Devtools Service

A WebdriverIO service that allows you to run Chrome DevTools commands in your tests

With Chrome v63 and up the browser started to support multi clients allowing arbitrary clients to access the Chrome DevTools Protocol. This provides interesting opportunities to automate Chrome beyond the WebDriver protocol. With this service you can enhance the wdio browser object to leverage that access and call Chrome DevTools commands within your tests to e.g. intercept requests, throttle network capabilities or take CSS/JS coverage.

Note: this service currently only supports Chrome v63 and up, and Chromium (Microsoft Edge is not yet supported)!


The easiest way is to keep @wdio/devtools-service as a devDependency in your package.json.

"devDependencies": {
"@wdio/devtools-service": "^6.3.7"

You can simple do it by:

npm install @wdio/devtools-service --save-dev

Instructions on how to install WebdriverIO can be found here.


In order to use the service you just need to add the service to your service list in your wdio.conf.js, like:

// wdio.conf.js
export.config = {
// ...
services: ['devtools'],
// ...


The @wdio/devtools-service offers you a variety of features that helps you to automate Chrome beyond the WebDriver protocol. It gives you access to the Chrome DevTools protocol as well as to a Puppeteer instance that you can use to automate Chrome with the Puppeteer automation interface.

Performance Testing#

The DevTools service allows you to capture performance data from every page load or page transition that was caused by a click. To enable it call browser.enablePerformanceAudits(<options>). After you are done capturing all necessary performance data disable it to revert the throttling settings, e.g.:

const assert = require('assert')
describe(' page', () => {
before(() => {
it('should load within performance budget', () => {
* this page load will take a bit longer as the DevTools service will
* capture all metrics in the background
let metrics = browser.getMetrics()
assert.ok(metrics.speedIndex < 1500) // check that speedIndex is below 1.5ms
let score = browser.getPerformanceScore() // get Lighthouse Performance score
assert.ok(score >= .99) // Lighthouse Performance score is at 99% or higher
metrics = browser.getMetrics()
assert.ok(metrics.speedIndex < 1500)
score = browser.getPerformanceScore()
assert.ok(score >= .99)
after(() => {

You can emulate a mobile device by using the emulateDevice command, throttling CPU and network as well as setting mobile as form factor:

browser.emulateDevice('iPhone X')
networkThrottling: 'Good 3G',
cpuThrottling: 4,
formFactor: 'mobile'

The following commands with their results are available:


Get most common used performance metrics.

* { estimatedInputLatency: 16,
* timeToFirstByte: 566,
* serverResponseTime: 566,
* domContentLoaded: 3397,
* firstVisualChange: 2610,
* firstPaint: 2822,
* firstContentfulPaint: 2822,
* firstMeaningfulPaint: 2822,
* largestContentfulPaint: 2822,
* lastVisualChange: 15572,
* firstCPUIdle: 6135,
* firstInteractive: 6135,
* load: 8429,
* speedIndex: 3259,
* totalBlockingTime: 31,
* cumulativeLayoutShift: 2822 }


Get some useful diagnostics about the page load.

* { numRequests: 8,
* numScripts: 0,
* numStylesheets: 0,
* numFonts: 0,
* numTasks: 237,
* numTasksOver10ms: 5,
* numTasksOver25ms: 2,
* numTasksOver50ms: 2,
* numTasksOver100ms: 0,
* numTasksOver500ms: 0,
* rtt: 147.20600000000002,
* throughput: 47729.68474448835,
* maxRtt: 176.085,
* maxServerLatency: 1016.813,
* totalByteWeight: 62929,
* totalTaskTime: 254.07899999999978,
* mainDocumentTransferSize: 8023 }


Returns a list with a breakdown of all main thread task and their total duration.

* [ { group: 'styleLayout', duration: 130.59099999999998 },
* { group: 'other', duration: 44.819 },
* { group: 'paintCompositeRender', duration: 13.732000000000005 },
* { group: 'parseHTML', duration: 3.9080000000000004 },
* { group: 'scriptEvaluation', duration: 2.437999999999999 },
* { group: 'scriptParseCompile', duration: 0.20800000000000002 } ]


Returns the Lighthouse Performance Score which is a weighted mean of the following metrics: firstContentfulPaint, speedIndex, largestContentfulPaint, cumulativeLayoutShift, totalBlockingTime, firstInteractive.

* 0.897826278457836


Enables auto performance audits for all page loads that are cause by calling the url command or clicking on a link or anything that causes a page load. You can pass in a config object to determine some throttling options. The default throttling profile is Good 3G network with a 4x CPU trottling.

networkThrottling: 'Good 3G',
cpuThrottling: 4,
cacheEnabled: true,
formFactor: 'mobile'

The following network throttling profiles are available: offline, GPRS, Regular 2G, Good 2G, Regular 3G, Good 3G, Regular 4G, DSL, Wifi and online (no throttling).

Device Emulation#

The service allows you to emulate a specific device type. If set, the browser viewport will be modified to fit the device capabilities as well as the user agent will set according to the device user agent. To set a predefined device profile you can run:

browser.emulateDevice('iPhone X')
// or `browser.emulateDevice('iPhone X', true)` if you want to be in landscape mode

Available predefined device profiles are: Blackberry PlayBook, BlackBerry Z30, Galaxy Note 3, Galaxy Note II, Galaxy S III, Galaxy S5, iPad, iPad Mini, iPad Pro, iPhone 4, iPhone 5, iPhone 6, iPhone 6 Plus, iPhone 7, iPhone 7 Plus, iPhone 8, iPhone 8 Plus, iPhone SE, iPhone X, JioPhone 2, Kindle Fire HDX, LG Optimus L70, Microsoft Lumia 550, Microsoft Lumia 950, Nexus 10, Nexus 4, Nexus 5, Nexus 5X, Nexus 6, Nexus 6P, Nexus 7, Nokia Lumia 520, Nokia N9, Pixel 2, Pixel 2 XL

You can also define your own device profile by providing an object as parameter like in the following example:

viewport: {
width: 550, // <number> page width in pixels.
height: 300, // <number> page height in pixels.
deviceScaleFactor: 1, // <number> Specify device scale factor (can be thought of as dpr). Defaults to 1
isMobile: true, // <boolean> Whether the meta viewport tag is taken into account. Defaults to false
hasTouch: true, // <boolean> Specifies if viewport supports touch events. Defaults to false
isLandscape: true // <boolean> Specifies if viewport is in landscape mode. Defaults to false
userAgent: 'my custom user agent'


This only works if you don't use mobileEmulation within capabilities['goog:chromeOptions']. If mobileEmulation is present the call to browser.emulateDevice() won't do anything.

PWA Testing#

With the checkPWA command you can validate if your webapp is compliant to latest web standards when it comes to progressive web apps. It checks:

  • whether your app is installable
  • provides a service worker
  • has a splash screen
  • provides apple touch and maskable icons
  • can be served on mobile devices

If you are not interested in one of these checks you can pass in a list of checks you like to run. The passed property will return true if all checks pass. If they fail you can use the details property to enrich your failure message with details of the failure.

// open page first
// validate PWA
const result = browser.checkPWA()

Capture Code Coverage#

The service offers you to capture the code coverage of your application under test. To do so you need to enable this feature as part of the service settings:

// wdio.conf.js
services: [
['devtools', {
coverageReporter: {
enable: true,
type: 'html',
logDir: __dirname + '/coverage'

Then you have access to a command that calculates the ratio of covered code lines and branches for you to assert within your test:

const coverage = browser.getCoverageReport()

Chrome DevTools Access#

For now the service allows two different ways to access the Chrome DevTools Protocol:

cdp Command#

The cdp command is a custom command added to the browser scope that allows you to call directly commands to the protocol.

browser.cdp(<domain>, <command>, <arguments>)

For example if you want to get the JavaScript coverage of your page you can do the following:

it('should take JS coverage', () => {
* enable necessary domains
browser.cdp('Profiler', 'enable')
browser.cdp('Debugger', 'enable')
* start test coverage profiler
browser.cdp('Profiler', 'startPreciseCoverage', {
callCount: true,
detailed: true
* capture test coverage
const { result } = browser.cdp('Profiler', 'takePreciseCoverage')
const coverage = result.filter((res) => res.url !== '')

getNodeId(selector) and getNodeIds(selector) Command#

Helper method to get the nodeId of an element in the page. NodeIds are similar like WebDriver node ids an identifier for a node. It can be used as a parameter for other Chrome DevTools methods, e.g. DOM.focus.

const nodeId = browser.getNodeId('body')
console.log(nodeId) // outputs: 4
const nodeId = browser.getNodeIds('img')
console.log(nodeId) // outputs: [ 40, 41, 42, 43, 44, 45 ]

startTracing(categories, samplingFrequency) Command#

Start tracing the browser. You can optionally pass in custom tracing categories (defaults to this list) and the sampling frequency (defaults to 10000).


endTracing Command#

Stop tracing the browser.


getTraceLogs Command#

Returns the tracelogs that was captured within the tracing period. You can use this command to store the trace logs on the file system to analyse the trace via Chrome DevTools interface.

fs.writeFileSync('/path/to/tracelog.json', JSON.stringify(browser.getTraceLogs()))

getPageWeight Command#

Returns page weight information of the last page load.

// outputs:
// { pageWeight: 2438485,
// transferred: 1139136,
// requestCount: 72,
// details: {
// Document: { size: 221705, encoded: 85386, count: 11 },
// Stylesheet: { size: 52712, encoded: 50130, count: 2 },
// Image: { size: 495023, encoded: 482433, count: 36 },
// Script: { size: 1073597, encoded: 322854, count: 15 },
// Font: { size: 84412, encoded: 84412, count: 5 },
// Other: { size: 1790, encoded: 1790, count: 2 },
// XHR: { size: 509246, encoded: 112131, count: 1 } }
// }

Setting Download Paths for the Browser#

The cdp command can be used to call the Page.setDownloadBehavior command of Devtools Protocol to set the behavior when downloading a file. Make sure the downloadPath is an absolute path and the browser.cdp() call is made before the file is downloaded.

browser.cdp('Page', 'setDownloadBehavior', {
behavior: 'allow',
downloadPath: '/home/root/webdriverio-project/',

Access Puppeteer Instance#

The service uses Puppeteer for its automation under the hood. You can get access to the used instance by calling the getPuppeteer command. Note: Puppeteer commands are async and either needs to be called within the call command or handled via async/await:

describe('use Puppeteer', () => {
it('by wrapping commands with call', () => {
const puppeteer = browser.getPuppeteer()
const page = => puppeteer.pages())[0]
console.log( => page.title()))
it('by using async/await', async () => {
const puppeteer = browser.getPuppeteer()
const page = (await puppeteer.pages())[0]
console.log(await page.title())

Event Listener#

In order to capture events in the browser you can register an event listener to a Chrome DevTools event like:

it('should listen on network events', () => {
browser.cdp('Network', 'enable')
browser.on('Network.responseReceived', (params) => {
console.log(`Loaded ${params.response.url}`)

For more information on WebdriverIO see the homepage.