Jenkins

Q: The job fails with "sh not found"

Windows has no sh built in. Either add Git for Windows' sh.exe to the machine PATH and restart the Jenkins service, or rewrite the pipeline's sh steps as bat.

Q: The job fails with "npm is not recognized" even after fixing sh

Jenkins runs as a Windows service that only sees the machine-wide PATH. Check whether Node is on it, and add Node's folder to the machine PATH the same way as the sh fix if it isn't.

Run Automate from a Jenkins pipeline with the access key in Jenkins credentials.

Store the access key

Manage Jenkins → Credentials → (global) → Add Credentials.
Kind Secret text, Secret = your TestKase access key, ID testkase-access-key. Save.

Add the job

Save this as Jenkinsfile in your repository, replacing [email protected] with your TestKase username:

One-time setup: run this locally and commit the result, before the pipeline needs it.

npm install --save-dev @testkase/playwright-reporter

The CI job only installs what's already committed — it can't add packages on its own. Skip this and the job either fails (no internet access) or runs slow (fetches the package fresh every time).

pipeline {
  agent any
  environment {
    TK_USERNAME    = '[email protected]'
    ACCESS_KEY     = credentials('testkase-access-key')   // Secret text credential
    TESTKASE_BUILD = "ci-${BUILD_NUMBER}"
  }
  stages {
    stage('E2E on TestKase') {
      steps {
        sh 'npm ci'
        sh 'npx testkase-playwright test'
      }
    }
  }
}

pipeline {
  agent any
  environment {
    TK_USERNAME    = '[email protected]'
    ACCESS_KEY     = credentials('testkase-access-key')   // Secret text credential
    TESTKASE_BUILD = "ci-${BUILD_NUMBER}"
  }
  stages {
    stage('E2E on TestKase') {
      steps {
        sh 'npm ci'
        sh 'npm test'   // your Selenium suite — see the note below
      }
    }
  }
}

pipeline {
  agent any
  environment {
    TK_USERNAME    = '[email protected]'
    ACCESS_KEY     = credentials('testkase-access-key')   // Secret text credential
    TESTKASE_BUILD = "ci-${BUILD_NUMBER}"
  }
  stages {
    stage('E2E on TestKase') {
      steps {
        sh 'npm ci'
        sh 'npm test'   // your Appium suite — see the note below
      }
    }
  }
}

Playwright runs through the testkase-playwright wrapper, which reads these environment variables for you. Selenium and Appium have no wrapper — your existing test code builds the Hub URL from TK_USERNAME / ACCESS_KEY itself, exactly as it already does outside CI. See Selenium: point your suite at TestKase and Appium: getting started.

What to check


Build name	`ci-${BUILD_NUMBER}` — sessions group under it on the Automate dashboard
Secret	read from `credentials('testkase-access-key')` into `ACCESS_KEY`; never in the repository
Tunnel	add `npx testkase-tunnel --token "$ACCESS_KEY" &` before the test step if the tests open a URL the farm can't reach

Jenkins masks any value bound with credentials() in the console log automatically.

Windows agents

The Jenkinsfile above uses sh, a Linux/macOS shell step, so a Windows agent — the machine that runs the job, not necessarily the machine hosting Jenkins — hits one of two failures. Both are one-time fixes, per agent.

▶The job fails with "sh not found"

Pick one.

Option A — add Git's shell to the PATH. Git for Windows ships sh.exe. Point the machine at it and the Jenkinsfile above needs no changes.

Manage Jenkins → System → Shell executable doesn't apply here. That setting is for Freestyle jobs only. This Jenkinsfile is a Pipeline job, and a Pipeline sh step looks for sh.exe on the machine-wide PATH — nowhere else.

Install Git for Windows on the agent, if it isn't there already.

Open PowerShell as Administrator and run:

$path = [Environment]::GetEnvironmentVariable('Path', 'Machine')
[Environment]::SetEnvironmentVariable('Path', "$path;C:\Program Files\Git\usr\bin", 'Machine')
Restart-Service Jenkins
Start-Sleep 5
(Get-Service Jenkins).Status

Wait about a minute, then click Build Now.

Option B — use bat instead of sh. No install, no restart, but every step changes:

pipeline {
  agent any
  environment {
    TK_USERNAME    = '[email protected]'
    ACCESS_KEY     = credentials('testkase-access-key')
    TESTKASE_BUILD = "ci-${BUILD_NUMBER}"
  }
  stages {
    stage('E2E on TestKase') {
      steps {
        bat 'npm ci'
        bat 'npx testkase-playwright test'
      }
    }
  }
}

▶The job fails with "npm is not recognized" even after fixing sh

Jenkins usually runs as a Windows service under LocalSystem, not as you. It only sees the machine-wide PATH — never your own. That's why npm works in your terminal but the job still fails.

Check whether Node is on the machine-wide PATH:

$path = [Environment]::GetEnvironmentVariable('Path', 'Machine')
$path -split ';' | Select-String -Pattern node

Prints a path, for example C:\Program Files\nodejs\ — Node is already visible to the service; look elsewhere for the failure.

Prints nothing — Node is on your user PATH only, or not installed. Find its folder (run Get-Command node in a terminal where node already works, or check C:\Program Files\nodejs), then add it the same way as the sh fix:

$path = [Environment]::GetEnvironmentVariable('Path', 'Machine')
[Environment]::SetEnvironmentVariable('Path', "$path;C:\Program Files\nodejs", 'Machine')
Restart-Service Jenkins

If Node isn't on the agent at all, install it as administrator — the official installer writes to the machine PATH by default — then restart the service the same way.

Run the pipeline once. Within a few seconds the session appears on Automate → Sessions; the job goes red if any test fails.

Store the access key

Add the job

What to check

Windows agents

On this page