Testland
Browse all skills & agents

pester-cli-testing

Configures Pester v5 for testing PowerShell CLIs, scripts, and cmdlets - Describe/Context/It blocks, Should assertions, Mock for isolating external dependencies, BeforeAll/BeforeEach setup hooks, Invoke-Pester with PesterConfiguration for tags, code coverage, and NUnit/JUnit XML output in CI. Use when the unit-under-test is a PowerShell script, function, or CLI tool invoked from pwsh on Windows or cross-platform.

Install with skills.sh (any agent)

npx skills add testland/qa --skill pester-cli-testing
View source

pester-cli-testing

Overview

Per pester-install (opens in new window), Pester is the standard PowerShell test framework for unit, integration, and acceptance testing of scripts, functions, modules, and CLI tools invoked from pwsh. It supports Windows PowerShell 3.0 - 5.1 and PowerShell 6.0.4+ across Windows, Linux, and macOS. When the unit-under-test is a shell script or binary on Linux/macOS, use bats-testing instead; Pester is the correct choice whenever the subject or test is PowerShell.

Step 1 - Install

Per pester-install (opens in new window), Windows 10 / Server 2016+ ship with Pester 3.4.0 built-in. The bundled version cannot be updated via Update-Module alone because its publisher certificate differs from the community-signed Gallery version. Install Pester v5 side-by-side:

# Windows (5.1 or pwsh 7+): -Force enables side-by-side; -SkipPublisherCheck
# accepts the newer certificate
Install-Module -Name Pester -Force -SkipPublisherCheck

# Linux / macOS (pwsh 7+): no bundled version to conflict with
Install-Module -Name Pester

# Verify
Import-Module Pester -PassThru

Step 2 - First test file

Per pester-quick-start (opens in new window):

Test files must follow the *.Tests.ps1 naming convention. The outer BeforeAll block dot-sources the script under test so its functions are available to every nested block:

# Get-Greeting.Tests.ps1
BeforeAll {
    . $PSScriptRoot/Get-Greeting.ps1
}

Describe 'Get-Greeting' {
    It 'returns a greeting for the given name' {
        Get-Greeting -Name 'Alice' | Should -Be 'Hello, Alice!'
    }
}

Run with:

Invoke-Pester -Output Detailed .\Get-Greeting.Tests.ps1

Step 3 - Describe / Context / It

Per pqs (opens in new window), Context and Describe are nearly interchangeable. Use Describe at the top level (function or CLI command name) and Context to group scenarios:

Describe 'Invoke-Deploy' {
    Context 'when the target environment is valid' {
        It 'exits 0 and logs a success message' {
            # ...
        }
    }

    Context 'when credentials are missing' {
        It 'throws a terminating error' {
            # ...
        }
    }
}

Step 4 - Should assertions

Per pqs (opens in new window), Should is the assertion command. Common matchers:

$result | Should -Be 'expected'           # strict equality
$result | Should -BeExactly 'Expected'    # case-sensitive equality
$result | Should -BeLike '*partial*'      # wildcard match
$result | Should -Match 'regex\d+'        # regex match
$result | Should -BeNullOrEmpty           # null or empty string/array
$result | Should -BeGreaterThan 0
{ risky-call } | Should -Throw            # expects a terminating error
{ risky-call } | Should -Throw '*message*' # error message wildcard
$result | Should -Not -Be $null           # negated form

Step 5 - BeforeAll / BeforeEach

Per pqs (opens in new window), lifecycle hooks load fixtures and reset state. BeforeAll runs once per block; BeforeEach runs before every It:

Describe 'Get-Report' {
    BeforeAll {
        . $PSScriptRoot/Get-Report.ps1
        $script:TempDir = Join-Path ([System.IO.Path]::GetTempPath()) ([System.IO.Path]::GetRandomFileName())
        New-Item -ItemType Directory -Path $script:TempDir | Out-Null
    }

    AfterAll {
        Remove-Item -Recurse -Force $script:TempDir
    }

    BeforeEach {
        # reset any per-test state here
    }

    It 'writes a report file to the output directory' {
        Get-Report -OutputPath $script:TempDir
        Test-Path (Join-Path $script:TempDir 'report.csv') | Should -BeTrue
    }
}

Use $script: scope when a variable set in BeforeAll must be read inside It blocks. Variables declared with plain $var in BeforeAll are not visible inside It due to PowerShell scoping rules.

Step 6 - Mock

Per pester-mock (opens in new window):

"Mock mocks the behavior of an existing command with an alternate implementation."

Describe 'Send-Notification' {
    BeforeAll {
        . $PSScriptRoot/Send-Notification.ps1

        Mock Invoke-RestMethod {
            return @{ status = 'ok' }
        }

        Mock Write-Error {}    # suppress error output in test runs
    }

    It 'calls Invoke-RestMethod once with the correct URI' {
        Send-Notification -Message 'deploy done'
        Should -Invoke Invoke-RestMethod -Times 1 -Exactly
    }

    It 'passes a ParameterFilter to scope the mock to specific arguments' {
        Mock Get-Date { return [datetime]'2024-01-01' } -ParameterFilter {
            $Format -eq 'yyyy-MM-dd'
        }
        $result = Get-FormattedDate
        $result | Should -Be '2024-01-01'
    }
}

Per pm (opens in new window), mocks placed in BeforeAll apply to all It blocks in the enclosing Describe/Context; a mock inside an It block scopes only to that test. Use -Scope to override. Should -Invoke verifies call count:

Should -Invoke -CommandName Invoke-RestMethod -Times 1 `
    -ParameterFilter { $Uri -like '*api.example.com*' }

Per pm (opens in new window), $PesterBoundParameters (available since Pester 5.2.0) replaces $PSBoundParameters inside mock script blocks.

Step 7 - Tags

Per pester-tags (opens in new window):

Tags can be applied to Describe, Context, and It blocks. Run a subset by filtering on tags:

Describe 'Invoke-Deploy' -Tag 'Integration' {
    Context 'slow path' -Tag 'Slow' {
        It 'completes a full deployment cycle' -Tag 'E2E' {
            # ...
        }
    }
}
# Run only Integration tests, excluding slow ones
Invoke-Pester $path -TagFilter 'Integration' -ExcludeTagFilter 'Slow', 'WindowsOnly'

Per pt (opens in new window), tag matching uses -like comparison, so wildcards work:

Invoke-Pester $path -ExcludeTagFilter 'Slow*', '*Only'

Step 8 - Invoke-Pester with PesterConfiguration

Per pester-config (opens in new window):

New-PesterConfiguration returns a typed configuration object. Cast from a hashtable for concise setup:

$config = [PesterConfiguration]@{
    Run    = @{
        Path = '.\tests'
        Exit = $true          # non-zero exit code on failure (required for CI)
    }
    Filter = @{
        Tag        = 'Unit'
        ExcludeTag = 'Slow', 'WindowsOnly'
    }
    Output = @{
        Verbosity = 'Detailed'
    }
}
Invoke-Pester -Configuration $config

Key Run properties and their defaults (Run.Path, Run.ExcludePath, Run.Exit, Run.TestExtension) are tabulated in references/ci-and-config.md.

Step 9 - Code coverage

Per pester-coverage (opens in new window):

$config = New-PesterConfiguration
$config.Run.Path                      = '.\tests'
$config.CodeCoverage.Enabled          = $true
$config.CodeCoverage.Path             = '.\src'
$config.CodeCoverage.CoveragePercentTarget = 75   # fail if below 75%
$config.CodeCoverage.OutputFormat     = 'JaCoCo'  # or 'CoverageGutters'
$config.CodeCoverage.OutputPath       = 'coverage.xml'
Invoke-Pester -Configuration $config

Per pcov (opens in new window), Pester does not traverse directories automatically for coverage; set CodeCoverage.Path explicitly when source files are not co-located with tests.

Step 10 - Test result XML and CI

Per pester-results (opens in new window):

$config = New-PesterConfiguration
$config.Run.Path               = '.\tests'
$config.Run.Exit               = $true
$config.TestResult.Enabled     = $true
$config.TestResult.OutputFormat = 'NUnitXml'   # or 'JUnitXml'
$config.TestResult.OutputPath  = 'testResults.xml'
Invoke-Pester -Configuration $config

Full CI pipeline definitions - a single cross-platform GitHub Actions matrix (with code coverage and artifact upload) and the Azure DevOps Publish Test Results task - are in references/ci-and-config.md.

Anti-patterns

Anti-patternWhy it failsFix
Dot-sourcing the script under test inside ItRe-executes on every test; slow and leaks stateDot-source once in BeforeAll
Using plain $var in BeforeAll, reading in ItPowerShell scope: It does not inherit BeforeAll localsUse $script:var
Mocking in It when needed across multiple testsMock scopes to that single test onlyMove mock to BeforeAll or BeforeEach
Omitting Run.Exit = $true in CI configPester exits 0 even on failures; CI pipeline passes on broken testsSet Run.Exit = $true
Setting CodeCoverage.Path = '.' without explicit pathsPester may miss source files not co-located with testsSet CodeCoverage.Path to the source directory explicitly
Using legacy Invoke-Pester -Script syntaxRemoved in Pester v5; breaks silently on older runnersUse PesterConfiguration with Run.Path

Limitations

  • Mocking scope is version-sensitive. Pester v5 scopes mocks to their placement block (not the entire Describe). Code written for v4 may need mock placement adjusted.
  • No built-in parallel execution. Pester v5 runs tests serially within a session. Parallelism requires splitting test files across separate pwsh processes or using a CI matrix.
  • Non-PowerShell binaries. Pester can invoke any executable, but assertion helpers are PowerShell-centric. For deep exit-code / stdout testing of Unix binaries, bats-testing is more idiomatic.
  • Windows PowerShell 5.1 vs pwsh 7+. Some Pester 5 features (e.g. $PesterBoundParameters) require a minimum Pester patch level; pin the version in CI with Install-Module -Name Pester -RequiredVersion 5.x.y.

References

  • pester-install (opens in new window) - compatibility, Windows side-by-side install, cross-platform pwsh install.
  • pqs (opens in new window) - quick-start, BeforeAll, Describe/Context/It, Should.
  • pm (opens in new window) - Mock, ParameterFilter, Should -Invoke, scope rules, $PesterBoundParameters.
  • pc (opens in new window) - New-PesterConfiguration, Run, Filter, Output, Invoke-Pester -Configuration.
  • pcov (opens in new window) - CodeCoverage.Enabled, CoveragePercentTarget, formats, path scoping.
  • pr (opens in new window) - TestResult.Enabled, OutputFormat, OutputPath, Azure DevOps / AppVeyor CI notes.
  • references/ci-and-config.md - cross-platform GitHub Actions matrix, Azure DevOps publish task, Run property table.
  • bats-testing - use for shell script / Unix binary testing; Pester is the correct choice when the subject or test is PowerShell.
  • cli-output-conventions - what to assert on (stable formats, exit codes, stderr vs stdout contracts).

Pester CI and configuration reference

View source (opens in new window)

Pester CI and configuration reference

Longer CI pipeline definitions and the PesterConfiguration property lookup. The core in-spine examples live in SKILL.md Steps 8 - 10; this file holds the full CI YAML and the property table.

Key Run properties

Per the configuration docs (opens in new window):

PropertyDefaultPurpose
Run.Path'.'Directory or file(s) to discover
Run.ExcludePath(none)Paths to skip
Run.Exit$falseNon-zero exit on failure
Run.TestExtension'.Tests.ps1'File filter for discovery

GitHub Actions (cross-platform matrix)

One matrix job covers all three GitHub-hosted runners via pwsh (PowerShell 7+). On Windows runners -SkipPublisherCheck overrides the bundled Pester's publisher certificate; it is a harmless no-op on Linux/macOS, where no bundled version exists to conflict with.

jobs:
  pester:
    strategy:
      matrix:
        os: [windows-latest, ubuntu-latest, macos-latest]
    runs-on: ${{ matrix.os }}
    steps:
      - uses: actions/checkout@v4
      - name: Install Pester
        shell: pwsh
        run: Install-Module -Name Pester -Force -SkipPublisherCheck
      - name: Run tests
        shell: pwsh
        run: |
          $config = New-PesterConfiguration
          $config.Run.Path               = '.\tests'
          $config.Run.Exit               = $true
          $config.TestResult.Enabled     = $true
          $config.TestResult.OutputFormat = 'NUnitXml'
          $config.TestResult.OutputPath  = 'testResults.xml'
          $config.CodeCoverage.Enabled   = $true
          $config.CodeCoverage.Path      = '.\src'
          $config.CodeCoverage.CoveragePercentTarget = 75
          Invoke-Pester -Configuration $config
      - uses: actions/upload-artifact@v4
        if: always()
        with:
          name: pester-results
          path: |
            testResults.xml
            coverage.xml

Azure DevOps

Per the test-results docs (opens in new window), after running Pester with NUnitXml output, add a Publish Test Results task with the NUnit format:

- task: PublishTestResults@2
  inputs:
    testResultsFormat: NUnit
    testResultsFiles: testResults.xml

Related skills

bats-testing

Configures Bats-core (Bash Automated Testing System) for testing CLI tools, shell scripts, and Unix programs - `.bats` test files with `@test` blocks, `run` to capture command exit + output, `[ "$status" -eq 0 ]` and `[ "$output" = ... ]` assertions, `setup`/`teardown` hooks, `load` for shared helpers, parallel execution via `--jobs N`, TAP-compliant output for CI integration. Use whenever the unit-under-test is a shell script, CLI binary, or anything invokable from Bash.

cli-output-conventions

Conventions for designing AND testing CLI output so it stays parseable and assertable - exit-code policy (0 success, non-zero failure with stable codes per failure mode), `stdout` for primary data / `stderr` for messages, `--json` / `--plain` for machine-readable output, deterministic ordering and timestamps, `NO_COLOR` / TTY-aware color, `-q` / `--verbose` discipline, and stable `--help` / `--version`. Built on the [Command Line Interface Guidelines][clig]. Use when designing or testing a CLI's output - a new flag, command, or error message, writing CLI assertions, or fixing flaky tests caused by non-deterministic output; it is the assertion contract for `bats-testing` and tells `tui-snapshot-tester` what does NOT need a snapshot.

tui-snapshot-tester

Snapshot testing for terminal UI apps (TUIs) - captures rendered terminal frames as deterministic SVG / text snapshots, diffs them on every run, surfaces a reviewable HTML report on failure, and supports `--snapshot-update` to accept changes intentionally. Wraps `pytest-textual-snapshot` for Python Textual apps; provides equivalent recipes for Ratatui (Rust) `insta` snapshots, Charm Bracelet (Go) `teatest` golden files, and Ink (Node) `ink-testing-library`. Use for any TUI where layout regressions otherwise reach users via `screenshot looks wrong in terminal`.