Gwen in Summary

Simpler web automation.

A Java runtime, a web browser, your favourite text editor, and the Gwen interpreter is all you need to start automating.

No clickety-click drag-drop mouse-pointy UI or dev-centric page-object code-compiling IDE or plugin required.

The interpreter is open source.

The automation is reusable, scalable, and shareable.

Many execution modes including sequential, parallel, dry run, and interactive REPL are provided.

Many reporting formats including JUnit XML, cucumber JSON, and pretty HTML are supported.

JavaScript can be injected in places if needed.

A CLI makes it easy to run locally and integrate with any on-premise or on-cloud build server.

Git ready workspaces make it easy for teams to collaborate.

A package manager downloads and installs native browser drivers for you.

Built on Gherkin and Selenium.

Easy to run on BrowserStack, Sauce Labs, Aerokube Selenoid and other grids.

Matured in industry for over 4 years.

Active user community is growing.

Gwen home page:

Share with your friends.

Feedback welcomed.

Running Gwen on BrowserStack

BrowserStack allows you to run automated Selenium tests on a variety of platforms, browsers, and devices on the cloud. Since Gwen is built on Selenium and also supports remote web driver, you can easily run your Gwen tests “as is” on BrowserStack too. As of gwen-web version 2.25.1, all you need to do is provide some runtime properties.

Remote URL

If you sign up for a BrowserStack account, you will receive a user name and access key. Add the following settings to your file to configure your connection to BrowserStack:

file: ~/

bs.user = your-browserstack-username
bs.key = your-browsetstack-access-key
gwen.web.remote.url = https://${bs.user}:${bs.key}

Target Capabilities

BrowserStack provides a neat capabilities page that can generate capabilities settings for you. Simply navigate to that page and select your desired target platform and browser. The capabilities settings will be generated for you and displayed as shown below:

This example shows the capabilities settings for a Chrome browser on the Mac OS X High Sierra Platform. To configure Gwen to use these settings, create a new properties file containing each of the settings above as follows:


gwen.web.capability.os = OS X
gwen.web.capability.os_version = High Sierra
gwen.web.capability.browser = Chrome
gwen.web.capability.browser_version = 67.0
gwen.web.capability.browserstack.local = false
gwen.web.capability.browserstack.selenium_version = 3.11.0

You can save this file anywhere you like. For example, you could save it in your current working directory as

You can create a separate properties file in the same manner for each other platform, browser, or device you wish to target.


Now to run your Gwen features on BrowserStack, simply invoke Gwen passing in the above properties file (through the -p option) as follows. This example will run all features in the features/google folder on BrowserStack targeting the Chrome browser on the Mac.

./gwen -b -p features/google

To target a different platform or browser, just pass in the properties file containing those capabilities instead.

If you logon to, you will see the results of all the remotely executed web driver steps and a video recording of the feature execution too.

That’s it!

Gwen Meta = Gherkin Automation Bindings

Declarative Features

From a business perspective, feature specifications should be declarative. What does does this mean? It means that features should describe the intended behavior of a feature independently of any implementation details (especially automation details). They should clearly express behavior in business terms and may also include some examples.

Imperative Features

Imperative specifications on the other hand consist of a step by step list of statements that describe a behavior in great detail and that can be followed robotically to emulate a process. You would ideally never write features in this manner when describing business specifications.

Meta Features

Declarative to Imperative Automation Bindings

Gwen is a Gherkin interpreter that was designed with behavior driven automation (BDA) in mind. It enables teams to bind declarative business features “as is” to imperative meta features to drive automation. Meta features are used to describe (or break down) declarative features into step definitions and bindings that can be used to drive automated testing and robotic processing. In Gwen, they are also defined in the Gherkin language and bind to declarative features at run time to make them executable.


Reuse can be maximised by sharing common meta across features


In the test-first approach, the team would first write feature specifications and then use those to drive development and automated testing at the same time. Executing features with Gwen prior to development would result in failures (as expected). As the feature is being implemented in code, the developers and/or testers can write meta features to define all the necessary step definitions and bindings. Loading the meta and executing features with Gwen during development would then enable the test execution to run. By progressively building up the required meta during development, the feature tests will eventually pass when implementation is complete.


In the test-later approach, Gherkin features can be used to drive user acceptance testing after development is complete. New acceptance features can be written if desired at this stage. Any meta that was created during development (if the test-first approach was practiced) could be reused here. Otherwise all the necessary step definitions and bindings would need to be defined in meta during this phase. The meta could then be loaded into Gwen to make the acceptance features perform the test execution.


Robotic process automation (RPA) involves the repeatable execution of tasks that would otherwise need to be performed by humans. A repetitive or monotonous process performed online by a person, for example, can be automated with Gwen by writing Gherkin specifications that describe that behavior. These too can be declarative with imperative bindings in meta, but can also be imperative only.

Gwen Web Engine

We have open sourced a web engine (Selenium wrapper) for Gwen that you can use to perform any sort of web UI automation using the approaches described above.

More on Gwen

Gwen for Behavior Driven Automation (BDA)

Gwen is first and foremost a BDA tool. BDD on the other hand has its purpose but it’s not automation. It’s a development practice. But it turns out that behavior in prose can be used to drive automation.

When you use behavioral specifications to drive automation of any kind, you are doing BDA. Be it for testing, robotic process automation (RPA) or any other purpose. It does not matter!

It just so happens that in BDD you are doing it to test-first at development time. In the case of test-later and also RPA, you are doing it after development time. Gwen can be used to transform behavioral specifications into automation operations in any case.

Gwen is about making behavior driven automation easier for everyone.

Core Gwen Interpreter

Web automation engine

Running Gwen Workspaces on Jenkins

In this post I’m going to walk though the basics of getting a Gwen Workspace up and running on Jenkins.


Be sure you have following setup before proceeding..

  • A running Jenkins environment with
    • Java 1.8+ installed
    • A browser installed (preferably Chrome)

Create a Gwen Workspace Repository

Next, you will need to convert your Gwen project over to a Gwen Workspace. If you have not already done this it is easy to do. Just follow these steps:

  • Download and unzip the latest to a location on your drive
  • Copy all your .feature and .meta files to the features folder
  • Check that your features execute successfully by running gwen features -b in the workspace root
    • You can tweak any Gwen properties or wrapper scripts in the workspace if required to tailor your execution.
  • Publish your workspace folder to your Git repository

The benefit of using a workspace is that it contains an embedded Gwen Package Manager that will automatically install and configure Gwen for you (so you don’t have to do this manually in the Jenkins environment). If you do not want to use a workspace but would rather utilise your current project setup “as is”, then you will have to do manual installation and configuration work on the Jenkins host to ensure that your features can execute (but the basic setting up of the Jenkins job will be similar to below).

Create a Jenkins Job

Once you have a workspace that is accessible from Git, you are ready to create a Jenkins job to run your workspace.

  • Logon to your Jenkins
  • Create a new “Freestyle” project and give it a name
  • In the “Source Code Management” section, select “Git” and provide your Gwen workspace repository URL and other Git settings
  • In the “Build” section, select “Execute Shell” (for linux) or “Execute Windows Batch Command” (for windows), and enter the following command to run Gwen
    • Linux: ./gwen features -b -f junit -Dgwen.web.browser.headless=true
    • Windows: gwen features -b -f junit -Dgwen.web.browser.headless=true

    -b tells Gwen to exit once execution is complete, -f junit tells Gwen to generate JUnit-XML reports, and -Dgwen.web.browser.headless=true tells Gwen to run the browser in headless mode. You can also pass additional Gwen options if required, like --parallel for example if you want to utilise all cores and perform parallel execution.

  • In the “Post-build Actions” section, select “Publish JUnit test result report”, and enter the following in the “Test Report XMLs” field.
    • target/reports/junit/TEST-*.xml
  • Click “Save” when you are done and then run your job

That’s it!!

Gwen Workspaces

Setting up Gwen for teams just got a whole lot easier!

Manually setting up and installing Gwen on multiple machines or build servers in a team environment can be tedious and can also result in inconsistent configurations across workstations. One of the reasons why we created gwen-gpm was to provide consistent installation across machines and platforms. But a team needs more than that. A team needs a consistent and seamless way of getting Gwen configured and running on any user workstation or build server too.

Introducing Gwen workspaces

Gwen workspaces solve this problem by defining a standard project structure on the file system complete with settings files and wrapper scripts that can easily be committed to Git and checked out on any machine. Any team member can then just checkout the workspace and start using Gwen straight away. Gwen will be automatically downloaded and installed on any user workstation or build server is not already present (through an embedded Gwen package manager in the workspace).

It is assumed that the target browser is already installed on the system. If not, you will need to manually install it.

The structure of this workspace is defined as follows:

|--/env : Put environment properties here
|--/features : Put your Gherkin feature and common meta files here
| gwen : Gwen launcher/wrapper script for linux
| gwen.bat : Gwen launcher/wrapper script for windows
| : Common Gwen properties
| : Gwen log settings
| gwen-gpm.jar : Gwen package manager
| .gitignore : Git ignore file

Create and Commit a Workspace to Git

To create a Gwen workspace for your team, perform the following (only one person in the team needs to do this):

  • Download and extract the┬áto a location on your computer
  • Tweak or add team wide settings to the file in the workspace root. You can also tailor the wrapper scripts if necessary (for example, if you want to change the default Gwen launch options or add some new ones).
  • Verify that it all works by launching gwen (on windows) or ./gwen (on linux). Type exit when done to quit the REPL session.
  • Commit and push the workspace to your remote Git repository (see online Git help if you are not sure how to do this)

Checkout Workspace from Git and Go..

To use the workspace on any machine, perform the following (all team members need to do this):

  • Ensure that the target browser is installed on the system
  • Checkout the workspace from Git
  • Open a command prompt to the root workspace location
  • Type..
    • gwen (on windows) or ./gwen (on linux) followed by the options you require
    • Gwen will execute using the settings in the workspace
      Note: this same command can be used on any build server that checks out the workspace too
    • Gwen and native web drivers will self install on the first call
    • Type exit when done to quit the REPL session (if you started Gwen in REPL mode).

The team can now manage all their Gwen settings, environment configurations, Gherkin feature files, and Gwen meta files in a single workspace that can easily be pulled down and executed on any machine.


For help, open a command prompt in the workspace root and type:

  • gwen help (on windows platforms)
  • ./gwen help (on linux platforms)

Each folder in the workspace also includes a README.txt file that can help to guide you.

Gwen and Selenoid reduced test execution from 50 minutes to 5

Thats 1000% (10 times) quicker than executing features in sequence!!

Gwen has always had the capability to make use of selenium grid which is particularly useful when running in parallel, and after discovering selenoid by the guys over at Aerokube I had to give it a shot.

Installing selenoid was a breeze with the configuration manager docker setup.  Once installed, selenoid downloads the last two driver versions of chrome, firefox and opera, and starts a grid ready to run tests.

Gwen comes with 235 features out of the box, ready for anyone to try.  These features cover a number of different scenarios including a challenge, navigating etsy, completing a todo, navigating blogs, and searching in google.    The features are also written using both imperative and declarative styles declarative-vs-imperative-gherkin.

Typically running 235 features takes 49 minutes and 33 seconds when running the features sequentially, however running these in parallel with selenoid took only 4 minutes and 41 seconds.  Both samples were run using a 34 core machine.


Gwen running 235 features sequentially
Gwen running 235 features in parallel

I have also included a screenshot of the running docker images, taken at a point in time where parallel was running full swing.

Thats all for now, hope you enjoyed a brief look into Selenoid and Gwen.