Generating Apex Tests
Generate production-ready Apex test classes and run disciplined test-fix loops with coverage analysis.
Core Principles
- One behavior per method — each test method validates a single scenario. Separate positive, negative, and bulk tests. NEVER combine related-but-distinct inputs (e.g., null and empty) in one method — create
_NullInput_ and _EmptyInput_ as separate test methods
- Bulkify tests — test with 251+ records to cross the 200-record trigger batch boundary. Batch Apex exception: in test context only one
execute() invocation runs, so set batchSize >= testRecordCount. See references/async-testing.md
- Isolate test data — every
@TestSetup must delegate record creation to a TestDataFactory class. If none exists, create one first. Never build record lists inline in @TestSetup. Never rely on org data (SeeAllData=false) or hardcoded IDs. For duplicate rule handling, see references/test-data-factory.md
- Assert meaningfully — use exact expected values computed from test data setup. NEVER use range assertions or approximate counts when the value is deterministic. Always include failure messages. See references/assertion-patterns.md
- Use
Assert class only — Assert.areEqual, Assert.isTrue, Assert.fail, etc. Never use legacy System.assert, System.assertEquals, or System.assertNotEquals
- Mock external boundaries — use
HttpCalloutMock for callouts, Test.setFixedSearchResults for SOSL, DML mock classes for database isolation. Design for testability via constructor injection. See references/mocking-patterns.md
- Test negative paths — validate error handling and exception scenarios, not just happy paths
- Wrap with start/stop — pair
Test.startTest() with Test.stopTest() to reset governor limits and force async execution
Test.startTest() / Test.stopTest()
Always wrap the code under test in Test.startTest() / Test.stopTest():
- Resets governor limits so the test measures only the code under test
- Executes async operations synchronously (queueables, batch, future methods)
- Fires scheduled jobs immediately
Test Code Anti-Patterns
| Anti-Pattern | Fix |
|---|
| SOQL/DML inside loops | Query once before the loop; use Map<Id, SObject> for lookups |
| Magic numbers in assertions | Derive expected values from setup constants |
| God test class (>500 lines) | Split into multiple test classes by behavior area |
| Long test methods (>30 lines) | Extract Given/When/Then into helper methods |
Generic Exception catch | Catch the specific expected type (e.g., DmlException) |
Workflow
Step 1 — Gather Context
Before generating or fixing tests, identify:
- the target production class(es) under test
- existing test classes, test data factories, and setup helpers
- desired test scope (single class, specific methods, suite, or local tests)
- coverage threshold (75% minimum for deploy, 90%+ recommended)
- org alias when running tests against an org
Step 2 — Generate the Test Class
Apply the structure, naming conventions, and patterns from the asset templates and reference docs.
MANDATORY — File Deliverables: For every test class, create BOTH files:
{ClassName}Test.cls — the test class (use assets/test-class-template.cls as starting point){ClassName}Test.cls-meta.xml — the metadata file:
<?xml version="1.0" encoding="UTF-8"?>
<ApexClass xmlns="http://soap.sforce.com/2006/04/metadata">
<apiVersion>66.0</apiVersion>
<status>Active</status>
</ApexClass>
If no TestDataFactory exists in the project, create TestDataFactory.cls + TestDataFactory.cls-meta.xml using assets/test-data-factory-template.cls.
@TestSetup Example
@TestSetup
static void setupTestData() {
List<Account> accounts = TestDataFactory.createAccounts(251, true);
}
Test Method Structure
Use Given/When/Then:
@isTest
static void shouldUpdateStatus_WhenValidInput() {
// Given
List<Account> accounts = [SELECT Id FROM Account];
// When
Test.startTest();
MyService.processAccounts(accounts);
Test.stopTest();
// Then
List<Account> updated = [SELECT Id, Status__c FROM Account];
Assert.areEqual(251, updated.size(), 'All accounts should be processed');
}
Negative Test — Exception Pattern
Use try/catch with Assert.fail to verify expected exceptions:
@isTest
static void shouldThrowException_WhenInvalidInput() {
// Given
List<Account> emptyList = new List<Account>();
// When/Then
Test.startTest();
try {
MyService.processAccounts(emptyList);
Assert.fail('Expected MyCustomException to be thrown');
} catch (MyCustomException e) {
Assert.isTrue(e.getMessage().contains('cannot be empty'),
'Exception message should indicate empty input');
}
Test.stopTest();
}
Naming Convention
should[ExpectedResult]_When[Scenario]: shouldSendNotification_WhenOpportunityClosedWon[SubjectOrAction]_[Scenario]_[ExpectedResult]: AccountUpdate_ChangeName_Success
Step 3 — Run Tests
Start narrow when debugging; widen after the fix is stable.
# Single test class
sf apex run test --class-names MyServiceTest --result-format human --code-coverage --target-org <alias>
# Specific test methods
sf apex run test --tests MyServiceTest.shouldUpdateStatus_WhenValidInput --result-format human --target-org <alias>
# All local tests
sf apex run test --test-level RunLocalTests --result-format human --code-coverage --target-org <alias>
Step 4 — Analyze Results
Focus on:
- failing methods — exception types and stack traces
- uncovered lines and weak coverage areas
- whether failures indicate bad test data, brittle assertions, or broken production logic
- when class metadata crosses from versions 66.0 and below to 67.0 and above, check explicit sharing, decide user/system mode per operation, grant required CRUD/FLS in test users or permission sets, and rerun affected tests
Step 5 — Fix Loop
When tests fail, run a disciplined fix loop (max 3 iterations — stop and surface root cause if still failing):
- Read the failing test class and the class under test
- Identify root cause from error messages and stack traces
- Apply fix — adjust test data or assertions for test-side issues; delegate production code issues to the
platform-apex-generate skill
- Rerun the focused test before broader regression
- Repeat until all tests pass, iteration limit reached, or root cause requires design change
Step 6 — Validate Coverage
| Level | Coverage | Purpose |
|---|
| Production deploy | 75% minimum | Required by Salesforce |
| Recommended | 90%+ | Best practice target |
| Critical paths | 100% | Business-critical code |
Cover all paths: positive, negative/exception, bulk (251+ records), callout/async.
What to Test by Component
| Component | Key Test Scenarios |
|---|
| Trigger | Bulk insert/update/delete, recursion guard, field change detection |
| Service | Valid/invalid inputs, bulk operations, exception handling |
| Controller | Page load, action methods, view state |
| Batch | start/execute/finish, scope matching (batch size >= record count), Database.Stateful tracking, error handling, chaining (separate methods — finish() calling Database.executeBatch() throws UnexpectedException) |
| Queueable | Chaining (only first job runs in tests), bulkification, error handling, callout mocks before Test.startTest() |
| Callout | Success response, error response, timeout |
| Selector | Valid/null/empty inputs, bulk (251+), field population, sort order, WITH USER_MODE via System.runAs |
| Scheduled | Direct execution via execute(null), CRON registration via CronTrigger query |
| Platform Event | Test.enableChangeDataCapture(), Test.getEventBus().deliver(), verify subscriber side effects |
Output Expectations
Deliverables per test class:
{ClassName}Test.cls + {ClassName}Test.cls-meta.xml (match API version of class under test; default 66.0)TestDataFactory.cls + TestDataFactory.cls-meta.xml (if not already present)
Reference Files
Load on demand for detailed patterns:
| Reference | When to use |
|---|
| references/test-data-factory.md | TestDataFactory patterns, field overrides, duplicate rule handling |
| references/assertion-patterns.md | Assertion best practices, anti-patterns, common pitfalls |
| references/mocking-patterns.md | HttpCalloutMock, DML mocking, StubProvider, SOSL, Email, Platform Events |
| references/async-testing.md | Batch, Queueable, Future, Scheduled job testing |
