Robot Framework

This document provides details about CumulusCI’s integration with Robot Framework for automating tests using the CumulusCI, Salesforce API’s, and Selenium.

Why Robot Framework?

Robot Framework provides an abstraction layer for writing automated test scenarios in Python and via text keywords in .robot files. Since Robot Framework is written in Python (like CumulusCI) and has a robust SeleniumLibrary for automated browser testing, it was an easy integration providing a lot of power.

CumulusCI’s integration with Robot Framework allows building automated test scenarios useful to Salesforce projects:

  • Browser testing with Selenium

  • API only tests interacting with the Salesforce REST, Bulk, and Tooling API’s

  • Complex org automation via CumulusCI

  • Combinations of all of the above

The ability to create rich, single file integration tests that interact with CumulusCI’s project specific automation, Salesforce’s APIs, and the Salesforce UI in a browser is the most exciting feature of the integration with Robot Framework.

The integration with Robot Framework adds a new dimension to CumulusCI. Before, automating the recreation of a test environment for an edge case bug reported in a custom org would have required creating new tasks in cumulusci.yml which pollute the project’s task list used by everyone on the project for an obscure scenario needed only for regression testing. Now, you can create the test scenario in a .robot test file and run it through the standard robot task in CumulusCI. Adding a new test scenario just adds a new file in the repository rather than a new task in CumulusCI.

Included Libraries

CumulusCI comes bundled with the following additional third-party keyword libraries in addition to the libraries that come with robot framework itself:

SeleniumLibrary is automatically imported when you import Salesforce.robot. To use RequestsLibrary you need to explicitly import it in the settings section of your robot test.

Example Robot Test

The following test file placed under robot/ExampleProject/tests/create_contact.robot in your project’s repository automates the testing of creating a Contact through the Salesforce UI in a browser and via the API. As an added convenience, it automatically deletes the created Contacts in the Suite Teardown step:

*** Settings ***

Resource        cumulusci/robotframework/Salesforce.robot
Library         cumulusci.robotframework.PageObjects

Suite Setup     Open Test Browser
Suite Teardown  Delete Records and Close Browser

*** Test Cases ***

    ${first_name} =       Get fake data  first_name
    ${last_name} =        Get fake data  last_name
    ${contact_id} =       Salesforce Insert  Contact
    ...                     FirstName=${first_name}
    ...                     LastName=${last_name}

    &{contact} =          Salesforce Get  Contact  ${contact_id}
    Validate Contact      ${contact_id}  ${first_name}  ${last_name}

Via UI
    ${first_name} =       Get fake data  first_name
    ${last_name} =        Get fake data  last_name

    Go to page            Home  Contact
    Click Object Button   New
    Wait for modal        New  Contact

    Populate Form
    ...                   First Name=${first_name}
    ...                   Last Name=${last_name}
    Click Modal Button    Save

    Wait Until Modal Is Closed

    ${contact_id} =       Get Current Record Id
    Store Session Record  Contact  ${contact_id}
    Validate Contact      ${contact_id}  ${first_name}  ${last_name}

*** Keywords ***

Validate Contact
    [Arguments]          ${contact_id}  ${first_name}  ${last_name}
    ...  Given a contact id, validate that the contact has the
    ...  expected first and last name both through the detail page in
    ...  the UI and via the API.

    # Validate via UI
    Go to page             Detail   Contact  ${contact_id}
    Page Should Contain    ${first_name} ${last_name}

    # Validate via API
    &{contact} =     Salesforce Get  Contact  ${contact_id}
    Should Be Equal  ${first_name}  ${contact}[FirstName]
    Should Be Equal  ${last_name}   ${contact}[LastName]


The Settings section of the robot file sets up the entire test suite. By including the Resource cumulusci/robotframework/Salesforce.robot which comes with CumulusCI, we inherit a lot of useful configuration and keywords for Salesforce testing automatically.

The Suite Setup and Suite Teardown are run at the start and end of the entire test suite. In the example test, we’re using the Open Test Browser keyword from the Salesforce.robot file to open a test browser. We’re also using the Delete Records and Close Browser keyword from Salesforce.robot to automatically delete all records created in the org during the session and close the test browser.

Test Cases

The two test cases test the same operation done through two different paths: the Salesforce REST API and the Salesforce UI in a browser.


This test case uses the Get fake data keyword to generate a first and last name. It then uses the Salesforce Insert keyword from the Salesforce Library (included via Salesforce.robot) to insert a Contact using the same technique for generating test data. Next, it uses Salesforce Get to retrieve the Contact’s information as a dictionary.

Finally, the test calls the Validate Contact keyword explained in the Keywords section below.

Via UI

This test case also uses Get fake data for the first and last name, but instead uses the test browser to create a Contact via the Salesforce UI. Using keywords from the Salesforce Library, it navigates to the Contact home page and clicks the New button to open a modal form. It then uses Populate Form to fill in the First Name and Last Name fields (selected by field label) and uses Click Modal Button to click the Save button and Wait Until Modal Is Closed to wait for the modal to close.

At this point, we should be on the record view for the new Contact. We use the Get Current Record Id keyword to parse the Contact’s ID from the url in the browser and the Store Session Record keyword to register the Contact in the session records list. The session records list stores the type and id of all records created in the session which is used by the Delete Records and Close Browser keyword on Suite Teardown to delete all the records created during the test. In the Via API test, we didn’t have to register the record since the Salesforce Insert keyword does that for us automatically. In the Via UI test, we created the Contact in the browser and thus need to store its ID manually for it to be deleted.


The Keywords section allows you to define keywords useful in the context of the current test suite. This allows you to encapsulate logic you want to reuse in multiple tests. In this case, we’ve defined the Validate Contact keyword which accepts the contact id, first, and last names as argument and validates the Contact via the UI in a browser and via the API via Salesforce Get. By abstracting out this keyword, we avoid duplication of logic in the test file and ensure that we’re validating the same thing in both test scenarios.

Running the Test Suite

This simple test file can then be run via the robot task in CumulusCI:

$ cd ~/dev/MyProject
$ cci task run robot -o suites robot/MyProject/tests/create_contact.robot -o vars BROWSER:firefox
2019-04-26 09:47:24: Getting scratch org info from Salesforce DX
2019-04-26 09:47:28: Beginning task: Robot
2019-04-26 09:47:28:        As user:
2019-04-26 09:47:28:         In org: 00DS0000003ORti
2019-04-26 09:47:28:
Create Contact
Via API                                                               | PASS |
[ WARN ] Retrying call to method _wait_until_modal_is_closed
Via UI                                                                | PASS |
Create Contact                                                        | PASS |
2 critical tests, 2 passed, 0 failed
2 tests total, 2 passed, 0 failed
Output:  /Users/boakley/dev/MyProject/robot/MyProject/results/output.xml
Log:     /Users/boakley/dev/MyProject/robot/MyProject/results/log.html
Report:  /Users/boakley/dev/MyProject/robot/MyProject/results/report.html


In the example output, the WARN line shows functionality from the Salesforce Library which helps handle retry scenarios common to testing against Salesforce’s Lightning UI. In this case, it automatically retried the wait for the modal window to close after creating a contact in a browser.

If you put all of your tests inside that robot/<project name>/tests folder you don’t have to use the suite option. By default the robot task will run all tests in the folder and all subfolders. For example, to run all tests and use the default browser you just have to issue the command cci task run robot.


Keywords can be defined in a test suite file, but they can also be defined in libraries and resource files. Libraries are written in python, and resource files are written in the robot syntax. Resource files are almost identical to a test file, except that they have no tests and can be imported into other test files. In addition to containing keywords, resource files can also define variables and they can import other libraries.

The file cumulusci/robotframework/Salesforce.robot was designed to be the way to import all of the keywords and variables provided by CumulusCI. It should be the first item imported in a test file. It will import the Salesforce Library and CumulusCI Library, as well as the most commonly used robot libraries (Collections, OperatingSystem, String, and XML)

Variables defined in resource files are accessible to all tests in a suite which imports the resource file. They can be set in your cumulusci.yml file, or specified with the vars option to the robot task. When doing so, the variables need to be referenced without the dollar sign and curly braces. Variable names are case-insensitive.

For example, here is how to set the browser to firefox and the default timeout to 20 seconds in a cumulusci.yml file:

        - BROWSER:firefox
        - TIMEOUT:20 seconds

The same variables can be set from the command line to override the config file for a single test run. This example shows that you can use the lowercase name for convenience:

$ cci task run robot -o vars browser:firefox,timeout:20

Supported Variables

The following variables defined in Salesforce.robot are all used by the Open Test Browser keyword:


Defines the browser to be used for testing. Supported values are chrome, firefox,`` headlesschrome``, and headlessfirefox. Default: chrome


This sets the preferred size of the browser. It is specified in the form of widthxheight, and the values are passed to the Set window size keyword. Default: 1280x1024


This is automatically passed to the Set Selenium Implicit Wait keyword. Default: 7 seconds


This defines a delay added after every selenium command. It is automatically passed to the Set Selenium Speed keyword. Default: 0 seconds


This sets the default amount of time selenium commands will wait before timing out. It is automatically passed to the Set Selenium Timeout keyword. Default: 30 seconds

CumulusCI Library

The CumulusCI Library for Robot Framework provides access to CumulusCI’s functionality from inside a robot test. It is mostly used to get credentials to a Salesforce org and to run more complex automation to set up the test environment in the org.

Logging Into An Org

The Login Url* keyword returns a url with an updated OAuth access token to automatically log into the CumulusCI org from CumulusCI’s project keychain.

Run Task

The Run Task keyword is used to run named CumulusCI tasks configured for the project. These can be any of CumulusCI’s built in tasks as well as project specific custom tasks from the project’s cumulusci.yml file.

Run Task accepts a single argument, the task name. It optionally accepts task options in the format option_name=value.

Run Task Class

The Run Task Class keyword is for use cases where you want to use one of CumulusCI’s Python task classes to automate part of a test scenario but don’t want to have to map a custom named task at the project level.

Run Task Class accepts a single argument, the class_path like would be entered into cumulusci.yml such as cumulusci.tasks.salesforce.Deploy. Like Run Task, you can also optionally pass task options in the format option_name=value.

Full Documentation

Full documentation of the keywords in the CumulusCI and Salesforce keyword libraries can be found here:

Salesforce Library

The Salesforce Library provides a set of useful keywords for interacting with Salesforce’s Lightning UI and Salesforce’s APIs to test Salesforce applications.

UI Keywords

The goal of the UI keywords in the Salesforce Library is to abstract out common interactions with Salesforce from interactions with your application’s UI. The Salesforce Library itself has an extensive suite of robot tests which are regularly run to alert us to any changes in the base Salesforce UI. By centralizing these interactions and regularly testing them, the Salesforce Library provides a more stable framework on which to build your product tests.

There are too many keywords relating to UI interactions to cover here. Please reference the full Salesforce Library documentation below.

Waiting for Lightning UI

A common challenge when writing end-to-end UI tests is the need to wait for asynchronous actions to complete before proceeding to run the next interaction. The Salesforce Library is aware of the Lightning UI and can handle this waiting automatically. After each click, it will wait for any pending requests to the server to complete. (Manually waiting using a “sleep” or waiting for a particular element to appear may still be necessary after other kinds of interactions and when interacting with pages that don’t use the Lightning UI.)

API Keywords

In addition to browser interactions, the Salesforce Library also provides the following keywords for interacting with the Salesforce REST API:

  • Salesforce Collection Insert: used for bulk creation of objects based on a template

  • Salesforce Collection Update: used for the bulk updating of objects

  • Salesforce Delete: Deletes a record using its type and ID

  • Salesforce Get: Gets a dictionary of a record from its ID

  • Salesforce Insert: Inserts a record using its type and field values. Returns the ID.

  • Salesforce Query: Runs a simple query using the object type and field=value syntax. Returns a list of matching record dictionaries.

  • Salesforce Update: Updates a record using its type, ID, and field=value syntax

  • SOQL Query: Runs a SOQL query and returns a REST API result dictionary

PageObjects Library

The PageObjects library provides support for page objects, Robot Framework-style. Even though robot is a keyword-driven framework, we’ve implemented a way to dynamically load in keywords that are unique to a page or an object on the page.

With this library, you can define classes which represent page objects. Each class provides keywords that are unique to a page or a component. These classes can be imported on demand only for tests which use these pages or components.

The pageobject Decorator

Page objects are normal Python classes which use the pageobject decorator provided by CumulusCI. Unlike traditional Robot Framework keyword libraries, you may define multiple sets of keywords in a single file.

When you create a page object class, you start by inheriting from one of the provided base classes. No matter which class your inherit from, your class gets the following predefined properties:

  • self.object_name is the name of the object related to the class. This is defined via the object_name parameter to the pageobject decorator. You should not add the namespace prefix in the decorator. This attribute will automatically add the prefix from cumulusci.yml when necessary.

  • self.builtin is a reference to the robot framework BuiltIn library, and can be used to directly call built-in keywords. Any built-in keyword can be called by converting the name to all lowercase, and replacing all spaces with underscores (eg: self.builtin.log, self.builtin.get_variable_value, etc).

  • self.cumulusci is a reference to the CumulusCI keyword library. You can call any keyword in this library by converting the name to all lowercase, and replacing all spaces with underscores (eg: self.cumulusci.get_org_info, etc).

  • self.salesforce is a reference to the Salesforce keyword library. You can call any keyword in this library by converting the name to all lowercase, and replacing all spaces with underscores (eg: self.salesforce.wait_until_loading_is_complete, etc).

  • self.selenium is a reference to SeleniumLibrary. You can call any keyword in this library by converting the name to all lowercase, and replacing all spaces with underscores (eg: self.selenim.wait_until_page_contains_element, etc)

Page Object Base Classes

Presently, cumulusci provides the following base classes, which should be used for all classes that use the pageobject decorator:

  • cumulusci.robotframework.pageobjects.BasePage - a generic base class used by the other base classes. It can be used when creating custom page objects when none of the other base classes make sense.

  • cumulusci.robotframework.pageobjects.DetailPage - a class for a page object which represents a detail page

  • cumulusci.robotframework.pageobjects.HomePage - a class for a page object which represents a home page

  • cumulusci.robotframework.pageobjects.ListingPage - a class for a page object which represents a listing page

  • cumulusci.robotframework.pageobject.NewModal - a class for a page object which represents the “new object” modal

  • cumulusci.robotframework.pageobject.ObjectManagerPage - a class for interacting with the object manager.

The BasePage class adds the following keyword to every page object:

  • Log current page object - this keyword is mostly useful while debugging tests. It will add to the log information about the currently loaded page object.

Example Page Object

The following example shows the definition of a page object for the listing page of a custom object named MyObject__c. It adds a new keyword named Click on the row with name:

from cumulusci.robotframework.pageobjects import pageobject, ListingPage

@pageobject(page_type="Listing", object_name="MyObject__c")
class MyObjectListingPage(ListingPage):

    def click_on_the_row_with_name(self, name):

The pageobject decorator takes two arguments: page_type and object_name. These two arguments are used to identify the page object (eg: Go To Page  Listing  Contact). The values can be any arbitrary string, but ordinarily should represent standard page types (“Detail”, “Home”, “Listing”, “New”), and standard object names.

Importing the library into a test

The PageObjects library is somewhat unique in that it is not only a keyword library, but also the mechanism by which you can import files which contain page object classes. This is done by providing the paths to one or more Python files which implement page objects. You may also import PageObjects without passing any files to it in order to take advantage of some general purpose page objects.

For example, consider the case where you’ve created two files that each have one or more page object definitions. For example, lets say in robot/MyProject/resources you have the files and You can import these page objects into a test suite like so:

*** Settings ***
Library         cumulusci.robotframework.PageObjects
...  robot/MyProject/resources/
...  robot/MyProject/resources/

Using Page Objects

There are two things that must be done in order to use the keywords in a page object. The first has already been covered, and that is to import the PageObjects library and any custom page object files you wish to use.

The second thing you must do is either explicitly load the keywords for a page object, or reference a page object with one of the generic keywords provided by the PageObjects library.

To explicitly load the keywords for a page object you can use the load page object keyword provided by the PageObjects library. Other keywords provided by that library will automatically import the keywords if they are successful. For example, you can call Go To Page followed by a page object reference, and if that page is able to be navigated to, its keywords will automatically be loaded.

Page Object Keywords

The PageObjects library provides the following keywords:

  • Current Page Should Be

  • Get Page Object

  • Go To Page Object

  • Load Page Object

  • Log Page Object Keywords

  • Wait For Modal

  • Wait For Page Object

Current Page Should Be

Example: Current Page Should Be  Listing  Contact

This keyword will attempt to validate that the given page object represents the current page. Each page object may use its own method for making the determination, but the built-in page objects all compare the page location to an expected pattern (eg: .../lightning/o/...). If the assertion passes, the keywords for that page object will autoamtically be loaded.

This keyword is useful if you get to a page via a button or some other form of navigation, in that it allows you to both assert that you are on the page you think you should be on, and load the keywords for that page, all with a single statement.

Get Page Object

Example: Get page object  Listing  Contact

This keyword is rarely used in a test. It is mostly useful to get the reference to a other keyword from another keyword. It is similar in function to robot’s built-in Get library instance keyword.

Go To Page

Example: Go to page  Listing  Contact

This will attempt to go to the listing page for the Contact object, and then load the keywords for that page.

Log Page Object Keywords

Example: Log Page Object Keywords

This keyword is primarily a debugging tool. When called it will log each of the keywords for the current page object.

Load Page Object

Example: Load page object  Listing  Contact

This will load the page object for the given page_type and object_name_. It is useful when you want to use the keywords from a page object without first navigating to that page (i.e. when you are already on the page and don’t want to navigate away).

Wait For Modal

Example: Wait for modal  New  Contact

This keyword can be used to wait for a modal, such as the one that pops up when creating a new object. The keyword will return once a modal appears, and has a title of “New _object_” (eg: “New Contact”).

Wait For Page Object

Example: Wait for page object  Popup  ActivityManager

Page objects don’t necessarily have to represent entire pages. You can use Wait for page object to wait for a page object representing a single element on a page such as a popup window.

Generic Page Objects

You do not need to create a page object in order to take advantage of the new page object keywords. If you use one of the page object keywords for a page that does not have its own page object, the PageObjects library will try to find a generic page.

For example, if you use Current page should be  Home  Event and there is no page object by that name, a generic Home page object will be loaded, and its object name will be set to Event.

Let’s say your project has created a custom object named Island. You don’t have a home page, but the object does have a standard listing page. Without creating any page objects, this test should work by using generic implementations of the Home and Listing page objects:

*** Test Cases ***
Example test which uses generic page objects
    # Go to the custom object home page, which should
    # redirect to the listing page
    Go To Page  Home  Islands

    # Verify that the redirect happened
    Current Page Should Be  Listing  Islands

CumulusCI provides the following generic page objects:

  • Detail (eg: Go to page  Detail  Contact  ${contact id}) Detail pages refer to pages with a URL that matches the pattern “<host>/lightning/r/<object name>/<object id>/view”

  • Home (eg: Go to page  Home  Contact) Home pages refer to pages with a URL that matches the pattern “<host>/lightning/o/<object name>/home”

  • Listing (eg: Go to  page  Listing  Contact) Listing pages refer to pages with a URL that matches the pattern “<host>b/lightning/o/<object name>/list”

  • New (eg: Wait for modal  New  Contact) The New page object refers to the modal that pops up when creating a new object.

Of course, the real power comes when you create your own page object class which implements keywords which can be used with your custom objects.

Keyword Documentation

Use the following links to download generated documentation for both the CumulusCI and Salesforce keywords

CumulusCI Robot Tasks

CumulusCI includes several tasks for working with Robot Framework tests and keyword libraries:

  • robot: Runs robot test suites. By default, recursively runs all tests located under the folder robot/<project name>/tests/. Test suites can be overridden via the suites keyword and variables inside robot files can be overridden using the vars option with the syntax VAR:value (ex: BROWSER:firefox).

  • robot_testdoc: Generates html documentation of your whole robot test suite and writes to robot/<project name>/doc/<project_name>.html.

  • robot_lint: Performs static analysis of robot files (files with .robot and .resource), flagging issues that may reduce the quality of the code.

  • robot_libdoc: This task can be wired up to generate library documentation if you choose to create a library of robot keywords for your project.

Configuring the libdoc task

If you have defined a robot resource file named MyProject.resource and placed it in the resources folder, you can add the following configuration to your cumulusci.yml file in order to enable the robot_libdoc task to generate documentation:

       description: Generates HTML documentation for the MyProject Robot Framework Keywords
           path: robot/MyProject/resources/MyProject.resource
           output: robot/MyProject/doc/MyProject_Library.html

You can generate documentation for more than one keyword file or library by giving a comma-separated list of files for the path option, or by defining path as a list in cumulusci.yml. In the following example, documentation will be generated for and MyLibrary.resource:

       description: Generates HTML documentation for the MyProject Robot Framework Keywords
             - robot/MyProject/resources/MyProject.resource
             - robot/MyProject/resources/
           output: robot/MyProject/doc/MyProject_Library.html

You can also use basic filesystem wildcards. For example, to document all robot files in robot/MyProject/resources you could configure your yaml file like this:

       description: Generates HTML documentation for the MyProject Robot Framework Keywords
           path: robot/MyProject/resources/*.resource
           output: robot/MyProject/doc/MyProject_Library.html

Robot Directory Structure

When you use cci project init, it creates a folder named robot at the root of your repository. Immediately under that is a folder for your project robot files. If your project depends on keywords from other projects, they would also be in the robot folder under their own project name.

├── robot
│   └── MyProject
│       ├── doc
│       ├── resources
│       ├── results
│       └── tests

With the project folder inside the robot folder are the following additional folders:

  • doc: the location where generated documentation will be placed.

  • resources: this folder is where you can put your own keyword files. You can create robot keyword files (.resource or .robot) as well as keyword libraries (.py). For keyword files we recommend using the .resource suffix.

  • results: this folder isn’t created by cci project init. Instead, it will automatically be created the first time you run your tests. It will contain all of the generated logs and screenshots.

  • tests: this is where you should put your test suites. You are free to organize this however you wish, including adding subfolders.

Creating Project Tests

Like in the example above, all project tests live in .robot files stored under the robot/<project name>/tests/ directory in the project. You can choose how you want to structure the .robot files into directories by just moving the files around. Directories are treated by robot as a parent test suite so a directory named “standard_objects” would become the “Standard Objects” test suite.

The following document is recommended reading:

Using Keywords and Tests from a Different Project

Much like you can use tasks and flows from a different project, you can also use keywords and tests from other projects. The keywords are brought into your repository the same way as with tasks and flows, via the sources configuration option in cumulusci.yml. However, they require a little extra configuration before they can be used.


This feature should not be used for general purpose sharing of keywords between multiple projects. This feature was designed specifically for the case where a product is being built on top of another project and needs access to product-specific keywords.

Using keywords

In order to use the resources from another project you must first configure the robot task to use one of the sources that have been defined for the project. To do this, add a sources option in the robot task, and add to it the name of one of the imported sources.

For exmple, if your project is built on top of NPSP and you want to use keywords from the NPSP project, you must first add the npsp repository as a source in the project’s cumulusci.yml like so:

        release: latest_beta

You must then add npsp under the sources option for the robot task. This is because the project as a whole may use tasks or flows from multiple projects, but robot only needs keywords from a single project.

          - npsp

When the robot task runs, it adds the directory which contains the code for the other repository to PYTHONPATH which robot uses when resolving references to libraries and keyword files.

Once this configuration has been saved, you can import the resources just as if you were in the NPSP repository. For example, in a project which has been configured to use npsp as a source, the following example shows how NPSP.robot can be imported into a test suite:

*** Settings ***
Resource   robot/Cumulus/resources/NPSP.robot


Even with proper configuration, some keywords or keyword libraries might not be usable. You must be careful to not try to use files that have the exact same name in multiple repositories.

Running Tests

Running a test from another project requires prefixing the path to the test with the source name. The path needs to be relative to the root of the other repo.

For example, starting from the previous example, to run the create_organization.robot test suite from NPSP, you would do it with something like this:

$ cci task run robot -o suites npsp:robot/Cumulus/tests/browser/contacts_accounts/create_organization.robot