create-test

name	create-test
description	Create backend tests for Rundeck (unit, API, functional). Auto-loads testing guidelines. For frontend unit tests use write-jest-tests.

Create Test Skill

When to Use

• Adding backend unit tests (Spock)
• Implementing test-first workflow
• Adding API tests for new/modified endpoints
• Writing functional/integration tests
• Fixing bugs (write failing test first, then fix)

For frontend unit tests: Use write-jest-tests skill

Process

Phase 1: Load Context

Automatically read:

• .claude/docs/testing-guidelines.md - Complete testing guide (all test types)
• .claude/docs/development-guidelines.md - Code standards, testing philosophy

Phase 2: Identify Test Type

Determine which test type(s) are needed:

Unit Tests - Functions, classes, services

• Spock (Groovy)
• Mock dependencies
• Test in isolation

API Tests - REST endpoints

• Testcontainers + Spock
• Test API version requirements
• Test request/response formats
• Located in module tests

Functional Tests - Feature integration with Docker

• Complete workflows with Rundeck in Docker
• Use API for setup/validation
• Test plugins, cluster, runners
• Located in functional-test

Integration Tests - External systems

• Docker containers (Testcontainers)
• LDAP, Email, Ansible, etc.
• Test external integrations

Phase 3: Write Unit Tests (Spock)

Spock Framework

Structure: Use given/when/then blocks

class UserServiceSpec extends Specification {
    
    def userService = new UserService()
    
    def "should return user when valid ID provided"() {
        given: "a valid user ID"
        def userId = 123
        
        when: "finding user by ID"
        def user = userService.findById(userId)
        
        then: "user is returned with correct ID"
        user != null
        user.id == userId
    }
    
    def "should throw exception when user not found"() {
        given: "an invalid user ID"
        def userId = 999
        
        when: "finding user by ID"
        userService.findById(userId)
        
        then: "exception is thrown"
        thrown(UserNotFoundException)
    }
}

Mocking with Spock:

def "should call external service"() {
    given: "a mocked external service"
    def mockService = Mock(ExternalService)
    def userService = new UserService(mockService)
    
    when: "getting user data"
    userService.getUserData(123)
    
    then: "external service is called once"
    1 * mockService.fetchData(123) >> "user data"
}

Data-Driven Tests with @Unroll:

@Unroll
def "should validate #scenario"() {
    expect:
    validator.isValid(input) == expected
    
    where:
    scenario        | input    | expected
    "valid email"   | "a@b.c"  | true
    "invalid email" | "abc"    | false
}

Phase 4: Write API Tests

Framework: Testcontainers + Spock

Purpose: Test Rundeck server API behavior with real server

CRITICAL: Test API version requirements

class ProjectApiSpec extends Specification {
    
    @Shared
    def container = new RundeckContainer()
    
    def "should return projects for API v44"() {
        when: "calling API v44"
        def response = client.get("/api/44/projects")
        
        then: "projects are returned"
        response.status == 200
        response.json.size() > 0
    }
    
    def "should reject for API v43"() {
        when: "calling API v43"
        def response = client.get("/api/43/projects")
        
        then: "request is rejected"
        response.status == 404
    }
}

Test Requirements:

1. ✅ New behavior works with NEW API version
2. ❌ New behavior does NOT work with OLD API version
3. ✅ Error conditions return correct responses
4. ✅ Success conditions work in all call patterns
5. ✅ Test multiple content types (JSON, XML)
6. ✅ Test multiple input formats

Phase 5: Write Functional Tests

Purpose: Test complete features/workflows with Rundeck running in Docker

Functional tests run in functional-test module using Testcontainers to launch Rundeck instances and verify functionality through APIs.

Test Annotations

@APITest - Standard functional tests

• Extends BaseContainer
• Single Rundeck instance in Docker
• Tests features via API calls

Base Test Structure

package org.rundeck.functional.api.myfeature

import org.rundeck.util.common.jobs.JobUtils
import org.rundeck.annotations.APITest
import org.rundeck.util.container.BaseContainer

@APITest
class MyFeatureSpec extends BaseContainer {
    
    static final String TEST_PROJECT = "test-project"
    
    def setupSpec() {
        startEnvironment()  // Starts Rundeck in Docker
        setupProject(TEST_PROJECT)  // Creates test project
        
        // Optional: Configure feature flags
        def config = [
            [
                "key": "rundeck.feature.myFeature.enabled",
                "value": "true",
                "strata": "default"
            ]
        ]
        client.post("/config/save", config, Map)
        waitFeatureFlag("rundeck.feature.myFeature.enabled")
    }
    
    def cleanupSpec() {
        deleteProject(TEST_PROJECT)
    }
    
    def "feature works as expected"() {
        given: "test data setup"
        // Setup via API
        
        when: "performing action"
        def response = client.get("/api/endpoint", Map)
        
        then: "expected result"
        response.status == "success"
    }
}

Key Methods

Environment Setup:

• startEnvironment() - Starts Rundeck container
• setupProject(projectName) - Creates project
• deleteProject(projectName) - Deletes project (cleanup)

API Calls:

• client.get(path, Type) - GET request, parse as Type
• client.post(path, body, Type) - POST request
• get(path, Type) - Shorthand for client.get
• doGet(path) - Raw HTTP GET
• doPost(path, body) - Raw HTTP POST

Utilities:

• waitFeatureFlag(key) - Wait for config to propagate
• loadKey(path, value, type) - Create key storage entry
• hold(seconds) - Wait fixed time (use sparingly)
• jsonValue(body) - Parse JSON response

Job Operations (JobUtils):

• JobUtils.createJob(project, yaml, client) - Import job
• JobUtils.waitForExecution(status, execId, client, timeout) - Wait for execution
• JobUtils.getExecutionOutputText(execId, client) - Get logs

Example: Basic API Test

@APITest
class SystemInfoSpec extends BaseContainer {
    
    def setupSpec() {
        startEnvironment()
    }
    
    def "system info returns valid response"() {
        when:
        def data = get("/system/info", Map)
        
        then:
        !data.error
        data.system.rundeck.apiversion.toInteger() >= 14
    }
}

Example: Plugin Integration Test

@APITest
class JiraPluginSpec extends BaseContainer {
    
    static final String TEST_PROJECT = "jira-test"
    
    def setupSpec() {
        startEnvironment()
        setupProject(TEST_PROJECT)
        
        // Configure Jira URL at system level
        def config = [
            ["key": "jira.url", "value": "https://jira.example.com", "strata": "default"]
        ]
        client.post("/config/save", config, Map)
        waitFeatureFlag("jira.url", "https://jira.example.com")
        
        // Store credentials
        loadKey("jira.pass", "test-password", "password")
    }
    
    def cleanupSpec() {
        deleteProject(TEST_PROJECT)
    }
    
    def "jira plugin reads system config"() {
        given: "a job using Jira plugin"
        def jobXml = """<joblist>
          <job>
            <name>jira test</name>
            <sequence>
              <command>
                <node-step-plugin type='jira-create-issue'>
                  <configuration>
                    <entry key='summary' value='Test Issue' />
                    <entry key='password' value='keys/jira.pass' />
                  </configuration>
                </node-step-plugin>
              </command>
            </sequence>
          </job>
        </joblist>"""
        
        when: "importing and running the job"
        def importResult = JobUtils.createJob(TEST_PROJECT, jobXml, client)
        def jobId = importResult.succeeded[0].id
        def response = doPost("/job/${jobId}/executions", ["loglevel": "DEBUG"])
        def execId = jsonValue(response.body()).id as String
        
        then: "job starts successfully"
        execId != null
        
        when: "waiting for completion"
        def exec = JobUtils.waitForExecution(
            ExecutionStatus.FAILED.state,  // Expected to fail (no real Jira)
            execId,
            client,
            WaitingTime.EXCESSIVE
        )
        
        then: "execution attempted to use system config URL"
        exec.status == 'failed'
        def fullLog = JobUtils.getExecutionOutputText(execId, client)
        fullLog.contains("jira.example.com")
    }
}

Important Patterns

Test Data Setup:

• Use API to create test data (jobs, keys, config)
• Don't rely on UI or manual setup
• Use setupProjectArchiveDirectoryResource() for complex project setup

Waiting:

• Use waitFeatureFlag() for config propagation
• Use JobUtils.waitForExecution() for job completion
• Avoid hold() except for cluster sync scenarios

Assertions:

• Test through API responses
• Verify job execution logs with getExecutionOutputText()
• Check execution status, output, side effects

Cleanup:

• Always cleanup in cleanupSpec()
• Delete projects, jobs, keys created
• Tests should not leave artifacts

Phase 6: Run Tests

# Backend unit tests
./gradlew test

# Specific module
./gradlew :module-name:test

# API tests
./gradlew :functional-test:apiTest

# Functional tests
./gradlew :functional-test:test

---

Testing Philosophy

How do you know you fixed the problem?

1. Write a test that exemplifies the problem
2. Test should FAIL before implementing fix
3. After fix, test should PASS

How do you know new behavior works?

Update existing tests to cover new behaviors

What if code has no existing tests?

CRITICAL: Add missing tests for existing code FIRST

1. Write tests for existing functionality
2. Verify code works as expected
3. You've now established baseline
4. Future changes that break functionality → tests fail

---

Checklist

Before considering tests complete:

• Test type identified (Unit, API, Functional)
• Testing guidelines loaded and followed
• Tests written before implementation
• Tests follow proper patterns (given/when/then, Spock structure)
• All tests pass locally
• Unit tests: Proper mocking, test isolation
• API tests: Version requirements tested, error conditions covered
• Functional tests: Proper annotation (@APITest), cleanup in cleanupSpec()
• Edge cases covered
• Error conditions tested
• No flaky tests introduced
• Test names clearly describe what they test

---

Common Mistakes to Avoid

❌ Don't:

• Skip tests for "simple" changes
• Write implementation before tests
• Test implementation details instead of behavior
• Leave commented-out test code
• Create flaky tests (random pass/fail)
• Over-mock (test real behavior when possible)
• Functional tests: Forget cleanupSpec() - leaves test data
• Functional tests: Use hold() instead of proper waits
• Functional tests: Create test data through UI (slow, brittle)
• API tests: Test only new API version (must test old version rejects)

✅ Do:

• Write test before implementation
• Test edge cases and error conditions
• Use clear test names that describe behavior
• Keep tests simple and focused
• Convert JUnit tests to Spock when modifying
• Make tests independent (no execution order dependencies)
• Unit tests: Mock at system boundaries, not internal interfaces
• API tests: Test version requirements (new works, old rejects)
• Functional tests: Setup via API, cleanup in cleanupSpec()
• Functional tests: Use proper annotations (@APITest)

---

Integration with Other Skills

• create-code: Use this skill when implementing test-first workflow
• create-api-endpoint: API endpoints MUST have API tests (test version requirements)
• create-plugin: Plugins should have unit tests for functionality

---

Resources

• .claude/docs/testing-guidelines.md - Complete testing guide
• .claude/docs/development-guidelines.md - Testing philosophy, code standards
• Spock Framework - Groovy testing framework
• Testcontainers - Docker containers for testing
• Example tests: functional-test/src/test/groovy/