevice scale factor passed to certain processes like renderers, etc.
--devtools-flagsPasses command line parameters to the DevTools front-end.
--devtools-greendev-uiEnables the experimental GreenDev UI in DevTools.
--diagnosticsTriggers a plethora of diagnostic modes.
--diagnostics-formatSets the output format for diagnostic modes enabled by diagnostics flag.
--diagnostics-recoveryTells the diagnostics mode to do the requested recovery step(s).
--disable--disable-2d-canvas-clip-aaDisable antialiasing on 2d canvas clips
--disable-2d-canvas-image-chromiumDisables Canvas2D rendering into a scanout buffer for overlay support.
--disable-3d-apisDisables client-visible 3D APIs, in particular WebGL. This is controlled by policy and is kept separate from the other enable/disable switches to avoid accidentally regressing the policy support for controlling access to these APIs.
--disable-accelerated-2d-canvasDisable gpu-accelerated 2d canvas.
--disable-angle-featuresDisables specified comma separated ANGLE features if found.
--disable-audio-inputForces input and output stream creation to use fake audio streams.
--disable-audio-output--disable-auto-maximize-for-testsDisables the auto maximize feature on ChromeOS so that a browser window always starts in normal state. This is used by tests that do not want this auto maximizing behavior.
--disable-auto-reloadDisable auto-reload of pages on top-level error.
--disable-back-forward-cacheDisables the BackForwardCache feature.
--disable-background-networkingDisable several subsystems which run network requests in the background. This is for use when doing network performance testing to avoid noise in the measurements.
--disable-backing-store-limitDisable limits on the number of backing stores. Can prevent blinking for users with many windows/tabs and lots of memory.
--disable-best-effort-tasksDelays execution of TaskPriority::BEST_EFFORT tasks until shutdown.
--disable-blink-featuresDisable one or more Blink runtime-enabled features. Use names from runtime_enabled_features.json5, separated by commas. Applied after kEnableBlinkFeatures, and after other flags that change these features.
--disable-breakpadDisables the crash reporting.
--disable-cancel-all-touchesDisable CancelAllTouches() function for the implementation on cancel single touches.
--disable-canvas-aaDisable antialiasing on 2d canvas.
--disable-checker-imagingDisabled defering all image decodes to the image decode service, ignoring DecodingMode preferences specified on PaintImage.
--disable-component-update--disable-crashpad-for-testingDisables crashpad initialization for testing. The crashpad binary will not run, and thus will not detect and symbolize crashes.
--disable-d3d11-warpExplicitly disable D3D11 WARP fallback. Some test suites prefer falling back to swiftshader.
--disable-dawn-featuresSet the Dawn features(toggles) disabled on the creation of Dawn devices.
--disable-default-appsDisables installation of default apps on first run. This is used during automated testing.
--disable-default-browser-promoDisable the default browser promo.
--disable-dev-shm-usageThe /dev/shm partition is too small in certain VM environments, causing Chrome to fail or crash (see http://crbug.com/715363). Use this flag to work-around this issue (a temporary directory will always be used to create anonymous shared memory files).
--disable-direct-compositionDisable DirectComposition.
--disable-domain-reliabilityDisables Domain Reliability Monitoring.
--disable-explicit-dma-fencesDisable explicit DMA-fences
--disable-featuresComma-separated list of feature names to disable. See also kEnableFeatures.
--disable-field-trial-configDisable field trial tests configured in fieldtrial_testing_config.json.
--disable-file-systemDisable FileSystem API.
--disable-gl-drawing-for-testsDisables GL drawing operations which produce pixel output. With this the GL output will not be correct but tests will run faster.
--disable-gl-error-limitDisable the GL error log limit.
--disable-gl-extensionsDisables specified comma separated GL Extensions if found.
--disable-glsl-translatorDisable the GLSL translator.
--disable-gpuDisables GPU hardware acceleration. If software renderer is not in place, then the GPU process won't launch.
--disable-gpu-compositingPrevent the compositor from using its GPU implementation.
--disable-gpu-early-initDisable proactive early init of GPU process.
--disable-gpu-process-crash-limitFor tests, to disable the limit on the number of times the GPU process may be restarted.
--disable-gpu-program-cacheTurn off gpu program caching
--disable-gpu-rasterizationDisable GPU rasterization, i.e. rasterize on the CPU only. Overrides the kEnableGpuRasterization flag.
--disable-gpu-sandboxDisables the GPU process sandbox.
--disable-gpu-shader-disk-cacheDisables the GPU shader on disk cache.
--disable-gpu-vsyncStop the GPU from synchronizing presentation with vblank.
--disable-gpu-watchdogDisable the thread that crashes the GPU process if it stops responding to messages.
--disable-highres-timerDisable high-resolution timer on Windows.
--disable-histogram-customizerDisable the RenderThread's HistogramCustomizer.
--disable-http2-grease-settings--disable-in-process-stack-tracesDisables the in-process stack traces.
--disable-ipc-flooding-protectionDisables the IPC flooding protection. It is activated by default. Some javascript functions can be used to flood the browser process with IPC. This protection limits the rate at which they can be used.
--disable-kill-after-bad-ipcDon't kill a child process when it sends a bad IPC message. Apart from testing, it is a bad idea from a security perspective to enable this switch.
--disable-landlock-sandboxDisables the Landlock sandbox (Android only).
--disable-lazy-loadingDisables lazy loading of images and frames.
--disable-lcd-textDisables LCD text.
--disable-legacy-windowDisable the Legacy Window which corresponds to the size of the WebContents.
--disable-local-storageDisable LocalStorage.
--disable-loggingForce logging to be disabled. Logging is enabled by default in debug builds.
--disable-logging-redirectDisables logging redirect for testing.
--disable-login-screen-appsDisables apps on the login screen. By default, they are allowed and can be installed through policy.
--disable-low-end-device-modeForce disabling of low-end device mode when set.
--disable-low-latency-dxvaDisables using CODECAPI_AVLowLatencyMode when creating DXVA decoders.
--disable-media-session-apiDisable Media Session API
--disable-metal-shader-cacheDisables Metal's shader cache, using the GPU sandbox to prevent access to it.
--disable-mipmap-generationDisables mipmap generation in Skia. Used a workaround for select low memory devices, see https://crbug.com/1138979 for details.
--disable-mojo-brokerDisables Mojo broker capabilities in the browser during Mojo initialization.
--disable-namespace-sandboxDisables usage of the namespace sandbox.
--disable-notificationsDisables the Web Notification and the Push APIs.
--disable-nv12-dxgi-videoDisables the video decoder from drawing to an NV12 textures instead of ARGB.
--disable-oopr-debug-crash-dumpDisables debug crash dumps for OOPR.
--disable-policy-key-verificationDisables the verification of policy signing keys. It just works on Chrome OS test images and crashes otherwise. better solution to enable policy tast test for Family Link account.
--disable-presentation-apiDisables the Presentation API.
--disable-print-previewDisables print preview (For testing, and for users who don't like us. :[ )
--disable-prompt-on-repostNormally when the user attempts to navigate to a page that was the result of a post we prompt to make sure they want to. This switch may be used to disable that check. This switch is used during automated testing.
--disable-pushstate-throttleDisables throttling of history.pushState/replaceState calls.
--disable-reading-from-canvasTaints all <canvas> elements, regardless of origin.
--disable-remote-fontsDisables remote web font support. SVG font should always work whether this option is specified or not.
--disable-remote-playback-apiDisables the RemotePlayback API.
--disable-renderer-backgroundingPrevent renderer process backgrounding when set.
--disable-resource-schedulerWhether the ResourceScheduler is disabled. Note this is only useful for C++ Headless embedders who need to implement their own resource scheduling.
--disable-screen-orientation-lockDisable the locking feature of the screen orientation API.
--disable-scroll-to-text-fragmentThis switch disables the ScrollToTextFragment feature.
--disable-seccomp-filter-sandboxDisable the seccomp filter sandbox (seccomp-bpf) (Linux only).
--disable-setuid-sandboxDisable the setuid sandbox (Linux only).
--disable-shader-name-hashingTurn off user-defined name hashing in shaders.
--disable-shared-workersDisable shared workers.
--disable-site-isolation-trialsDisables site isolation. Note that the opt-in (to site-per-process, isolate-origins, etc.) via enterprise policy and/or cmdline takes precedence over the kDisableSiteIsolation switch (i.e. the opt-in takes effect despite potential presence of kDisableSiteIsolation switch). Note that for historic reasons the name of the switch misleadingly mentions "trials", but the switch also disables the default site isolation that ships on desktop since M67. The name of the switch is preserved for backcompatibility of chrome://flags.
--disable-skia-graphiteForce disabling/enabling Skia Graphite. Disabling will take precedence over enabling if both are specified.
--disable-skia-runtime-optsDo not use runtime-detected high-end CPU optimizations in Skia. This is useful for forcing a baseline code path for e.g. web tests.
--disable-smooth-scrollingDisable smooth scrolling for testing.
--disable-software-rasterizerDisables the use of a 3D software rasterizer.
--disable-speech-apiDisables the Web Speech API (both speech recognition and synthesis).
--disable-speech-synthesis-apiDisables the speech synthesis part of Web Speech API.
--disable-stack-profilerDisable stack profiling. Stack profiling may change performance. Disabling stack profiling is beneficial when comparing performance metrics with a build that has it disabled by default.
--disable-threaded-animation--disable-threaded-compositingDisable multithreaded GPU compositing of web content.
--disable-timeouts-for-profilingDisable timeouts that may cause the browser to die when running slowly. This is useful if running with profiling (such as debug malloc).
--disable-usb-keyboard-detectDisables the USB keyboard detection for blocking the OSK on Windows.
--disable-v8-idle-tasksDisable V8 idle tasks.
--disable-variations-safe-modeDisable variations safe mode.
--disable-variations-seed-fetchDisable fetching of variations seed from the server for testing.
--disable-vsync-for-tests--disable-vulkan-surfaceDisables VK_KHR_surface extension. Instead of using swapchain, bitblt will be used for present render result on screen.
--disable-wayland-imeDisable wayland input method editor.
--disable-web-securityDon't enforce the same-origin policy; meant for website testing only. This switch has no effect unless --user-data-dir (as defined by the content embedder) is also present.
--disable-webglDisable all versions of WebGL.
--disable-webgl-image-chromiumDisables WebGL rendering into a scanout buffer for overlay support.
--disable-webgl2Disable WebGL2.
--disable-webrtc-encryptionDisables encryption of RTP Media for WebRTC. When Chrome embeds Content, it ignores this switch on its stable and beta channels.
--disable-yuv-image-decodingDisable YUV image decoding for those formats and cases where it's supported. Has no effect unless GPU rasterization is enabled.
--disable-zero-copy-dxgi-videoDisable the video decoder from drawing directly to a texture.
--disableddisabled: touch events are disabled.
--discoverability--disk-cache-dirUse a specific disk cache location, rather than one derived from the UserDatadir.
--disk-cache-sizeForces the maximum disk space to be used by the disk cache, in bytes.
--displayWhich X11 display to connect to. Emulates the GTK+ "--display=" command line argument. In use only with Ozone/X11.
--display-propertiesAdditional display properties are provided through this switch that are beyond what is available via EDID encoded as JSON. Please see `https://chromium.googlesource.com/chromiumos/platform2/+/dd10a5ae3618bb9dc5fb47ac415ebef6e9a3827d/chromeos-config/README.md#displays` for the data format.
--dmg-deviceWhen switches::kProcessType is switches::kRelauncherProcess, if this switch is also present, the relauncher process will unmount and eject a mounted disk image and move its disk image file to the trash. The argument's value must be a BSD device name of the form "diskN" or "diskNsM".
--do-not-create-nsapp-for-testsSkips initializing the shares NSApplication instance in ChromeTestSuite.
--do-not-de-elevateDo not de-elevate the browser on launch. Used after de-elevating to prevent infinite loops.
--dom-automationSpecifies if the |DOMAutomationController| needs to be bound in the renderer. This binding happens on per-frame basis and hence can potentially be a performance bottleneck. One should only enable it when automating dom based tests.
--dump-browser-histogramsRequests that a running browser process dump its collected histograms to a given file. The file is overwritten if it exists.
--dump-compositor-frameSave CompositorFrame data in both renderer and gpu process for TreesInViz and TreeAnimationsInViz. This doesn't work without --no-sandbox. If no parameters are specified, by default, it records frame 1 - 10. If --dump-compositor-frame=begin,end is specified, it records frame [begin, end] inclusive.
--edge-touch-filteringTells Chrome to do edge touch filtering. Useful for convertible tablet.
--egl--enable--enable-angle-featuresANGLE features are defined per-backend in third_party/angle/include/platform Enables specified comma separated ANGLE features if found.
--enable-auto-reloadEnable auto-reload of pages on top-level error.
--enable-automationEnable indication that browser is controlled by automation.
--enable-benchmarkingSee flag_descriptions.cc for more details.
--enable-benchmarking-apiEnables the benchmarking JavaScript API.
--enable-blink-featuresEnable one or more Blink runtime-enabled features. Use names from runtime_enabled_features.json5, separated by commas. Applied before kDisableBlinkFeatures, and after other flags that change these features.
--enable-blink-test-featuresEnables blink runtime enabled features with status:"test" or status:"experimental", which are enabled when running web tests.
--enable-bookmark-undoEnables the multi-level undo system for bookmarks.
--enable-caret-browsingEnable native caret browsing, in which a moveable cursor is placed on a web page, allowing a user to select and navigate through non-editable text using just a keyboard. See https://crbug.com/977390 for links to i2i.
--enable-cloud-print-proxyThis applies only when the process type is "service". Enables the Cloud Print Proxy component within the service process.
--enable-crash-reporterIndicates that crash reporting should be enabled. On platforms where helper processes cannot access to files needed to make this decision, this flag is generated internally.
--enable-dawn-backend-validationEnable validation layers in Dawn backends.
--enable-dawn-featuresSet the Dawn features(toggles) enabled on the creation of Dawn devices.
--enable-devtools-pwa-handlerIf enabled, DevTools will allow creating pwa_handler, to enable executing CDP methods (i.e. PWA.install) on browsers connected remotely
--enable-distillability-service--enable-dom-distiller--enable-domain-reliabilityEnables Domain Reliability Monitoring.
--enable-exclusive-audioUse exclusive mode audio streaming for Windows Vista and higher. Leads to lower latencies for audio streams which uses the AudioParameters::AUDIO_PCM_LOW_LATENCY audio path. See http://msdn.microsoft.com/en-us/library/windows/desktop/dd370844.aspx for details.
--enable-extension-actor-apiName of the command line flag to allow the experimental actor API.
--enable-extension-ai-data-collectionName of the command line flag to allow the ai data collection extension API.
--enable-featuresComma-separated list of feature names to enable. See also kDisableFeatures.
--enable-field-trial-configEnable field trial tests configured in fieldtrial_testing_config.json. If the "disable_fieldtrial_testing_config" GN flag is set to true, then this switch is a no-op. Otherwise, for non-Chrome branded builds, the testing config is already applied by default, unless the "--disable-field-trial-config", "--force-fieldtrials", and/or "--variations-server-url" switches are passed. It is however possible to apply the testing config as well as specify additional field trials (using "--force-fieldtrials") by using this switch. For Chrome-branded builds, the testing config is not enabled by default, so this switch is required to enable it.
--enable-gpu-benchmarkingEnables the GPU benchmarking extension
--enable-gpu-blocked-timeEnables measures of how long GPU Main Thread was blocked between SwapBuffers
--enable-gpu-command-loggingTurn on Logging GPU commands.
--enable-gpu-debuggingTurn on Calling GL Error after every command.
--enable-gpu-driver-debug-loggingEnable logging of GPU driver debug messages.
--enable-gpu-rasterizationAllow heuristics to determine when a layer tile should be drawn with the Skia GPU backend. Only valid with GPU accelerated compositing.
--enable-gpu-service-loggingEnable GPU service logging. Note: This is the same switch as the one in gl_switches.cc. It's defined here again to avoid dependencies between dlls.
--enable-gpu-service-tracingTurns on calling TRACE for every GL call.
--enable-idle-tracingAdds additional thread idle time information into the trace event output.
--enable-lcd-textEnables LCD text.
--enable-loggingForce logging to be enabled. Logging is disabled by default in release builds.
--enable-longpress-drag-selectionEnable drag manipulation of longpress-triggered text selections.
--enable-low-end-device-modeForce low-end device mode when set.
--enable-native-gpu-memory-buffersEnable native CPU-mappable GPU memory buffer support on Linux.
--enable-net-benchmarkingEnables the network-related benchmarking extensions.
--enable-new-app-menu-icon--enable-pixel-output-in-testsForces tests to produce pixel output when they normally wouldn't.
--enable-precise-memory-infoMake the values returned to window.performance.memory more granular and more up to date in shared worker. Without this flag, the memory information is still available, but it is bucketized and updated less frequently. This flag also applys to workers.
--enable-privacy-sandbox-ads-apisEnables Privacy Sandbox APIs: Attribution Reporting, Fledge, Topics, Fenced Frames, Shared Storage, Private Aggregation, and their associated features.
--enable-profile-shortcut-managerForce-enables the profile shortcut manager. This is needed for tests since they use a custom-user-data-dir which disables this.
--enable-protected-video-buffersEnables protected buffers for encrypted video streams.
--enable-sandbox-loggingCause the OS X sandbox write to syslog every time an access to a resource is denied by the sandbox.
--enable-scaling-clipped-imagesAllows scaling clipped images in GpuImageDecodeCache. Note that this may cause color-bleeding. cache problems are solved.
--enable-service-binary-launcherIf true the ServiceProcessLauncher is used to launch services. This allows for service binaries to be loaded rather than using the utility process. This is only useful for tests.
--enable-sgi-video-syncEnable use of the SGI_video_sync extension, which can have driver/sandbox/window manager compatibility issues.
--enable-skia-benchmarkingEnables the Skia benchmarking extension.
--enable-skia-graphite--enable-smooth-scrollingOn platforms that support it, enables smooth scroll animation.
--enable-spatial-navigationEnable spatial navigation
--enable-speech-dispatcherAllows sending text-to-speech requests to speech-dispatcher, a common Linux speech service. Because it's buggy, the user must explicitly enable it so that visiting a random webpage can't cause instability.
--enable-swap-buffers-with-boundsEnables SwapBuffersWithBounds if it is supported.
--enable-tracing-fractionWhen specified along with a value in the range (0,1] will --enable-tracing for (roughly) that percentage of tests being run. This is done in a stable manner such that the same tests are chosen each run, and under the assumption that tests hash equally across the range of possible values. The flag will enable all tracing categories for those tests, and none for the rest. This flag could be used with other tracing switches like --enable-tracing-format, but any other switches that will enable tracing will turn tracing on for all tests.
--enable-unsafe-swiftshaderAllow usage of SwiftShader for WebGL
--enable-unsafe-webgpu--enable-user-metricsEnable user metrics from within the installer.
--enable-viewportEnables the use of the @viewport CSS rule, which allows pages to control aspects of their own layout. This also turns on touch-screen pinch gestures.
--enable-vtune-supportEnable the Vtune profiler support.
--enable-vulkan-protected-memoryEnables using protected memory for vulkan resources.
--enable-wayland-imeTry to enable wayland input method editor.
--enable-webgl-draft-extensionsEnables WebGL extensions not yet approved by the community.
--enable-webgl-image-chromiumEnables WebGL rendering into a scanout buffer for overlay support.
--enabledenabled: touch events always enabled.
--encrypted-reporting-urlSpecifies the URL at which to upload encrypted reports.
--enforce--enforce-gl-minimumsEnforce GL minimums.
--ensure-forced-color-profileCrash the browser at startup if the display's color profile does not match the forced color profile. This is necessary on Mac because Chrome's pixel output is always subject to the color conversion performed by the operating system. On all other platforms, this is a no-op.
--experimental-ai-stable-channelAllows experimental ai extension APIs to be used in stable channel. This disables chrome sign-in if set, regardless of channel.
--explicitly-allowed-portsAllows overriding the list of restricted ports by passing a comma-separated list of port numbers.
--extension-content-verificationName of the command line flag to force content verification to be on in one of various modes.
--fail-audio-stream-creationCauses the AudioManager to fail creating audio streams. Used when testing various failure cases.
--fake-variations-channelFakes the channel of the browser for purposes of Variations filtering. This is to be used for testing only. Possible values are "stable", "beta", "dev" and "canary". This works for official builds as well.
--field-trial-handleHandle to the shared memory segment containing field trial state that is to be shared between processes. The argument to this switch is made of segments separated by commas: - The platform-specific handle id for the shared memory as a string. - (Windows only) i=inherited by duplication or p=child must open parent. - The high 64 bits of the shared memory block GUID. - The low 64 bits of the shared memory block GUID. - The size of the shared memory segment as a string.
--file-storage-server-upload-urlSpecifies the URL at which to communicate with File Storage Server (go/crosman-file-storage-server) to upload log and support packet files.
--file-url-path-aliasDefine an alias root directory which is replaced with the replacement string in file URLs. The format is "/alias=/replacement", which would turn file:///alias/some/path.html into file:///replacement/some/path.html.
--focusActivates an existing tab or app window by URL or app id before creating anything new. Syntax: comma-ordered selectors. Bare URLs are exact. Add a trailing * for prefix. app:<app-id> targets PWAs. Example: --focus=https://meet.google.com/*,app:abc123
--focus-result-fileSpecifies a file path to write JSON focus result information.
--font-cache-shared-handleDirectWrite FontCache is shared by browser to renderers using shared memory. This switch allows us to pass the shared memory handle to the renderer.
--force-app-modeForces application mode. This hides certain system UI elements and forces the app to be installed if it hasn't been already.
--force-browser-crash-on-gpu-crashCrash Chrome if GPU process crashes. This is to force a test to fail when GPU process crashes unexpectedly.
--force-color-profileForce all monitors to be treated as though they have the specified color profile. Accepted values are "srgb" and "generic-rgb" (currently used by Mac layout tests) and "color-spin-gamma24" (used by layout tests).
--force-desktop-androidForces the DeviceInfo.isDesktop() check to return true. Can be used to enable desktop-only features on other form factors.
--force-device-ownershipForces the device to report being owned by an enterprise. This mimics the presence of an app signaling device ownership.
--force-device-scale-factorOverrides the device scale factor for the browser UI and the contents.
--force-devtools-availableForces developer tools availability, no matter what values the enterprise policies DeveloperToolsDisabled and DeveloperToolsAvailability are set to.
--force-disable-variation-idsForces to remove Chrome Variation Ids from being sent in X-Client-Data header, specified as a 64-bit encoded list of numeric experiment ids. Ids prefixed with the character "t" will be treated as Trigger Variation Ids.
--force-effective-connection-typeForces Network Quality Estimator (NQE) to return a specific effective connection type.
--force-enable-night-modeForces the night mode to be enabled.
--force-fieldtrial-paramsThis option can be used to force parameters of field trials when testing changes locally. The argument is a param list of (key, value) pairs prefixed by an associated (trial, group) pair. You specify the param list for multiple (trial, group) pairs with a comma separator. Example: "Trial1.Group1:k1/v1/k2/v2,Trial2.Group2:k3/v3/k4/v4" Trial names, groups names, parameter names, and value should all be URL escaped for all non-alphanumeric characters.
--force-fieldtrialsThis option can be used to force field trials when testing changes locally. The argument is a list of name and value pairs, separated by slashes. If a trial name is prefixed with an asterisk, that trial will start activated. For example, the following argument defines two trials, with the second one activated: "GoogleNow/Enable/*MaterialDesignNTP/Default/" This option can also be used by the browser process to send the list of trials to a non-browser process, using the same format. See FieldTrialList::CreateTrialsFromString() in field_trial.h for details.
--force-first-runDisplays the First Run experience when the browser is started, regardless of whether or not it's actually the First Run (this overrides kNoFirstRun).
--force-fre-default-browser-stepForce enable the default browser step in the first run experience on Desktop.
--force-high-performance-gpu--force-high-res-timeticksForces the use of QPC for TimeTicks even if cpuid doesn't report the presence of an invariant TSC.
--force-max-texture-sizeSets the maximum texture size in pixels.
--force-prefers-no-reduced-motionForces whether the user desires no reduced motion, regardless of system settings.
--force-prefers-reduced-motionForces whether the user desires reduced motion, regardless of system settings.
--force-raster-color-profileForce rastering to take place in the specified color profile. Accepted values are the same as for the kForceDisplayColorProfile case above.
--force-show-update-menu-badgeForces the update menu badge to show.
--force-update-menu-typeForces the update menu type to a specific type.
--force-variation-idsForces additional Chrome Variation Ids that will be sent in X-Client-Data header, specified as a 64-bit encoded list of numeric experiment ids. Ids prefixed with the character "t" will be treated as Trigger Variation Ids. IMPORTANT: You can use this switch for test purposes (e.g. a manual command line run or from a unit test), but NOT for production code in the browser, as the latter is not allowed for privacy reasons (except for the current use by about:flags code).
--force-video-overlaysForce media player using SurfaceView instead of SurfaceTexture on Android. Note: This is used by the Cast playback pipeline and must be kept.
--force-wave-audioUse Windows WaveOut/In audio API even if Core Audio is supported.
--force-webgpu-compatForce all WebGPU content to run in WebGPU Compatibility mode.
--force-webrtc-ip-handling-policyOverride WebRTC IP handling policy to mimic the behavior when WebRTC IP handling policy is specified in Preferences.
--force-webxr-runtimeForcibly enable and select the specified runtime for webxr. Note that this provides an alternative means of enabling a runtime, and will also functionally disable all other runtimes.
--force-whats-newDisplays the What's New experience when the browser is started if it has not yet been shown for the current milestone (this overrides kNoFirstRun, without showing the First Run experience).
--from-browser-switcherIndicates that this launch of the browser originated from the Legacy Browser Support for Edge extension's native host. This is recorded in UMA.
--from-installerIndicates that this launch of the browser originated from the installer (i.e., following a successful new install or over-install). This triggers browser behaviors for this specific launch, such as a welcome announcement for accessibility software (see https://crbug.com/1072735).
--full-memory-crash-reportGenerates full memory crash dump.
--gl-egl--gl-null--gl-shader-interm-outputInclude ANGLE's intermediate representation (AST) output in shader compilation info logs.
--gles--gles-egl--gles-null--glic-admin-redirect-patternsList of URL patterns in the glic webview to redirect to an admin blocked panel, as a space-separated list.
--glic-always-open-fre--glic-automationAutomation is intended to be passed in addition to glic-dev. It further disables functionality to make basic testing easier.
--glic-devDev mode for glic only exposed via command line flag.
--glic-force-g1-for-miOverride actual AI subscription tier by forcing G1 status, specifically for multi-instance enablement. Intended for manual testing only.
--glic-fre-url--glic-guest-urlOverrides the glic guest URL.
--glic-guest-url-preset-autopushConfigure preset guest URLs for manual testing. These are saved to profile prefs and can be selected to override the default glic guest URL through corresponding entries in chrome://flags.
--glic-guest-url-preset-preprod--glic-guest-url-preset-prod--glic-host-loggingWhether additional logging is enabled in the glic api host.
--glic-open-on-startupUse --glic-open-on-startup=attached or --glic-open-on-startup=detached.
--glic-shortcuts-learn-more-url--glic-webui-allowed-originsList of allowed origins in the glic webview, as a space-separated list.
--gpu-blocklist-test-groupSelect a different set of GPU blocklist entries with the specified test_group ID.
--gpu-device-idPasses the active graphics device id from browser process to info collection GPU process.
--gpu-disk-cache-size-kbAllows explicitly specifying the shader disk cache size for embedded devices. Default value is 6MB. On Android, 2MB is default and 128KB for low-end devices.
--gpu-driver-bug-list-test-groupEnable an extra set of GPU driver bug list entries with the specified test_group ID. Note the default test group (group 0) is still active.
--gpu-driver-versionPasses the active graphics driver version from browser process to info collection GPU process.
--gpu-launcherExtra command line options for launching the GPU process (normally used for debugging). Use like renderer-cmd-prefix.
--gpu-no-context-lostInform Chrome that a GPU context will not be lost in power saving mode, screen saving mode, etc. Note that this flag does not ensure that a GPU context will never be lost in any situations, say, a GPU reset.
--gpu-preferencesPasses encoded GpuPreferences to GPU process.
--gpu-processMakes this process a GPU sub-process.
--gpu-program-cache-size-kbSets the maximum size of the in-memory gpu program cache, in kb
--gpu-revisionPasses the active graphics revision info from browser process to info collection GPU process.
--gpu-sandbox-allow-sysv-shmAllows shmat() system call in the GPU sandbox.
--gpu-sandbox-failures-fatalMakes GPU sandbox failures fatal.
--gpu-sandbox-start-earlyStarts the GPU sandbox before creating a GL context.
--gpu-startup-dialogCauses the GPU process to display a dialog on launch.
--gpu-sub-system-idPasses the active graphics sub system id from browser process to info collection GPU process.
--gpu-vendor-idPasses the active graphics vendor id from browser process to info collection GPU process.
--gpu-watchdog-timeout-secondsOverride value for the GPU watchdog timeout in seconds.
--gpu2-startup-dialogCauses the second GPU process used for gpu info collection to display a dialog on launch.
--guestCauses the browser to launch directly in guest mode.
--hardware-video-decode-framerateSome (Qualcomm only at the moment) V4L2 video decoders require setting the framerate so that the hardware decoder can scale the clocks efficiently. This provides a mechanism during testing to lock the decoder framerate to a specific value.
--headlessRun in headless mode, i.e., without a UI or display server dependencies.
--helpThese flags show the man page on Linux. They are equivalent to each other.
--hide-crash-restore-bubbleDoes not show the crash restore bubble when the browser is started during the system startup phase in ChromeOS, if the ChromeOS full restore feature is enabled, because the ChromeOS full restore notification is shown for the user to select restore or not.
--hide-iconsMakes Windows happy by allowing it to show "Enable access to this program" checkbox in Add/Remove Programs->Set Program Access and Defaults. This only shows an error box because the only way to hide Chrome is by uninstalling it.
--hide-scrollbarsPrevents creating scrollbars for web content. Useful for taking consistent screenshots.
--homepageSpecifies which page will be displayed in newly-opened tabs. We need this for testing purposes so that the UI tests don't depend on what comes up for http://google.com.
--host-package-label--host-package-nameWhen we retrieve the package name within the SDK Runtime, we need to use a bit of a hack to do this by taking advantage of the fact that the pid is the same pid as the application's pid + 10000. see: https://cs.android.com/android/platform/superproject/main/+/main:frameworks/base/core/java/android/os/Process.java;l=292;drc=47fffdd53115a9af1820e3f89d8108745be4b55d When the render process is created however, it is just a regular isolated process with no particular association so we can't perform the same hack. When creating minidumps, the package name is retrieved from the process meaning the render process minidumps would end up reporting a generic process name not associated with the app. We work around this by feeding through the host package information to the render process when launching it.
--host-resolver-rulesThese mappings only apply to the host resolver.
--host-rulesDeprecated alias for backwards compatibility, now does the same thing as --host-resolver-rules which should be used instead.
--host-version-code--ignore-bad-message-for-testingThe switch to ignore bad mojo message reports. Only for testing.
--ignore-gpu-blocklistIgnores GPU blocklist.
--import-passwordsTriggers the import of passwords on startup.
--in-process-gpuRun the GPU process as a thread in the browser process.
--incognitoCauses the initial browser opened to be in incognito mode. Further browsers may or may not be in incognito mode; see `IncognitoModePrefs`.
--init-isolate-as-foregroundSpecifies that the main-thread Isolate should initialize in foreground mode. If not specified, the the Isolate will start in background mode for extension processes and foreground mode otherwise.
--install-autogenerated-themeInstalls an autogenerated theme based on the given RGB value. The format is "r,g,b", where r, g, b are a numeric values from 0 to 255.
--install-chrome-appCauses Chrome to initiate an installation flow for the given app.
--instant-processMarks a renderer as an Instant process.
--ip-address-space-overridesSpecifies manual overrides to the IP endpoint -> IP address space mapping. This allows running local tests against "public" and "local" IP addresses. This switch is specified as a comma-separated list of overrides. Each override is an equals-separated "<endpoint|ip-range>=<address space>" pair. Grammar, in pseudo-BNF format: switch := override-list override-list := override “,” override-list | <nil> override := (ip-endpoint | ip_range) “=” address-space address-space := “public” | “private” | “local” | "loopback" ip-endpoint := ip-address ":" port ip-address := see `net::ParseURLHostnameToAddress()` for details port := integer in the [0-65535] range ip-range := ip-address "/" bitmask bitmask := integer in the [0-128] range Any invalid entries in the comma-separated list are ignored. If the port specified is 0, all ports for the given ip-address will be overridden. See also the design doc: https://docs.google.com/document/d/1-umCGylIOuSG02k9KGDwKayt3bzBXtGwVlCQHHkIcnQ/edit# And the Web Platform Test RFC #72 behind it: https://github.com/web-platform-tests/rfcs/blob/master/rfcs/address_space_overrides.md Note that since the doc and the RFC were written, the address space names have changed slightly due to Local Network Access (LNA) replacing Private Network Access (PNA).
--ipc-connection-timeoutOverrides the timeout, in seconds, that a child process waits for a connection from the browser before killing itself.
--ipc-dump-directoryDumps IPC messages sent from renderer processes to the browser process to the given directory. Used primarily to gather samples for IPC fuzzing.
--ipc-fuzzer-testcaseSpecifies the testcase used by the IPC fuzzer.
--isolate-originsRequire dedicated processes for a set of origins, specified as a comma-separated list. For example: --isolate-origins=https://www.foo.com,https://www.bar.com
--javascript-harmonyEnables experimental Harmony (ECMAScript 6) features.
--keep-alive-for-testUsed for testing - keeps browser alive after last browser window closes.
--kioskEnable kiosk mode. Please note this is not Chrome OS kiosk mode.
--kiosk-printingEnable automatically pressing the print button in print preview.
--launch-time-ticksTime the browser launched the renderer process (in TimeTicks).
--layer--list-appsWrites open and installed web apps for each profile to the specified file without launching a new browser window or tab. Pass a absolute file path to specify where to output the information. Can be used together with optional --profile-base-name switch to only write information for a given profile.
--load-webui-from-diskFlag used to load WebUI files directly from disk instead of pak files. Meant to be used during local development only (requires a local checkout and build), and only works if used along with the GN load_webui_from_disk=true GN flag.
--log-best-effort-tasksLogs information about all tasks posted with TaskPriority::BEST_EFFORT. Use this to diagnose issues that are thought to be caused by TaskPriority::BEST_EFFORT execution fences. Note: Tasks posted to a non-BEST_EFFORT UpdateableSequencedTaskRunner whose priority is later lowered to BEST_EFFORT are not logged.
--log-fileOverrides the default file name to use for general-purpose logging (does not affect which events are logged).
--log-gpu-control-list-decisionsLogs GPU control list decisions when enforcing blocklist rules.
--log-levelSets the minimum log level. Valid values are from 0 to 3: INFO = 0, WARNING = 1, LOG_ERROR = 2, LOG_FATAL = 3.
--log-missing-unload-ackLog an error whenever the unload timeout for a render frame is exceeded.
--log-net-logEnables saving net log events to a file. If a value is given, it used as the path the the file, otherwise the file is named netlog.json and placed in the user data directory.
--log-on-ui-double-background-blurChecks and logs double background blur as an error if any.
--make-chrome-defaultIndicates whether Chrome should be set as the default browser during installation.
--make-default-browserMakes Chrome default browser
--market-url-for-testingSets the market URL for Chrome for use in testing.
--max-active-webgl-contextsAllows user to override maximum number of active WebGL contexts per renderer process.
--max-decoded-image-size-mbSets the maximium decoded image size limitation.
--max-gum-fpsOverride the maximum framerate as can be specified in calls to getUserMedia. This flag expects a value. Example: --max-gum-fps=17.5
--max-web-media-player-countSets the maximum number of WebMediaPlayers allowed per frame.
--message-loop-type-uiIndicates the utility process should run with a message loop type of UI.
--metal--metal-null--metrics-client-idThis is how the metrics client ID is passed from the browser process to its children. With Crashpad, the metrics client ID is distinct from the crash client ID.
--metrics-shmem-handleHandle to the shared memory segment a child process should use to transmit histograms back to the browser process.
--mock--mojo-local-storageUse a Mojo-based LocalStorage implementation.
--monitoring-destination-idAllows setting a different destination ID for connection-monitoring GCM messages. Useful when running against a non-prod management server.
--mse-audio-buffer-size-limit-mbAllows explicitly specifying MSE audio/video buffer sizes as megabytes. Default values are 150M for video and 12M for audio.
--mse-video-buffer-size-limit-mb--mute-audioMutes audio sent to the audio device so it is not audible during automated testing.
--native--native-messaging-connect-hostRequests a native messaging connection be established between the native messaging host named by this switch and the extension with ID specified by kNativeMessagingConnectExtension.
--native-messaging-connect-idIf set when kNativeMessagingConnectHost and kNativeMessagingConnectExtension are specified, is reflected to the native messaging host as a command line parameter.
--net-log-capture-modeSets the granularity of events to capture in the network log. The mode can be set to one of the following values: "Default" "IncludeSensitive" "Everything" See the enums of the corresponding name in net_log_capture_mode.h for a description of their meanings.
--net-log-durationSpecifies the duration (in seconds) for network logging. When this flag is provided with a positive integer value X, Chrome will automatically stop collecting NetLog events after X seconds and flush the log to disk.
--net-log-max-size-mbSets the maximum size, in megabytes. The log file can grow to before older data is overwritten. Do not use this flag if you want an unlimited file size.
--new-windowLaunches URL in new browser window.
--no-default-browser-checkDisables the default browser check. Useful for UI/browser tests where we want to avoid having the default browser info-bar displayed.
--no-experimentsDisables all experiments set on about:flags. Does not disable about:flags itself. Useful if an experiment makes chrome crash at startup: One can start chrome with --no-experiments, disable the problematic lab at about:flags and then restart chrome without this switch again.
--no-first-runSkip First Run tasks as well as not showing additional dialogs, prompts or bubbles. Suppressing dialogs, prompts, and bubbles is important as this switch is used by automation (including performance benchmarks) where it's important only a browser window is shown. This may not actually be the first run or the What's New page. Its effect can be partially ignored by adding kForceFirstRun (for FRE), kForceWhatsNew (for What's New) and/or kIgnoreNoFirstRunForSearchEngineChoiceScreen (for the DSE choice screen). This does not drop the First Run sentinel and thus doesn't prevent first run from occurring the next time chrome is launched without this flag. It also does not update the last What's New milestone, so does not prevent What's New from occurring the next time chrome is launched without this flag.
--no-network-profile-warningWhether or not the browser should warn if the profile is on a network share. This flag is only relevant for Windows currently.
--no-pingsDon't send hyperlink auditing pings
--no-pre-read-main-dllWhether this process should PrefetchVirtualMemory on the contents of Chrome.dll. This warms up the pages in memory to speed up startup but might not be required in later renderers and/or GPU. For experiment info see crbug.com/1350257.
--no-proxy-serverDon't use a proxy server, always make direct connections. Overrides any other proxy server flags that are passed.
--no-sandboxDisables the sandbox for all process types that are normally sandboxed. Meant to be used as a browser-level switch for testing purposes only.
--no-service-autorunDisables the service process from adding itself as an autorun process. This does not delete existing autorun registrations, it just prevents the service from registering a new one.
--no-startup-windowDoes not automatically open a browser window on startup (used when launching Chrome for the purpose of hosting background apps).
--no-unsandboxed-zygoteDisables the unsandboxed zygote. Note: this flag should not be used on most platforms. It is introduced because some platforms (e.g. Cast) have very limited memory and binaries won't be updated when the browser process is running.
--no-user-gesture-requiredAutoplay policy that does not require any user gesture.
--no-xr-runtimeTell WebXr to assume that it does not support any runtimes.
--no-xshmDisables MIT-SHM extension. In use only with Ozone/X11.
--no-zygoteDisables the use of a zygote process for forking child processes. Instead, child processes will be forked and exec'd directly. Note that --no-sandbox should also be used together with this flag because the sandbox needs the zygote to work.
--no-zygote-sandboxInstructs the zygote to launch without a sandbox. Processes forked from this type of zygote will apply their own custom sandboxes later.
--noerrdialogsSuppresses all error dialogs when present.
--none--notification-inline-replyUsed in combination with kNotificationLaunchId to specify the inline reply entered in the toast in the Windows Action Center.
--notification-launch-idUsed for launching Chrome when a toast displayed in the Windows Action Center has been activated. Should contain the launch ID encoded by Chrome.
--null--num-raster-threadsControls the number of threads to use for raster tasks.
--nv12Swap chain formats for direct composition SDR video overlays.
--offer-in-settings--opengraph--openxr--orientation-sensors--override-language-detectionOverrides the language detection result determined based on the page contents.
--ozone-dump-fileSpecify location for image dumps.
--ozone-override-screen-sizeSpecifies ozone screen size.
--ozone-platformSpecify ozone platform implementation to use.
--p010--pack-extensionPackages an extension to a .crx installable file from a given directory.
--pack-extension-keyOptional PEM private key to use in signing packaged .crx.
--package-name--package-version-name--passthrough--pdf-rendererRenderer process that runs the non-PPAPI PDF plugin.
--pen-devicesTells chrome to interpret events from these devices as pen events. Only available with XInput 2 (i.e. X server 1.8 or above). The id's of the devices can be retrieved from 'xinput list'.
--policySet policy value by command line.
--policy-verification-keyReplace the original verification_key with the one provided by the command line flag. Can be used only for unit tests or browser tests.
--pre-crashpad-crash-testCauses the browser process to crash very early in startup, just before crashpad (or breakpad) is initialized.
--preinstalled-web-apps-dirA directory where Chrome looks for json files describing default/preinstalled web apps. This overrides any default directory to load preinstalled web apps from.
--privet-ipv6-onlyUse IPv6 only for privet HTTP.
--process-per-siteEnable the "Process Per Site" process model for all domains. This mode consolidates same-site pages so that they share a single process. More details here: - https://www.chromium.org/developers/design-documents/process-models - The class comment in site_instance.h, listing the supported process models. IMPORTANT: This isn't to be confused with --site-per-process (which is about isolation, not consolidation). You probably want the other one.
--process-per-tabRuns each set of script-connected tabs (i.e., a BrowsingInstance) in its own renderer process. We default to using a renderer process for each site instance (i.e., group of pages from the same registered domain with script connections to each other). "unnecessary" process swaps for cross-site navigations but still swap when needed for security (e.g., isolated origins).
--product-versionOutputs the product version information and quit. Used as an internal api to detect the installed version of Chrome on Linux.
--profile-base-namePass the basename of the profile directory to specify which profile to get information. Only relevant when used with --list-apps switch.
--profile-directorySelects directory of profile to associate with the first browser launched.
--profile-emailLike kProfileDirectory, but selects the profile by email address. If the email is not found in any existing profile, this switch has no effect. If both kProfileDirectory and kProfileEmail are specified, kProfileDirectory takes priority.
--profile-management-attributesDomains and associated SAML attributes for which third-party profile management should be enabled. Input should be in JSON format.
--profiling-at-startStarts the sampling based profiler for the browser process at startup. This will only work if chrome has been built with the gn arg enable_profiling = true. The output will go to the value of kProfilingFile.
--profiling-fileSpecifies a location for profiling output. This will only work if chrome has been built with the gyp variable profiling=1 or gn arg enable_profiling=true. {pid} if present will be replaced by the pid of the process. {count} if present will be incremented each time a profile is generated for this process. The default is chrome-profile-{pid} for the browser and test-profile-{pid} for tests.
--profiling-flushControls whether profile data is periodically flushed to a file. Normally the data gets written on exit but cases exist where chromium doesn't exit cleanly (especially when using single-process). A time in seconds can be specified.
--proxy-auto-detectForces proxy auto-detection.
--proxy-bypass-listSpecifies a list of hosts for whom we bypass proxy settings and use direct connections. Ignored if --proxy-auto-detect or --no-proxy-server are also specified. This is a comma-separated list of bypass rules. See: "net/proxy_resolution/proxy_host_matching_rules.h" for the format of these rules.
--proxy-pac-urlUses the pac script at the given URL
--proxy-serverUses a specified proxy server, overrides system settings.
--pseudonymization-salt-handleHandle to shared memory containing the pseudonymization salt, passed to child processes at launch. See https://crbug.com/40850085.
--pull-to-refreshEnables or disables pull-to-refresh gesture in response to vertical overscroll. Set the value to '0' to disable the feature, set to '1' to enable it for both touchpad and touchscreen, and set to '2' to enable it only for touchscreen. Defaults to disabled.
--pwa-launcher-versionSpecifies the version of the Progressive-Web-App launcher that launched Chrome, used to determine whether to update all launchers. method, as older PWA launchers still using this switch will rely on Chrome to update them to use the new method.
--raise-timer-frequencyRaise the timer interrupt frequency in all Chrome processes, for experimental purposes. This feature is needed because as of Windows 10 2004 the scheduling effects of changing the timer interrupt frequency are not global, and this lets us prove/disprove whether this matters. See https://crbug.com/1128917
--reader-mode-feedback--reader-mode-heuristics--realtime-reporting-urlSpecifies the URL at which to upload real-time reports.
--reduce-accept-languageReduce the accept-language for HTTP header and JS navigator.languages, and only most preferred language: https://github.com/Tanych/accept-language.
--reduce-accept-language-httpReduce the accept-language for HTTP header, and only send most preferred language in the request header: https://github.com/Tanych/accept-language.
--reduce-user-agent-minor-versionReduce the minor version number in the User-Agent string. This flag implements phase 4 of User-Agent reduction: https://blog.chromium.org/2021/09/user-agent-reduction-origin-trial-and-dates.html.
--refresh-platform-policyForces immediate platform policy refresh (not cloud policy) when Chrome is already running. The switch prevents a new browser window from opening and only triggers the policy refresh. Useful for testing and automation to avoid waiting for the next scheduled refresh interval. No-op if Chrome is not already running.
--relauncherA process type (switches::kProcessType) that relaunches the browser. See chrome/browser/mac/relauncher.h.
--remote-allow-originsEnables web socket connections from the specified origins only. '*' allows any origin.
--remote-debugging-io-pipesSpecifies pipe names for the incoming and outbound messages on the Windows platform. This is a comma separated list of two pipe handles serialized as unsigned integers, e.g. "--remote-debugging-io-pipes=3,4".
--remote-debugging-pipeEnables remote debug over stdio pipes [in=3, out=4] or over the remote pipes specified in the 'remote-debugging-io-pipes' switch. Optionally, specifies the format for the protocol messages, can be either "JSON" (the default) or "CBOR".
--remote-debugging-portEnables remote debug over HTTP on the specified port.
--remote-debugging-socket-nameEnables remote debug over HTTP on the specified socket name.
--remote-debugging-targetsProvides a list of addresses to discover DevTools remote debugging targets. The format is <host>:<port>,...,<host>:port.
--render-node-overrideAllows explicitly picking a DRM render node to create gbm_device for rendering.
--rendererCauses the process to run as renderer instead of as browser.
--renderer-client-id--renderer-cmd-prefixThe contents of this flag are prepended to the renderer command line. Useful values might be "valgrind" or "xterm -e gdb --args".
--renderer-process-limitOverrides the default/calculated limit to the number of renderer processes. Very high values for this setting can lead to high memory/resource usage or instability.
--renderer-startup-dialogCauses the renderer process to display a dialog on launch. Passing this flag also adds sandbox::policy::kNoSandbox on Windows non-official builds, since that's needed to show a dialog.
--renderer-wait-for-java-debuggerBlock ChildProcessMain thread of the renderer's ChildProcessService until a Java debugger is attached.
--renderpassParameters for kUIShowCompositedLayerBorders.
--repair-all-valid-extensionsIndicates that all corrupted extensions should be repaired if they are are enabled by policy. This is mainly used after a user data downgrade.
--request-desktop-sitesForce enable user agent overrides to request desktop sites in Clank.
--restartIndicates that Chrome was restarted (e.g., after a flag change). This is used to ignore the launch when recording the Launch.Mode2 metric.
--restore-last-sessionIndicates the last session should be restored on startup. This overrides the preferences value. Note that this does not force automatic session restore following a crash, so as to prevent a crash loop. This switch is used to implement support for OS-specific "continue where you left off" functionality on OS X and Windows.
--run-manualManual tests only run when --run-manual is specified. This allows writing tests that don't run automatically but are still in the same test binary. This is useful so that a team that wants to run a few tests doesn't have to add a new binary that must be compiled on all builds.
--same-tabIndicates that the URL in the command line should open in the active tab instead of a new tab. In case of multiple URLS given as arguments, the first one will replace the active tab.
--sandbox-ipcCauses the process to run as a sandbox IPC subprocess.
--save-page-as-mhtmlEnables saving webpages as MHTML (Webpage, Single) by default, instead of saving as HTML with a directory of sub-resources. (Webpage, Complete). See http://crbug.com/40179885 for how to remove this switch.
--scheduler-boost-urgentOverride the default scheduling boosting value for urgent tasks. This can be adjusted if a specific chromeos device shows better perf/power ratio (e.g. by running video conference tests). Currently, this values directs to linux scheduler's utilization min clamp. Range is 0(no biased load) ~ 100(mamximum load value).
--screen-configSpecifies the initial screen configuration, or state of all displays, for FakeDisplayDelegate, see class for format details.
--screen-infoHeadless screen info in the format: {0,0 800x600}{800,0 600x800}. See //components/headless/screen_info/README.md for more details.
--secondary-display-layoutSpecifies the layout mode and offsets for the secondary display for testing. The format is "<t|r|b|l>,<offset>" where t=TOP, r=RIGHT, b=BOTTOM and L=LEFT. For example, 'r,-100' means the secondary display is positioned on the right with -100 offset. (above than primary)
--secure-connect-api-urlSpecifies the base URL to contact the secure connect Api.
--service-sandbox-typeType of sandbox to apply to the process running the service, one of the values in the next block.
--shared-filesDescribes the file descriptors passed to a child process in the following list format: <file_id>:<descriptor_id>,<file_id>:<descriptor_id>,... where <file_id> is an ID string from the manifest of the service being launched and <descriptor_id> is the numeric identifier of the descriptor for the child process can use to retrieve the file descriptor from the global descriptor table.
--show-autofill-signaturesAnnotates forms and fields with Autofill signatures.
--show-autofill-type-predictionsAnnotates forms with Autofill field type predictions.
--show-composited-layer-bordersRenders a border around compositor layers to help debug and study layer compositing.
--show-fps-counterDraws a heads-up-display showing Frames Per Second as well as GPU memory usage. If you also use --enable-logging=stderr --vmodule="head*=1" then FPS will also be output to the console log.
--show-iconsSee kHideIcons.
--show-layer-animation-boundsRenders a border that represents the bounding box for the layer's animation.
--show-property-changed-rectsShow rects in the HUD around layers whose properties have changed.
--show-screenspace-rectsShow rects in the HUD around the screen-space transformed bounds of every layer.
--show-surface-damage-rectsShow rects in the HUD around damage as it is recorded into each render surface.
--silent-debugger-extension-apiDoes not show an infobar when an extension attaches to a page using chrome.debugger page. Required to attach to extension background pages.
--silent-launchCauses Chrome to launch without opening any windows by default. Useful if one wishes to use Chrome as an ash server.
--simulate-browsing-data-lifetimeSets the BrowsingDataLifetime policy to a very short value (shorter than normally possible) for testing purposes.
--simulate-critical-updateSimulates a critical update being available.
--simulate-elevated-recoverySimulates that elevation is needed to recover upgrade channel.
--simulate-idle-timeoutSets the IdleTimeout policy to a very short value (shorter than normally possible) for testing purposes.
--simulate-outdatedSimulates that current version is outdated.
--simulate-outdated-no-auSimulates that current version is outdated and auto-update is off.
--simulate-upgradeSimulates an update being available.
--single-processRuns the renderer and plugins in the same process as the browser
--site-per-processEnforces a one-site-per-process security policy: * Each renderer process, for its whole lifetime, is dedicated to rendering pages for just one site. * Thus, pages from different sites are never in the same process. * A renderer process's access rights are restricted based on its site. * All cross-site navigations force process swaps. * <iframe>s are rendered out-of-process whenever the src= is cross-site. More details here: - https://www.chromium.org/developers/design-documents/site-isolation - https://www.chromium.org/developers/design-documents/process-models - The class comment in site_instance.h, listing the supported process models. IMPORTANT: this isn't to be confused with --process-per-site (which is about process consolidation, not isolation). You probably want this one.
--skia-font-cache-limit-mbSpecifies the max number of bytes that should be used by the skia font cache. If the cache needs to allocate more, skia will purge previous entries.
--skia-graphite-backendSpecify which backend to use for Skia Graphite - "dawn" (default) or "metal" (only allowed on non-official developer builds).
--skia-resource-cache-limit-mbSpecifies the max number of bytes that should be used by the skia resource cache. The previous entries are purged from the cache when the memory useage exceeds this limit.
--slow-down-raster-scale-factorRe-rasters everything multiple times to simulate a much slower machine. Give a scale factor to cause raster to take that many times longer to complete, such as --slow-down-raster-scale-factor=25.
--source-app-idWhen rendezvousing with an existing process, used to indicate that the StartupInfoW of the new Chrome process had dwFlags == STARTF_TITLEISAPPID. This is used to record launch metrics.
--source-shortcutWhen rendezvousing with an existing process, used to pass the path of the shortcut that launched the new Chrome process. This is used to record launch metrics.
--ssl-key-log-fileCauses SSL key material to be logged to the specified file for debugging purposes. See https://developer.mozilla.org/en-US/docs/Mozilla/Projects/NSS/Key_Log_Format for the format.
--ssl-version-maxSpecifies the maximum SSL/TLS version ("tls1.2" or "tls1.3").
--ssl-version-minSpecifies the minimum SSL/TLS version ("tls1.2" or "tls1.3").
--start-fullscreenSpecifies if the browser should start in fullscreen mode, like if the user had pressed F11 right after startup.
--start-maximizedStarts the browser maximized, regardless of any previous settings.
--start-stack-profilerStarts the stack sampling profiler in the child process.
--startup-foreground-launchIdentifies Chrome instances that start in foreground mode at startup to record related metrics.
--stub--suppress-performance-logsSuppresses GL_DEBUG_TYPE_PERFORMANCE log messages for web tests that can get sent to the JS console and cause unnecessary test failures due test output log expectation comparisons.
--surface--swiftshader--swiftshader-webgl--system-aec-enabledEnforce system audio echo cancellation.
--system-log-upload-frequencyFrequency in Milliseconds for system log uploads. Should only be used for testing purposes.
--target-device-scale-for-testingAllows web tests to specify the target device scale for the test cases.
--test-child-processWhen running certain tests that spawn child processes, this switch indicates to the test framework that the current process is a child process.
--test-gl-libFlag used for Linux tests: for desktop GL bindings, try to load this GL library first, but fall back to regular library if loading fails.
--test-memory-log-delay-in-minutesCustom delay for memory log. This should be used only for testing purpose.
--test-namePasses the name of the current running automated test to Chrome.
--test-third-party-cookie-phaseout--test-typeType of the current test harness ("browser" or "ui" or "gpu").
--time-ticks-at-unix-epochAccepts a number representing the time-ticks value at the Unix epoch. Since different processes can produce a different value for this due to system clock changes, this allows synchronizing them to a single value.
--tint-dc-layerTint `SwapChainPresenter` with the following colors: - Decode swap chain: blue - VP blit: magenta - VP blit w/ staging texture: orange - MF proxy surface: green This is similar to `HKLM\Software\Microsoft\Windows\DWM` `OverlayTestMode=1` in DWM, but to help understand `SwapChainPresenter` state.
--top-controls-hide-thresholdPercentage of the browser controls need to be hidden before they will auto hide.
--top-controls-show-thresholdPercentage of the browser controls need to be shown before they will auto show.
--touch-devicesTells chrome to interpret events from these devices as touch events. Only available with XInput 2 (i.e. X server 1.8 or above). The id's of the devices can be retrieved from 'xinput list'.
--touch-eventsEnable support for touch event feature detection.
--touch-slop-distanceOverrides touch slop distance for gesture detection. The touch slop distance is the maximum distance from the starting point of a touch sequence that a gesture can travel before it can no longer be considered a tap. Scroll gestures can only begin after this distance has been travelled. The switch value is a floating point number that is interpreted as a distance in pixels.
--trace-to-fileSends trace events from these categories to a file. --trace-to-file on its own sends to default categories.
--trace-to-file-nameSpecifies the file name for --trace-to-file. If unspecified, it will go to a default file name.
--translate-ranker-model-urlOverrides the URL from which the translate ranker model is downloaded.
--translate-script-urlOverrides the default server used for Google Translate.
--translate-security-originOverrides security-origin with which Translate runs in an isolated world.
--trusted-download-sourcesIdentifies a list of download sources as trusted, but only if proper group policy is set.
--try-supported-channel-layoutsInstead of always using the hardware channel layout, check if a driver supports the source channel layout. Avoids outputting empty channels and permits drivers to enable stereo to multichannel expansion. Kept behind a flag since some drivers lie about supported layouts and hang when used. See http://crbug.com/259165 for more details.
--typeThe value of this switch determines whether the process is started as a renderer or plugin host. If it's empty, it's the browser.
--ui-disable-zero-copy--ui-enable-rgba-4444-textures--ui-enable-zero-copy--ui-show-composited-layer-borders--ui-show-fps-counter--ui-show-layer-animation-bounds--ui-show-paint-rects--ui-show-property-changed-rects--ui-show-screenspace-rects--ui-show-surface-damage-rects--ui-slow-animations--uninstallRuns un-installation steps that were done by chrome first-run.
--uninstall-app-idSpecifies that the WebApp with the specified id should be uninstalled.
--unique-temp-dir-suffixWhen switches::kProcessType is switches::kCodeSignCloneCleanupProcess this switch is required. The value must be the unique suffix portion of the temporary directory that contains the clone. The full path will be reconstructed by the cleanup process.
--unlimited-storageOverrides per-origin quota settings to unlimited storage for any apps/origins. This should be used only for testing purpose.
--use-adapter-luidInitialize the GPU process using the adapter with the specified LUID. This is only used on Windows, as LUID is a Windows specific structure.
--use-angleSelect which ANGLE backend to use. Options are: default: Attempts several ANGLE renderers until one successfully initializes, varying ES support by platform. d3d9: Legacy D3D9 renderer, ES2 only. d3d11: D3D11 renderer, ES2 and ES3. warp: D3D11 renderer using software rasterization, ES2 and ES3. gl: Desktop GL renderer, ES2 and ES3. gles: GLES renderer, ES2 and ES3.
--use-cmd-decoderUse the Pass-through command decoder, skipping all validation and state tracking. Switch lives in ui/gl because it affects the GL binding initialization on platforms that would otherwise not default to using EGL bindings.
--use-context-snapshotSwitch supplied to the renderer if the feature `kUseContextSnapshot` is enabled. A switch is used as at the time the renderer needs this information features have not yet been loaded.
--use-crasUse CRAS, the ChromeOS audio server.
--use-fake-device-for-media-streamUse fake device for Media Stream to replace actual camera and microphone. For the list of allowed parameters, see FakeVideoCaptureDeviceFactory::ParseFakeDevicesConfigFromOptionsString().
--use-fake-ui-for-digital-identityBypass the digital-identity-credential OS call. Simulate the user accepting the OS-presented dialog.
--use-fake-ui-for-fedcmBypass the FedCM account selection dialog. If a value is provided for this switch, that account ID is selected, otherwise the first account is chosen.
--use-fake-ui-for-media-streamBypass the media stream infobar by selecting the default device for media streams (e.g. WebRTC). Works with --use-fake-device-for-media-stream. Prefer --auto-accept-camera-and-microphone-capture which does not interact with screen/tab capture.
--use-file-for-fake-audio-capturePlay a .wav file as the microphone. Note that for WebRTC calls we'll treat the bits as if they came from the microphone, which means you should disable audio processing (lest your audio file will play back distorted). The input file is converted to suit Chrome's audio buses if necessary, so most sane .wav files should work. You can pass either <path> to play the file looping or <path>%noloop to stop after playing the file to completion. Must also be used with kDisableAudioInput or kUseFakeDeviceForMediaStream.
--use-file-for-fake-video-captureUse an .y4m file to play as the webcam. See the comments in media/capture/video/file_video_capture_device.h for more details.
--use-first-display-as-internalUses the 1st display in --ash-host-window-bounds as internal display. This is for debugging on linux desktop.
--use-first-party-setAllows the manual specification of a First-Party Set. The format is the same as that of `--use-related-website-set`. DEPRECATED(crbug.com/1486689): This switch is under deprecation due to renaming "First-Party Set" to "Related Website Set". Please use `kUseRelatedWebsiteSet` instead.
--use-glSelect which implementation of GL the GPU process should use. Options are: desktop: whatever desktop OpenGL the user has installed (Linux and Mac default). egl: whatever EGL / GLES2 the user has installed (Windows default - actually ANGLE). swiftshader: The SwiftShader software renderer.
--use-gpu-in-testsUse hardware gpu, if available, for tests.
--use-redist-dmlTry to use a redistributable DirectML.dll. Used for testing WebNN against newer DirectML release before it is integrated into Windows OS. Please see more info about DirectML releases at: https://learn.microsoft.com/en-us/windows/ai/directml/dml-version-history
--use-related-website-setAllows the manual specification of a Related Website Set. The set should be provided as a stringified JSON object, whose format matches the format of the JSON in https://github.com/GoogleChrome/related-website-sets.
--use-system-default-printerUses the system default printer as the initially selected destination in print preview, instead of the most recently used destination.
--use-system-proxy-resolverUses WinHttp to resolve proxies instead of using Chromium's normal proxy resolution logic. This is only supported in Windows. exclusively using system proxy configs.
--use-vulkanEnable Vulkan support and select Vulkan implementation, must also have ENABLE_VULKAN defined. This only initializes Vulkan, the flag --enable-features=Vulkan must also be used to select Vulkan for compositing and rasterization.
--use-wayland-explicit-grabUse explicit grab when opening popup windows. See https://crbug.com/1220274
--use-webgpu-adapterThe adapter to use for WebGPU content.
--use-webgpu-power-preferenceThe adapter selecting strategy related to GPUPowerPreference.
--user-data-dirSpecifies the user data directory, which is where the browser will look for all of its state.
--user-data-migratedIndicates that this process is the product of a relaunch following migration of User Data.
--user-gesture-requiredAutoplay policy to require a user gesture in order to play.
--utilityCauses the process to run as a utility subprocess.
--utility-cmd-prefixThe contents of this flag are prepended to the utility process command line. Useful values might be "valgrind" or "xterm -e gdb --args".
--utility-startup-dialogCauses the utility process to display a dialog on launch.
--utility-sub-typeThis switch indicates the type of a utility process. It does not affect the services offered by the process, but is added to the command line to make it easier to identify the purpose of the utility process.
--v8-cache-optionsSet options to cache V8 data. (none, code, or default)
--validate-crxExamines a .crx for validity and prints the result.
--validatingThe command decoder names that can be passed to --use-cmd-decoder.
--variations-insecure-server-urlSpecifies a custom URL for the server to use as an insecure fallback when requests to |kVariationsServerURL| fail. Requests to this URL will be encrypted.
--variations-override-countryAllows overriding the country used for evaluating variations. This is similar to the "Override Variations Country" entry on chrome://translate-internals, but is exposed as a command-line flag to allow testing First Run scenarios. Additionally, unlike chrome://translate-internals, the value isn't persisted across sessions.
--variations-seed-corpusSpecifies the value of the variations seed corpus to use. When unspecified or unrecognized, the default corpus will be used.
--variations-seed-fetch-intervalOverride the time interval between each variation seed fetches. Unit is in minutes. The minimum is 1 minute. The default is 30 minutes.
--variations-seed-versionUsed to share variations seed version with child processes.
--variations-server-urlSpecifies a custom URL for the server which reports variation data to the client. Specifying this switch enables the Variations service on unofficial builds. See variations_service.cc.
--variations-state-fileUse features defined in the value. Use this flag to reproduce experiments related issues. Copy 'Command-line Variations' value from chrome://version page. Save it to a file and pass it via this flag. The value is a base64 encoded JSON format produced by `variations::VariationsCommandLine::WriteToString`.
--variations-test-seed-pathSpecifies the location of a seed file for Local State's seed to be populated from. The seed file must be in json format with the keys |kVariationsCompressedSeed| and |kVariationsSeedSignature|.
--versionPrints version information and quits.
--video-image-texture-targetTexture target for CHROMIUM_image backed video frame textures.
--video-threadsSet number of threads to use for video decoding.
--vmoduleGives the per-module maximal V-logging levels to override the value given by --v. E.g. "my_module=2,foo*=3" would change the logging level for all code in source files "my_module.*" and "foo*.*" ("-inl" suffixes are also disregarded for this matching). Any pattern containing a forward or backward slash will be tested against the whole pathname and not just the module. E.g., "*/foo/bar/*=2" would change the logging level for all code in source files under a "foo/bar" directory.
--vulkan--vulkan-heap-memory-limit-mbSpecifies the heap limit for Vulkan memory.
--vulkan-null--vulkan-sync-cpu-memory-limit-mbSpecifies the sync CPU limit for total Vulkan memory.
--wait-for-debuggerWill wait for 60 seconds for a debugger to come to attach to the process.
--wait-for-debugger-childrenWill add kWaitForDebugger to every child processes. If a value is passed, it will be used as a filter to determine if the child process should have the kWaitForDebugger flag passed on or not.
--wait-for-debugger-on-navigationOn every navigation a message with the renderer's URL will be logged and the renderer will wait for a debugger to be attached or SIGUSR1 to be sent to continue execution.
--wait-for-debugger-webuiFlag used by WebUI test runners to wait for debugger to be attached.
--wallet-service-use-sandboxUse the sandbox Online Wallet service URL (for developer testing).
--waveout-buffersNumber of buffers to use for WaveOut.
--wayland-text-input-versionSpecify wayland text-input protocol version. Defaults to "1" for text-input-v1. Can specify value "3" for experimental text-input-v3 support.
--web-otp-backendEnables specified backend for the Web OTP API.
--web-otp-backend-autoEnables auto backend selection for Web OTP API.
--web-otp-backend-sms-verificationEnables Sms Verification backend for Web OTP API which requires app hash in SMS body.
--web-otp-backend-user-consentEnables User Consent backend for Web OTP API.
--web-settings-for-testingAllows web tests to specify additional web settings for the test cases.
--webapk-server-urlCustom WebAPK server URL for the sake of testing.
--webauthn-remote-desktop-supportAllows trusted remote desktop clients to make WebAuthn requests on behalf of other origins. This switch only controls availability of the `remoteDesktopClientOverride` extension but doesn't by itself enable any origin to use it.
--webgl-antialiasing-modeSet the antialiasing method used for webgl. (none, explicit, implicit)
--webgl-msaa-sample-countSet a default sample count for webgl if msaa is enabled.
--webrtc-event-loggingEnable capture and local storage of WebRTC event logs without visiting chrome://webrtc-internals. This is useful for automated testing. It accepts the path to which the local logs would be stored. Disabling is not possible without restarting the browser and relaunching without this flag.
--webrtc-ip-handling-policyOverride WebRTC IP handling policy to mimic the behavior when WebRTC IP handling policy is specified in Preferences.
--webview-draw-functor-uses-vulkanIndicate that the this is being used by Android WebView and its draw functor is using vulkan.
--whats-new-use-stagingForce What's New on Desktop to request from the staging environment.
--win-jumplist-actionSpecifies which category option was clicked in the Windows Jumplist that resulted in a browser startup.
--window-nameSpecify the initial window user title: --window-name="My custom title"
--window-positionSpecify the initial window position: --window-position=x,y
--window-sizeSpecify the initial window size: --window-size=w,h
--window-workspaceSpecify the initial window workspace: --window-workspace=id
--winhttp-proxy-resolverUses WinHTTP to fetch and evaluate PAC scripts. Otherwise the default is to use Chromium's network stack to fetch, and V8 to evaluate.
--yuy2--zygoteCauses the process to run as a zygote.
--zygote-cmd-prefixThe prefix used when starting the zygote process. (i.e. 'gdb --args')