Testing

🧪 Testing

Test Stack

Tool	Purpose
xUnit	Test framework
NSubstitute	Mocking library
FluentAssertions	Assertion library

---

▶️ Running Tests

# Run all tests
dotnet test

# Run with verbose output
dotnet test --verbosity normal

# Run a specific test class
dotnet test --filter "FullyQualifiedName~CollectionParserTests"

# Run tests from a specific namespace
dotnet test --filter "Namespace~CodeGenesis.Engine.Tests.Claude"

---

📊 Test Coverage

Test Class	Target	Tests
UsePipelineStepTests	Sub-pipeline execution, input mapping, output merging, circular detection	20
DynamicStepTests	Success/failure execution, retry, rate limit, fail_if, clone	20
PipelineConfigLoaderTests	Template resolution, YAML loading, validation	20
ClaudeResponseTests	JSON parsing, FailureKind classification, Failure factory	15
CollectionParserTests	JSON array, comma-separated, newline-separated, fallbacks	12
StepEntryTests	Discriminators (IsSimpleStep, IsForeach, etc.), ToStepConfig	11
McpServerConfigTests	Template resolution, immutability	10
McpContextBuilderTests	MCP tools section generation for system prompts	9
PipelineContextTests	Initialization, metrics accumulation, step outputs	6
StepResultTests	Record equality, outcome values	6
RetryPolicyTests	Cascading resolution (step → global → defaults)	4
	Total	133

---

🔄 CI

Tests run automatically via GitHub Actions on every push to main and on pull requests.

The workflow:

1. ☑️ Sets up .NET 10 SDK (10.0.103)
2. ☑️ Restores dependencies
3. ☑️ Builds in Release configuration
4. ☑️ Runs all tests

---

✍️ Writing New Tests

Conventions

Convention	Example
File location	Mirror source namespace (Claude/, Config/, Pipeline/, Steps/)
Test naming	MethodName_Scenario_ExpectedResult
Mocking	Use NSubstitute for interfaces (IClaudeRunner, IStepExecutor)
Assertions	Use FluentAssertions (result.Should().Be(...))

Example

[Fact]
public async Task ExecuteAsync_Success_ReturnsSuccessResult()
{
    // Arrange
    _claude.RunAsync(Arg.Any<ClaudeRequest>(), Arg.Any<CancellationToken>())
        .Returns(new ClaudeResponse
        {
            Success = true,
            Result = "done!",
            InputTokens = 100,
            OutputTokens = 50,
            CostUsd = 0.01,
            Duration = TimeSpan.FromSeconds(2)
        });

    var step = MakeStep();
    var ctx = MakeContext();

    // Act
    var result = await step.ExecuteAsync(ctx, CancellationToken.None);

    // Assert
    result.Outcome.Should().Be(StepOutcome.Success);
    result.Output.Should().Be("done!");
}

Tip

When testing steps that depend on IClaudeRunner, use NSubstitute.Returns() to simulate different Claude responses (success, timeout, rate limit, max_turns, etc.).