Getting Started

CodeceptJS is a modern end to end testing framework with a special BDD-style syntax. The tests are written as a linear scenario of the user's action on a site.

Feature('CodeceptJS demo')

Scenario('check Welcome page on site', ({ I }) => {
  I.amOnPage('/')
  I.see('Welcome')
})

Tests are written as ES modules using modern JavaScript syntax.

Each test is described inside a Scenario function with the I object passed into it. The I object is an actor, an abstraction for a testing user. The I is a proxy object for currently enabled Helpers.

How It Works

Command Delegation to Helpers

CodeceptJS delegates all test commands to helper backends. Tests written with the I object (the actor) don't directly execute actions. Instead, CodeceptJS routes them through configurable helpers:

Playwright - Chromium, Firefox, WebKit automation
WebDriver - Native browser automation via WebDriver Protocol
Appium - Mobile testing on iOS/Android
Puppeteer - Chromium automation via DevTools Protocol

All helpers share the same API, so it's easy to switch backends. However, due to backend differences and limitations, they aren't guaranteed to be compatible with each other. For example, you can set request headers in Playwright or Puppeteer, but not in WebDriver.

Pick one helper to define how your tests execute. If requirements change, it's straightforward to migrate to another.

Promise Chain & Async/Await

Tests appear synchronous but all actions are wrapped in promises and chained together in a global promise chain. This means:

You usually don't need await for regular actions - commands are automatically queued
Each I.* command is appended to the promise chain
Setup, teardown, and all test steps execute in sequence

// These execute in order WITHOUT await
I.amOnPage('/')
I.click('Login')
I.see('Welcome')

Behind the scenes, this is equivalent to:

Promise.resolve()
  .then(() => I.amOnPage('/'))
  .then(() => I.click('Login'))
  .then(() => I.see('Welcome'))

When You DO Need await:

Use await only with grab actions (methods that retrieve data from the page):

Scenario('use data from page', async ({ I }) => {
  I.fillField('email', '[email protected]')
  I.click('Generate Password')
  // grab actions return data - use await here
  const password = await I.grabTextFrom('#password')
  I.fillField('password', password)
  I.click('Login')
})

Also use await with imported functions and page object methods, as they may contain async operations that aren't part of the promise chain (e.g., await loginPage.login() if it contains I.grab operations inside).

Rule: If an action starts with grab, or if calling an imported function/page object method, you must await it. Regular actions (I.click(), I.fillField(), I.see()) don't need await.

Writing Tests

Tests are written from a user's perspective. There is an actor (represented as I) which contains actions taken from helpers. A test is written as a sequence of actions performed by an actor:

I.amOnPage('/')
I.click('Login')
I.see('Please Login', 'h1')
// ...

A complete test file looks like this:

// suite declaration, like describe() in other frameworks
Feature('User Authentication')

// before each hook
Before(({ I }) => {
  I.amOnPage('/')
})

// a test
Scenario('user can login with valid credentials', ({ I }) => {
  I.click('Login')
  I.fillField('email', '[email protected]')
  I.fillField('password', 'password123')
  I.click('Sign In')
  I.see('Welcome, User')
})

// after each hook
After(({ I }) => {
  // ...
})

CodeceptJS doesn't allow nested suites or multiple suites in one file.

Opening a Page

A test should usually start by navigating the browser to a website.

Start a test by opening a page. Use the I.amOnPage() command for this:

// When "http://site.com" is url in config
I.amOnPage('/') // -> opens http://site.com/
I.amOnPage('/about') // -> opens http://site.com/about
I.amOnPage('https://google.com') // -> https://google.com

When an URL doesn't start with a protocol (http:// or https://) it is considered to be a relative URL and will be appended to the URL which was initially set-up in the config.

It is recommended to use a relative URL and keep the base URL in the config file, so you can easily switch between development, stage, and production environments.

Locating Elements

CodeceptJS supports multiple locating strategies. Most actions accept locators as strings or objects:

// Basic CSS/XPath
I.click('.login-btn')
I.fillField('//input[@type="email"]', '[email protected]')

// Semantic - by text/labels (searches multiple strategies)
I.click('Login')
I.fillField('Email', '[email protected]')

// ARIA locators - by role and accessible name (most reliable)
I.click({ role: 'button', name: 'Login' })
I.fillField({ role: 'textbox', name: 'Email Address' }, '[email protected]')
I.seeElement({ role: 'button', name: 'Submit' })

// Accessibility ID or aria-label
I.click('~login-button')
I.seeElement({ aria: 'Username' })

// Complex locators with locate()
I.click(locate('button').withText('Save').inside('.modal'))
I.seeElement(locate({ role: 'button', name: 'Delete' }).inside('.toolbar'))

// ARIA + Context - combine role with section to target specific area
I.click({ role: 'button', name: 'Save' }, '#detail-panel')
I.fillField({ role: 'textbox', name: 'Title' }, 'New Task', '.modal')
I.seeElement({ role: 'button', name: 'Delete' }, { css: '.toolbar' })
I.click({ role: 'button', name: 'Close' }, '.sidebar')

Best practices:

Use ARIA locators { role: 'button', name: '...' } - resilient to CSS changes and support accessibility
Combine with context when multiple similar elements exist: I.click({ role: 'button', name: 'Delete' }, '.toolbar')
Use semantic CSS classes and IDs when available (e.g., .btn-save, #login-form); avoid style-based names like .bg-green
Use locate() function for complex queries: locate(selector).withText(...).inside(...)

▶ Learn more about using locators in CodeceptJS.

Clicking Elements

Use the locator strategies from Locating Elements section to click any element:

I.click('Login')                              // by text
I.click({ role: 'button', name: 'Save' })   // by ARIA role
I.click('#signup')                            // by ID
I.click('Delete', '.toolbar')                 // with context


click	forceClick	doubleClick	rightClick
forceRightClick	moveCursorTo	dragAndDrop	dragSlider

Use forceClick when standard click fails (e.g., hidden elements, animations). Use rightClick for context menus, doubleClick for multi-select.

Interacting with Forms

Use form methods to interact with inputs, selects, checkboxes, and other form elements. Fields can be located by label, name, CSS, XPath, or aria-label:

// Fill fields - by label, name, CSS, or aria-label
I.fillField('Email', '[email protected]')
I.fillField('My Address', 'Home Sweet Home')   // matches aria-label or aria-labelledby
I.fillField('LoginForm[username]', 'davert')   // by field name attribute
I.fillField('Password', secret('123456'))       // use secret() for sensitive data

// Use context (3rd parameter) to narrow search to specific form
I.fillField('Email', '[email protected]', '#login-form')
I.fillField('Email', '[email protected]', '#registration-form')

// Append or clear
I.appendField('Comments', ' — updated')
I.appendField('Message', ' P.S. Thank you!', '.contact-form')  // with context
I.clearField('#search-input')

// Type into focused field - use when fillField doesn't trigger JS events
I.click('Card Number')
I.type('4111111111111111', 100)  // optional delay in ms between keystrokes

// Selects - pass array for multi-select
I.selectOption('Role', 'Admin')
I.selectOption('Role', 'User', '.user-form')                // with context
I.selectOption('Tags', ['Important', 'Urgent'])

// Checkboxes and radios - supports context
I.checkOption('I Agree to Terms and Conditions')
I.checkOption('Remember me', '#login-form')                 // with context
I.uncheckOption('Subscribe')


fillField	clearField	appendField	type
selectOption	checkOption	uncheckOption	focus
blur

Use secret() for sensitive data: I.fillField('password', secret('123456')) - won't expose in logs.

selectOption works with native <select> elements as well as custom components using role="combobox" or role="listbox".

Assertions

CodeceptJS provides built-in browser assertions instead of generic expect() calls. This keeps tests readable and produces clear failure messages without extra assertion libraries.

I.see(text) - checks that text is visible on the page
I.seeElement(locator) - checks that element exists and is visible in DOM
All assertions have a dontSee / dontSee* counterpart

// Text visibility
I.see('Welcome, Miles')
I.see('Error', '.alert')           // with context
I.dontSee('Loading...')

// Element presence
I.seeElement({ role: 'button', name: 'Submit' })
I.seeElement('.success-message', '#checkout')   // with context
I.dontSeeElement('.error')
I.seeElementInDOM('#hidden-input')              // in DOM but may be invisible

// URL, title, fields
I.seeInCurrentUrl('/dashboard')
I.seeInTitle('My App')
I.seeInField('Email', '[email protected]')
I.seeCheckboxIsChecked('Accept Terms')
I.seeCookie('session')
I.seeNumberOfElements('.item', 5)


see	seeElement	seeElementInDOM	seeInField
dontSee	dontSeeElement	dontSeeElementInDOM	dontSeeInField
seeInCurrentUrl	seeInTitle	seeCheckboxIsChecked	seeCookie
seeNumberOfElements	seeNumberOfVisibleElements	seeTextEquals	seeAttributesOnElements

For custom assertions use grab* methods to retrieve data, then assert with any library

const title = await I.grabTitle(); 
expect(title).toEqual('My App')`

Fetching Content from Page

Grabbers retrieve data from the page for use in subsequent steps. They are the equivalent of Playwright's textContent(), Cypress's cy.get().invoke(), or WebDriver's getText() — but integrated into the CodeceptJS promise chain.

Grabbers always require await since they return data out of the known promise chain.

const title = await I.grabTitle()
const text = await I.grabTextFrom('.user-name')
const val = await I.grabValueFrom('#email-input')
const href = await I.grabAttributeFrom('a.logo', 'href')
const count = await I.grabNumberOfVisibleElements('.item')

In complex scenarios, grabbers are used to extract data from page and pass it to next steps.

Scenario('login with generated password', async ({ I }) => {
  I.click('Generate Password')
  const password = await I.grabTextFrom('#generated-password')
  I.fillField('Password', password)
  I.click('Sign In')
  I.see('Welcome')
})


grabTextFrom	grabTextFromAll	grabValueFrom	grabValueFromAll
grabHTMLFrom	grabAttributeFrom	grabAttributeFromAll	grabNumberOfVisibleElements
grabCurrentUrl	grabTitle	grabCookie	grabSource
grabWebElement	grabWebElements	grabElementBoundingRect	grabCssPropertyFrom

Waiting

CodeceptJS automatically waits for elements before clicking, filling, and most other interactions — so explicit waits are rarely needed. Failed steps are also automatically retried.

Use wait* methods when you need to explicitly wait for a UI change, such as a modal appearing, a spinner hiding, or a value updating:

I.waitForVisible('.modal')               // wait for modal to appear
I.waitForInvisible('.spinner')           // wait for spinner to hide
I.waitForText('Success', 5, '.alert')   // wait for text in element (5s timeout)
I.waitForEnabled('#submit-btn')          // wait for button to become enabled
I.waitForElement('.results li', 10)      // wait for results to load

I.wait(2)                                // explicit pause in seconds (last resort)

Example usage inside scenario:

Scenario('submit and wait for confirmation', ({ I }) => {
  I.click('Submit Order')
  I.waitForVisible('.order-confirmation', 10)
  I.see('Order placed successfully', '.order-confirmation')
})


waitForVisible	waitForInvisible	waitForElement	waitForDetached
waitForText	waitForValue	waitForClickable	waitForEnabled
waitForDisabled	waitForFunction	waitToHide	waitInUrl
waitUrlEquals	waitNumberOfVisibleElements	wait

Running Tests

bash

npx codeceptjs run                      # run all tests
npx codeceptjs run tests/login_test.js        # run a single file
npx codeceptjs run --grep "checkout"    # run tests matching name pattern
npx codeceptjs run --grep "@smoke"      # run tests by tag

Run tests in parallel using NodeJS workers. Tests are split by scenarios, results aggregated in the main process:

bash

npx codeceptjs run-workers 3    # run with 3 parallel workers

Output Verbosity

bash

npx codeceptjs run --steps     # print each step
npx codeceptjs run --debug     # print steps + additional debug info
npx codeceptjs run --verbose   # print everything including promises

It is recommended to always launch tests in --debug mode when developing tests.

Headless / Headed Mode

By default tests run headless (no browser window). To open a browser during test execution set show: true in helper config, or use @codeceptjs/configure:

import { setHeadlessWhen } from '@codeceptjs/configure'

setHeadlessWhen(process.env.CI)  // headless only on CI, show browser locally

For a single run without editing config, use the browser plugin:

npx codeceptjs run -p browser:show   # force visible browser
npx codeceptjs run -p browser:hide   # force headless

See Plugin Arguments.

Configuration

Configuration is set in codecept.conf.js. The two most important settings are the helper (which browser engine to use) and the base URL of your application:

export const config = {
  helpers: {
    Playwright: {
      url: 'http://localhost:3000',  // base URL for I.amOnPage('/')
      show: !process.env.CI,         // show browser locally, headless on CI
      browser: 'chromium',
    },
  },
  tests: './**/*_test.js',
  output: './output',
}

Use -c to switch between config files:

bash

npx codeceptjs run -c codecept.ci.conf.js

▶ See complete configuration reference.