Custom Commands

Cypress comes with its own API for creating custom commands and overwriting existing commands. The built in Cypress commands use the very same API that's defined below.

Syntax

Cypress.Commands.add(name, callbackFn)
Cypress.Commands.add(name, options, callbackFn)
Cypress.Commands.addAll(callbackObj)
Cypress.Commands.addAll(options, callbackObj)
Cypress.Commands.overwrite(name, callbackFn)

Usage

Correct Usage

Cypress.Commands.add('login', (email, pw) => {})
Cypress.Commands.addAll({
  login(email, pw) {},
  visit(orig, url, options) {},
})
Cypress.Commands.overwrite('visit', (orig, url, options) => {})

Arguments

name (String)

The name of the command you're either adding or overwriting.

callbackFn (Function)

Pass a function that receives the arguments passed to the command.

callbackObj (Object)

An object with callbackFns as properties.

options (Object)

Pass in an options object to define the implicit behavior of the custom command.

OptionAcceptsDefaultDescription
prevSubjectBoolean, String or Arrayfalsehow to handle the previously yielded subject.

The prevSubject accepts the following values:

  • false: ignore any previous subjects: (parent command)
  • true: receives the previous subject: (child command)
  • optional: may start a chain, or use an existing chain: (dual command)

In addition to controlling the command's implicit behavior you can also add declarative subject validations such as:

  • element: requires the previous subject be a DOM element
  • document: requires the previous subject be the document
  • window: requires the previous subject be the window

Examples

Parent Commands

Parent commands always begin a new chain of commands. Even if you've chained it off of a previous command, parent commands will always start a new chain, and ignore previously yielded subjects.

Examples of parent commands:

Cypress.Commands.add('clickLink', (label) => {
  cy.get('a').contains(label).click()
})
cy.clickLink('Buy Now')

Check a token

Cypress.Commands.add('checkToken', (token) => {
  cy.window().its('localStorage.token').should('eq', token)
})
cy.checkToken('abc123')

Download a file

Originally used in cypress-downloadfile, this command calls other Cypress commands.

Cypress.Commands.add('downloadFile', (url, directory, fileName) => {
  return cy.getCookies().then((cookies) => {
    return cy.task('downloadFile', {
      url,
      directory,
      cookies,
      fileName,
    })
  })
})
cy.downloadFile('https://path_to_file.pdf', 'mydownloads', 'demo.pdf')

Commands to work with sessionStorage

Cypress.Commands.add('getSessionStorage', (key) => {
  cy.window().then((window) => window.sessionStorage.getItem(key))
})

Cypress.Commands.add('setSessionStorage', (key, value) => {
  cy.window().then((window) => {
    window.sessionStorage.setItem(key, value)
  })
})
cy.setSessionStorage('token', 'abc123')
cy.getSessionStorage('token').should('eq', 'abc123')

Log in command using UI

Cypress.Commands.add('typeLogin', (user) => {
  cy.get('input[name=email]').type(user.email)

  cy.get('input[name=password]').type(user.password)
})
cy.typeLogin({ email: 'fake@email.com', password: 'Secret1' })

Log in command using request

Cypress.Commands.add('login', (userType, options = {}) => {
  // this is an example of skipping your UI and logging in programmatically

  // setup some basic types
  // and user properties
  const types = {
    admin: {
      name: 'Jane Lane',
      admin: true,
    },
    user: {
      name: 'Jim Bob',
      admin: false,
    },
  }

  // grab the user
  const user = types[userType]

  // create the user first in the DB
  cy.request({
    url: '/seed/users', // assuming you've exposed a seeds route
    method: 'POST',
    body: user,
  })
    .its('body')
    .then((body) => {
      // assuming the server sends back the user details
      // including a randomly generated password
      //
      // we can now login as this newly created user
      cy.request({
        url: '/login',
        method: 'POST',
        body: {
          email: body.email,
          password: body.password,
        },
      })
    })
})
// can start a chain off of cy
cy.login('admin')

// can be chained but will not receive the previous subject
cy.get('button').login('user')

Log out command using UI

Cypress.Commands.add('logout', () => {
  cy.contains('Login').should('not.exist')
  cy.get('.avatar').click()
  cy.contains('Logout').click()
})

Log out command using localStorage
End-to-End Only

Cypress.Commands.add('logout', () => {
  cy.window().its('localStorage').invoke('removeItem', 'session')

  cy.visit('/login')
})
cy.logout()

Create a user

Cypress.Commands.add('createUser', (user) => {
  cy.request({
    method: 'POST',
    url: 'https://www.example.com/tokens',
    body: {
      email: 'admin_username',
      password: 'admin_password',
    },
  }).then((resp) => {
    cy.request({
      method: 'POST',
      url: 'https://www.example.com/users',
      headers: { Authorization: 'Bearer ' + resp.body.token },
      body: user,
    })
  })
})
cy.createUser({
  id: 123,
  name: 'Jane Lane',
})

Child Commands

Child commands are always chained off of a parent command, or another child command.

The previous subject will automatically be yielded to the callback function.

Examples of child commands:

Custom console command

// not a super useful custom command
// but demonstrates how subject is passed
// and how the arguments are shifted
Cypress.Commands.add(
  'console',
  {
    prevSubject: true,
  },
  (subject, method) => {
    // the previous subject is automatically received
    // and the commands arguments are shifted

    // allow us to change the console method used
    method = method || 'log'

    // log the subject to the console
    console[method]('The subject is', subject)

    // whatever we return becomes the new subject
    //
    // we don't want to change the subject so
    // we return whatever was passed in
    return subject
  }
)
cy.get('button')
  .console('info')
  .then(($button) => {
    // subject is still $button
  })

By setting the { prevSubject: true }, our new .console() command will require a subject.

Invoking it like this would error:

cy.console() // error about how you can't call console without a subject

Dual Commands

A dual command can either start a chain of commands or be chained off of an existing one. It is basically the hybrid between both a parent and a child command. You will likely rarely use this, and only a handful of our internal commands use this.

Nevertheless, it is useful if your command can work in multiple ways - either with an existing subject or without one.

Examples of dual commands:

Custom Dual Command

Cypress.Commands.add('dismiss', {
  prevSubject: 'optional'
}, (subject, arg1, arg2) => {
  // subject may be defined or undefined
  // so you likely want to branch the logic
  // based off of that

  if (subject) {
    // wrap the existing subject
    // and do something with it
    cy.wrap(subject)
    ...
  } else {
    ...
  }
})
cy.dismiss() // no subject
cy.get('#dialog').dismiss() // with subject

Overwrite Existing Commands

You can also modify the behavior of existing Cypress commands. This is useful to always set some defaults to avoid creating another command that ends up using the original.

Overwrite visit command

Cypress.Commands.overwrite('visit', (originalFn, url, options) => {
  const domain = Cypress.env('BASE_DOMAIN')

  if (domain === '...') {
    url = '...'
  }

  if (options.something === 'else') {
    url = '...'
  }

  // originalFn is the existing `visit` command that you need to call
  // and it will receive whatever you pass in here.
  //
  // make sure to add a return here!
  return originalFn(url, options)
})

Overwrite type command

If you are typing into a password field, the password input is masked automatically within your application. But .type() automatically logs any typed content into the Cypress Command Log.

cy.get('#username').type('username@email.com')
cy.get('#password').type('superSecret123')

You may want to mask some values passed to the .type() command so that sensitive data does not display in screenshots or videos of your test run. This example overwrites the .type() command to allow you to mask sensitive data in the Cypress Command Log.

Cypress.Commands.overwrite('type', (originalFn, element, text, options) => {
  if (options && options.sensitive) {
    // turn off original log
    options.log = false
    // create our own log with masked message
    Cypress.log({
      $el: element,
      name: 'type',
      message: '*'.repeat(text.length),
    })
  }

  return originalFn(element, text, options)
})
cy.get('#username').type('username@email.com')
cy.get('#password').type('superSecret123', { sensitive: true })

Now our sensitive password is not printed to the Cypress Command Log when sensitive: true is passed as an option to .type().

Overwrite screenshot command

This example overwrites cy.screenshot() to always wait until a certain element is visible.

Cypress.Commands.overwrite(
  'screenshot',
  (originalFn, subject, name, options) => {
    // call another command, no need to return as it is managed
    cy.get('.app')
      .should('be.visible')

      // overwrite the default timeout, because screenshot does that internally
      // otherwise the `then` is limited to the default command timeout
      .then({ timeout: Cypress.config('responseTimeout') }, () => {
        // return the original function so that cypress waits for it
        return originalFn(subject, name, options)
      })
  }
)

Overwrite click command

This example overwrites .click() to always have the waitForAnimations option set to false.

Cypress.Commands.overwrite(
  'click',
  (originalFn, subject, positionOrX, y, options = {}) => {
    options.waitForAnimations = false
    return originalFn(subject, positionOrX, y, options)
  }
)

Validations

As noted in the Arguments above, you can also set prevSubject to one of:

  • element
  • document
  • window

When doing so Cypress will automatically validate your subject to ensure it conforms to one of those types.

Require Element

Require subject be of type: element.

// this is how .click() is implemented
Cypress.Commands.add(
  'click',
  {
    prevSubject: 'element',
  },
  (subject, options) => {
    // receives the previous subject and it's
    // guaranteed to be an element
  }
)

Valid Usage

cy.get('button').click() // has subject, and is `element`

Invalid Usage

cy.click() // no subject, will error
cy.wrap([]).click() // has subject, but not `element`, will error

Allow Multiple Types

.trigger()

Require subject be one of the following types: element, document or window

// this is how .trigger() is implemented
Cypress.Commands.add(
  'trigger',
  {
    prevSubject: ['element', 'document', 'window'],
  },
  (subject, eventName, options) => {
    // receives the previous subject and it's
    // guaranteed to be an element, document, or window
  }
)

Valid Usage

cy.get('button').trigger() // has subject, and is `element`
cy.document().trigger() // has subject, and is `document`
cy.window().trigger() // has subject, and is `window`

Invalid Usage

cy.trigger() // no subject, will error
cy.wrap(true).trigger() // has subject, but not `element`, will error

Validations always work as "or" not "and".

Optional with Types

You can also mix optional commands with validations.

// this is how .scrollTo() is implemented
Cypress.Commands.add(
  'scrollTo',
  {
    prevSubject: ['optional', 'element', 'window'],
  },
  (subject, ...args) => {
    // subject could be undefined
    // since it's optional.
    //
    // if it's present then it's and element or window.
    // - when window, we'll scroll to a position on the page.
    // - when element, we'll scroll to a position related to the element.
    if (subject) {
      // ...
    } else {
      // ...
    }
  }
)

Valid Usage

cy.scrollTo() // no subject, but valid because it's optional
cy.get('#main').scrollTo() // has subject, and is `element`
cy.visit().scrollTo() // has subject, and since visit yields `window` it's ok

Invalid Usage

cy.document().scrollTo() // has subject, but it's a `document`, will error
cy.wrap(null).scrollTo() // has subject, but it's `null`, will error

Notes

Command Logging

When creating your own custom command, you can control how it appears and behaves in the Command Log.

Take advantage of the Cypress.log() API. When you're issuing many internal Cypress commands, consider passing { log: false } to those commands, and programmatically controlling your custom command. This will cleanup the Command Log and be much more visually appealing and understandable.

cy.hover() and cy.mount()

Cypress does not have cy.hover() or cy.mount() commands out-of-the-box. See how to craft your own cy.hover() and cy.mount() custom commands.

Best Practices

1. Don't make everything a custom command

Custom commands work well when you're needing to describe behavior that's desirable across all of your tests. Examples would be a cy.setup() or cy.login() or extending your application's behavior like cy.get('.dropdown').dropdown('Apples'). These are specific to your application and can be used everywhere.

However, this pattern can be used and abused. Let's not forget - writing Cypress tests is JavaScript, and it's often more efficient to write a function for repeatable behavior that's specific to only a single spec file.

If you're working on a search.cy.js file and want to compose several repeatable actions together, you should first ask yourself:

Can this be written as a function?

The answer is usually yes. Here's an example:

// There's no reason to create something like a cy.search() custom
// command because this behavior is only applicable to a single spec file
//
// Use a regular ol' javascript function folks!
const search = (term, options = {}) => {
  // example massaging to defaults
  _.defaults(options, {
    headers: {},
  })

  const { fixture, headers } = options

  // return cy chain here so we can
  // chain off this function below
  return cy
    .log(`Searching for: ${term} `)
    .intercept('GET', '/search/**', (req) => {
      req.reply({
        statusCode: 200,
        body: `fixture:${fixture}`,
        headers: headers,
      })
    })
    .as('getSearchResults')
    .get('#search')
    .type(term)
    .wait('@getSearchResults')
}

it('displays a list of search results', () => {
  cy.visit('/page')
    .then(() => {
      search('cypress.io', {
        fixture: 'list',
      }).then((reqRes) => {
        // do something with the '@getSearchResults'
        // request such as make assertions on the
        // request body or url params
        // {
        //   url: 'http://app.com/search?cypress.io'
        //   method: 'GET',
        //   duration: 123,
        //   request: {...},
        //   response: {...},
        // }
      })
    })
    .get('#results li')
    .should('have.length', 5)
    .get('#pagination')
    .should('not.exist')
})

it('displays no search results', () => {
  cy.visit('/page')
    .then(() => {
      search('cypress.io', {
        fixture: 'zero',
      })
    })
    .get('#results')
    .should('contain', 'No results found')
})

it('paginates many search results', () => {
  cy.visit('/page')
    .then(() => {
      search('cypress.io', {
        fixture: 'list',
        headers: {
          // trick our app into thinking
          // there's a bunch of pages
          'x-pagination-total': 3,
        },
      })
    })
    .get('#pagination')
    .should(($pagination) => {
      // should offer to goto next page
      expect($pagination).to.contain('Next')

      // should have provided 3 page links
      expect($pagination.find('li.page')).to.have.length(3)
    })
})

2. Don't overcomplicate things

Custom commands you write are generally an abstraction over a series of internal commands. That means you and your team members exert much more mental effort to understand what your custom command does.

There's no reason to add this level of complexity when you're only wrapping a couple commands.

Don't do things like:

  • cy.clickButton(selector)
  • .shouldBeVisible()

This first custom command is wrapping cy.get(selector).click(). Going down this route would lead to creating dozens or even hundreds of custom commands to cover every possible combination of element interactions. It's completely unnecessary.

The .shouldBeVisible() custom command isn't worth the trouble or abstraction when you can already use: .should('be.visible')

Testing in Cypress is all about readability and simplicity. You don't have to do that much actual programming to get a lot done. You also don't need to worry about keeping your code as DRY as possible. Test code serves a different purpose than app code. Understandability and debuggability should be prioritized above all else.

Try not to overcomplicate things and create too many abstractions. When in doubt, use a regular function for individual spec files.

3. Don't do too much in a single command

Make your custom commands composable and as unopinionated as possible. Cramming too much into them makes them inflexible and requires more and more options passing to control their behavior.

Try to add either zero or as few assertions as possible in your custom command. Those tend to shape your command into a much more rigid structure. Sometimes this is unavoidable, but a best practice is to let the calling code choose when and how to use assertions.

4. Skip your UI as much as possible

Custom commands are a great way to abstract away setup (specific to your app). When doing those kinds of tasks, skip as much of the UI as possible. Use cy.request() to login, set cookies or localStorage directly, stub and mock your applications functions, and / or trigger events programmatically.

Having custom commands repeat the same UI actions over and over again is slow, and unnecessary. Try to take as many shortcuts as possible.

5. Write TypeScript definitions

You can describe the method signature for your custom command, allowing IntelliSense to show helpful documentation. See the cypress-example-todomvc repository for a working example.

History

VersionChanges
0.20.0Cypress.Commands API added

See also