agentleFS
VincentH-NetVincentH-Net/dotnet-agentic-engineering/.agents/skills/writing-mstest-tests/SKILL.md

writing-mstest-tests

Skill9 starsChanged 6 months agoMITView on GitHub

Best practices for writing new MSTest 3.x/4.x unit tests and implementing concrete fixes in existing MSTest code. Use when the user asks to write, create, implement, repair, or modernize tests (including fix-it prompts such as 'something seems off, fix issues'). Primary fit for direct code changes like correcting swapped Assert.AreEqual argument order, replacing outdated assertion patterns, and converting DynamicData from IEnumerable<object[]> to ValueTuple-based data sets. Covers modern assertions, data-driven tests, test lifecycle, MSTest.Sdk, sealed classes, Assert.Throws, DynamicData with ValueTuples, TestContext, and conditional execution. Do NOT use for broad test quality audits, flaky-test investigations, or test smell detection reports — use test-anti-patterns instead.

1 other project uses a version of this, 1 of them edited.

Adopted by

Most recently taken up by

The file

Show its rules and text

We split this skill into the rules it gives an agent

15 rules. About 0 are common, 3 are distinctive, and 12 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈3

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 12

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Check the test project for MSTest version and configuration:

    line 44
  2. Very specific:

    If using MSTest.Sdk (<Sdk Name="MSTest.Sdk">): modern setup, all features available

    line 46
  3. Very specific:

    If using MSTest metapackage: modern setup (MSTest 3.x+)

    line 47
  4. Very specific:

    If using MSTest.TestFramework + MSTest.TestAdapter: check version for feature availability

    line 48
  5. Very specific:

    When using MSTest.Sdk, put the version in global.json instead of the project file so all test projects get bumped together:

    line 61
  6. Very specific:

    Use [TestClass] on the class and [TestMethod] on test methods

    line 88
  7. Distinctive:

    Follow the Arrange-Act-Assert (AAA) pattern

    line 89
  8. Very specific:

    Use separate test projects with naming convention [ProjectName].Tests

    line 91
  9. Very specific:

    Use the correct assertion for each scenario.

    line 115
  10. Very specific:

    Prefer ValueTuple return types over IEnumerable<object[]> for type safety:

    line 207
  11. Very specific:

    When you need metadata per test case, use TestDataRow<T>:

    line 227
  12. Very specific:

    Use [TestInitialize] only for async initialization, combined with the constructor for sync parts

    line 241
  13. Very specific:

    Use [TestCleanup] for cleanup that must run even on failure

    line 242
  14. Very specific:

    Always use TestContext.CancellationToken with [Timeout]:

    line 284
  15. Distinctive:

    Use only for genuinely flaky external dependencies (network, file system), not to paper over race conditions or shared state issues.

    line 300

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Writing MSTest Tests

Help users write effective, modern unit tests with MSTest 3.x/4.x using current APIs and best practices.

When to Use

  • User wants to write new MSTest unit tests
  • User wants to improve or modernize existing MSTest tests by implementing concrete fixes
  • User asks about MSTest assertion APIs, data-driven patterns, or test lifecycle
  • User needs help fixing a specific MSTest test bug or failing assertion
  • User asks to fix swapped Assert.AreEqual argument order (expected first, actual second)
  • User asks to convert DynamicData from IEnumerable<object[]> to ValueTuple-based data

When Not to Use

  • User needs a test quality audit, anti-pattern detection, or flaky-test investigation (use test-anti-patterns)
  • User needs to run or execute tests (use the run-tests skill)
  • User needs to upgrade from MSTest v1/v2 to v3 (use migrate-mstest-v1v2-to-v3)
  • User needs to upgrade from MSTest v3 to v4 (use migrate-mstest-v3-to-v4)
  • User needs CI/CD pipeline configuration
  • User is using xUnit, NUnit, or TUnit (not MSTest)

Inputs

InputRequiredDescription
Code under testNoThe production code to be tested
Existing test codeNoCurrent tests to fix, update, or modernize
Test scenario descriptionNoWhat behavior the user wants to test

Workflow

Step 1: Determine project setup

Check the test project for MSTest version and configuration:

  • If using MSTest.Sdk (<Sdk Name="MSTest.Sdk">): modern setup, all features available
  • If using MSTest metapackage: modern setup (MSTest 3.x+)
  • If using MSTest.TestFramework + MSTest.TestAdapter: check version for feature availability

Recommend MSTest.Sdk or the MSTest metapackage for new projects:

<!-- Option 1: MSTest SDK (simplest, recommended for new projects) -->
<Project Sdk="MSTest.Sdk">
  <PropertyGroup>
    <TargetFramework>net9.0</TargetFramework>
  </PropertyGroup>
</Project>

When using MSTest.Sdk, put the version in global.json instead of the project file so all test projects get bumped together:

{
  "msbuild-sdks": {
    "MSTest.Sdk": "3.8.2"
  }
}
<!-- Option 2: MSTest metapackage -->
<Project Sdk="Microsoft.NET.Sdk">
  <PropertyGroup>
    <TargetFramework>net9.0</TargetFramework>
  </PropertyGroup>
  <ItemGroup>
    <PackageReference Include="MSTest" Version="3.8.2" />
  </ItemGroup>
</Project>

Step 2: Write test classes following conventions

Apply these structural conventions:

  • Seal test classes with sealed for performance and design clarity
  • Use [TestClass] on the class and [TestMethod] on test methods
  • Follow the Arrange-Act-Assert (AAA) pattern
  • Name tests using MethodName_Scenario_ExpectedBehavior
  • Use separate test projects with naming convention [ProjectName].Tests
[TestClass]
public sealed class OrderServiceTests
{
    [TestMethod]
    public void CalculateTotal_WithDiscount_ReturnsReducedPrice()
    {
        // Arrange
        var service = new OrderService();
        var order = new Order { Price = 100m, DiscountPercent = 10 };

        // Act
        var total = service.CalculateTotal(order);

        // Assert
        Assert.AreEqual(90m, total);
    }
}

Step 3: Use modern assertion APIs

Use the correct assertion for each scenario. Prefer Assert class methods over StringAssert or CollectionAssert where both exist.

Equality and null checks

Assert.AreEqual(expected, actual);      // Value equality
Assert.AreSame(expected, actual);       // Reference equality
Assert.IsNull(value);
Assert.IsNotNull(value);

Exception testing -- use Assert.Throws instead of [ExpectedException]

// Synchronous
var ex = Assert.ThrowsExactly<ArgumentNullException>(() => service.Process(null));
Assert.AreEqual("input", ex.ParamName);

// Async
var ex = await Assert.ThrowsExactlyAsync<InvalidOperationException>(
    async () => await service.ProcessAsync(null));
  • Assert.Throws<T> matches T or any derived type
  • Assert.ThrowsExactly<T> matches only the exact type T

Collection assertions

Assert.Contains(expectedItem, collection);
Assert.DoesNotContain(unexpectedItem, collection);
var single = Assert.ContainsSingle(collection);  // Returns the single element
Assert.HasCount(3, collection);
Assert.IsEmpty(collection);
Assert.IsNotEmpty(collection);

Replace generic Assert.IsTrue with specialized assertions -- they give better failure messages:

Instead ofUse
Assert.IsTrue(list.Count > 0)Assert.IsNotEmpty(list)
Assert.IsTrue(list.Count() == 3)Assert.HasCount(3, list)
Assert.IsTrue(x != null)Assert.IsNotNull(x)
list.Single(predicate) + Assert.IsNotNullAssert.ContainsSingle(list)
Assert.IsTrue(list.Contains(item))Assert.Contains(item, list)

String assertions

Assert.Contains("expected", actualString);
Assert.StartsWith("prefix", actualString);
Assert.EndsWith("suffix", actualString);
Assert.MatchesRegex(@"\d{3}-\d{4}", phoneNumber);

Type assertions

// MSTest 3.x -- out parameter
Assert.IsInstanceOfType<MyHandler>(result, out var typed);
typed.Handle();

// MSTest 4.x -- returns directly
var typed = Assert.IsInstanceOfType<MyHandler>(result);

Comparison assertions

Assert.IsGreaterThan(lowerBound, actual);
Assert.IsLessThan(upperBound, actual);
Assert.IsInRange(actual, low, high);

Step 4: Use data-driven tests for multiple inputs

DataRow for inline values

[TestMethod]
[DataRow(1, 2, 3)]
[DataRow(0, 0, 0, DisplayName = "Zeros")]
[DataRow(-1, 1, 0)]
public void Add_ReturnsExpectedSum(int a, int b, int expected)
{
    Assert.AreEqual(expected, Calculator.Add(a, b));
}

DynamicData with ValueTuples (preferred for complex data)

Prefer ValueTuple return types over IEnumerable<object[]> for type safety:

[TestMethod]
[DynamicData(nameof(DiscountTestData))]
public void ApplyDiscount_ReturnsExpectedPrice(decimal price, int percent, decimal expected)
{
    var result = PriceCalculator.ApplyDiscount(price, percent);
    Assert.AreEqual(expected, result);
}

// ValueTuple -- preferred (MSTest 3.7+)
public static IEnumerable<(decimal price, int percent, decimal expected)> DiscountTestData =>
[
    (100m, 10, 90m),
    (200m, 25, 150m),
    (50m, 0, 50m),
];

When you need metadata per test case, use TestDataRow<T>:

public static IEnumerable<TestDataRow<(decimal price, int percent, decimal expected)>> DiscountTestDataWithMetadata =>
[
    new((100m, 10, 90m)) { DisplayName = "10% discount" },
    new((200m, 25, 150m)) { DisplayName = "25% discount" },
    new((50m, 0, 50m)) { DisplayName = "No discount" },
];

Step 5: Handle test lifecycle correctly

  • Always initialize in the constructor -- this enables readonly fields and works correctly with nullability analyzers (fields are guaranteed non-null after construction)
  • Use [TestInitialize] only for async initialization, combined with the constructor for sync parts
  • Use [TestCleanup] for cleanup that must run even on failure
  • Inject TestContext via constructor (MSTest 3.6+)
[TestClass]
public sealed class RepositoryTests
{
    private readonly TestContext _testContext;
    private readonly FakeDatabase _db;  // readonly -- guaranteed by constructor

    public RepositoryTests(TestContext testContext)
    {
        _testContext = testContext;
        _db = new FakeDatabase();  // sync init in ctor
    }

    [TestInitialize]
    public async Task InitAsync()
    {
        // Use TestInitialize ONLY for async setup
        await _db.SeedAsync();
    }

    [TestCleanup]
    public void Cleanup() => _db.Reset();
}

Execution order

  1. [AssemblyInitialize] -- once per assembly
  2. [ClassInitialize] -- once per class
  3. Per test:
    • With TestContext property injection: Constructor -> set TestContext property -> [TestInitialize]
    • With constructor injection of TestContext: Constructor (receives TestContext) -> [TestInitialize]
  4. Test method
  5. [TestCleanup] -> DisposeAsync -> Dispose -- per test
  6. [ClassCleanup] -- once per class
  7. [AssemblyCleanup] -- once per assembly

Step 6: Apply cancellation and timeout patterns

Always use TestContext.CancellationToken with [Timeout]:

[TestMethod]
[Timeout(5000)]
public async Task FetchData_ReturnsWithinTimeout()
{
    var result = await _client.GetDataAsync(_testContext.CancellationToken);
    Assert.IsNotNull(result);
}

Step 7: Use advanced features where appropriate

Retry flaky tests (MSTest 3.9+)

Use only for genuinely flaky external dependencies (network, file system), not to paper over race conditions or shared state issues.

[TestMethod]
[Retry(3)]
public void ExternalService_EventuallyResponds() { }

Conditional execution (MSTest 3.10+)

[TestMethod]
[OSCondition(OperatingSystems.Windows)]
public void WindowsRegistry_ReadsValue() { }

[TestMethod]
[CICondition(ConditionMode.Exclude)]
public void LocalOnly_InteractiveTest() { }

Parallelization

[assembly: Parallelize(Workers = 4, Scope = ExecutionScope.MethodLevel)]

[TestClass]
[DoNotParallelize]  // Opt out specific classes
public sealed class DatabaseIntegrationTests { }

Validation

  • Test classes are sealed
  • Test methods follow MethodName_Scenario_ExpectedBehavior naming
  • Assert.ThrowsExactly<T> used instead of [ExpectedException]
  • Specialized assertions used instead of Assert.IsTrue (e.g., Assert.IsNotNull, Assert.AreEqual)
  • DynamicData uses ValueTuple return types instead of IEnumerable<object[]>
  • Sync initialization done in the constructor, not [TestInitialize]
  • TestContext.CancellationToken passed to async calls in tests with [Timeout]
  • Project builds with zero errors and all tests pass

Common Pitfalls

PitfallSolution
Assert.AreEqual(actual, expected) -- swapped argumentsAlways put expected first: Assert.AreEqual(expected, actual). Failure messages show "Expected: X, Actual: Y" so wrong order makes messages confusing
[ExpectedException] -- obsolete, cannot assert messageUse Assert.Throws<T> or Assert.ThrowsExactly<T>
items.Single() -- unclear exception on failureUse Assert.ContainsSingle(items) for better failure messages
Hard cast (MyType)result -- unclear exceptionUse Assert.IsInstanceOfType<MyType>(result)
IEnumerable<object[]> for DynamicDataUse IEnumerable<(T1, T2, ...)> ValueTuples for type safety
Sync setup in [TestInitialize]Initialize in the constructor instead -- enables readonly fields and satisfies nullability analyzers
CancellationToken.None in async testsUse TestContext.CancellationToken for cooperative timeout
public TestContext? TestContext { get; set; }Drop the ? -- MSTest suppresses CS8618 for this property
TestContext TestContext { get; set; } = null!Remove = null! -- unnecessary, MSTest handles assignment
Non-sealed test classesSeal test classes by default for performance

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.