agentleFS
kevmookevmoo/dash_skills/skills/dart-test-fundamentals/SKILL.md

dart-test-fundamentals

Skill148 starsChanged 13 days agoApache-2.0View on GitHub

Core concepts and best practices for `package:test`. Covers `test`, `group`, lifecycle methods (`setUp`, `tearDown`), and configuration (`dart_test.yaml`).

The file

Show its rules and text

We split this skill into the rules it gives an agent

16 rules. About 0 are common, 5 are distinctive, and 11 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈5

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 11

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Use this skill when:

    line 18
  2. Distinctive:

    Do NOT apply this skill or refactor existing tests when:

    line 27
  3. Very specific:

    Legacy Single-Group Churn: Do NOT remove or reformat existing group hierarchies in untouched existing tests unless explicitly asked, as …

    line 29
  4. Very specific:

    Search for tests that use try-finally for cleanup instead of addTearDown:

    line 45
  5. Very specific:

    Naming: Use PascalCase for groups that correspond to a class name (e.g., group('MyClient', ...)).

    line 64
  6. Very specific:

    Avoid Single Groups: Do not wrap all tests in a file with a single group call if it's the only one.

    line 66
  7. Distinctive:

    NOTE: DO NOT remove groups when doing a cleanup on existing code you didn't create unless explicitly asked to.

    line 68
  8. Very specific:

    Avoid redundant "test" prefixes.

    line 73
  9. Very specific:

    Use setUp for resetting state to ensure test isolation.

    line 98
  10. Distinctive:

    Avoid sharing mutable state between tests without resetting it.

    line 99
  11. Very specific:

    Naming: Test files must end in _test.dart to be picked up by the test runner, and should be placed in the test/ directory.

    line 180
  12. Very specific:

    Shared Fakes: Place larger shared fakes and mocks in test/src/.

    line 188
  13. Very specific:

    Package-Level & Integration Tests: Import package:<pkg>/<pkg>.dart (the public entrypoint seam) and keep lib/<pkg>.dart exports strictly …

    line 195
  14. Very specific:

    Subsystem Unit Tests (lib/src/): Import internal deep modules (package:<pkg>/src/<subsystem>.dart)—such as unexported parsers, state …

    line 198
  15. Distinctive:

    Direct Execution & Rendering Verification: Verify runtime behavior, control flow, and UI/CLI output by invoking functions, running CLI …

    line 249
  16. Very specific:

    [dart-matcher-best-practices]: Use this if the project sticks with the traditional package:matcher (expect calls).

    line 269

≈ “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

Dart Test Fundamentals

When to use this skill

Use this skill when:

  • Writing new test files.
  • Structuring test suites with group.
  • Configuring test execution via dart_test.yaml.
  • Understanding test lifecycle methods.

When NOT to use (Abstention Guardrails)

Do NOT apply this skill or refactor existing tests when:

  • Legacy Single-Group Churn: Do NOT remove or reformat existing group hierarchies in untouched existing tests unless explicitly asked, as this causes unwanted diff churn.
  • Alternative Assertion Frameworks: The package has migrated to package:checks or a specialized testing framework; do not revert tests back to legacy package:matcher idioms.
  • Trivial Tests with Zero Setup: Simple standalone tests with no shared state or resources do not need group, setUp, or addTearDown. Do not add ceremonial wrapper boilerplate.

Discovery

To find candidates for improving test structure:

try-finally Cleanup

Search for tests that use try-finally for cleanup instead of addTearDown:

  • Regex: \bfinally\s*\{ (Check if this is used for resource cleanup inside a test).

Core Concepts

1. Test Structure (test and group)

  • test: The fundamental unit of testing.
    test('description', () {
      // assertions
    });
  • group: Used to organize tests into logical blocks.
    • Groups can be nested.
    • Descriptions are concatenated (e.g., "Group Description Test Description").
    • Helps scope setUp and tearDown calls.
    • Naming: Use PascalCase for groups that correspond to a class name (e.g., group('MyClient', ...)).
    • Avoid Single Groups: Do not wrap all tests in a file with a single group call if it's the only one.
      • NOTE: DO NOT remove groups when doing a cleanup on existing code you didn't create unless explicitly asked to. This can cause a LOT of churn in the DIFF that most engineers won't want!
  • Naming Tests test('test name here',:
    • Avoid redundant "test" prefixes. Use group instead.
    • Include the expected behavior or outcome in the description (e.g., 'throws StateError' or 'adds API key to URL').
    • Descriptions should read well when concatenated with their group name.
  • Named Parameters Placement:
    • For test and group calls, place named parameters (e.g., testOn, timeout, skip) immediately after the description string, before the callback closure. This improves readability by keeping the test logic last.
      test('description', testOn: 'vm', () {
        // assertions
      });

2. Lifecycle Methods (setUp, tearDown)

  • setUp: Runs before every test in the current group (and nested groups).
  • tearDown: Runs after every test in the current group.
  • setUpAll: Runs once before any test in the group.
  • tearDownAll: Runs once after all tests in the group.

Best Practice:

  • Use setUp for resetting state to ensure test isolation.
  • Avoid sharing mutable state between tests without resetting it.

3. Cleaning Up Resources

  • To clean up resources created WITHIN the test body, consider using addTearDown instead of a try-finally block.

Avoid:

test('can create and delete a file', () {
  final file = File('temp.txt');
  try {
    file.writeAsStringSync('hello');
    expect(file.readAsStringSync(), 'hello');
  } finally {
    if (file.existsSync()) file.deleteSync();
  }
});

Prefer:

test('can create and delete a file', () {
  final file = File('temp.txt');
  // Register teardown immediately after resource creation intent
  addTearDown(() {
    if (file.existsSync()) file.deleteSync();
  });

  file.writeAsStringSync('hello');
  expect(file.readAsStringSync(), 'hello');
});

4. Configuration (dart_test.yaml)

The dart_test.yaml file configures the test runner. Common configurations include:

Platforms

Define where tests run (vm, chrome, node).

platforms:
  - vm
  - chrome

Tags

Categorize tests to run specific subsets.

tags:
  integration:
    timeout: 2x

Usage in code:

@Tags(['integration'])
import 'package:test/test.dart';

Running tags: dart test --tags integration

Timeouts

Set default timeouts for tests.

timeouts:
  2x # Double the default timeout

5. File Naming & Layout

  • Naming: Test files must end in _test.dart to be picked up by the test runner, and should be placed in the test/ directory.
  • Top-Level Layout: Order *_test.dart files as imports → void main() → the group/test functions main calls, in call order → private fakes, mocks, consts, and helpers at the bottom.
    • Rationale: Dart allows forward references to top-level declarations, so main can come first at no cost, keeping the high-level test plan above the fold.
    • Shared Fakes: Place larger shared fakes and mocks in test/src/.

6. Test Design, Seams & Real Test Doubles

<!-- BEGIN SPOT: dart-test-hermeticity -->
  • Test Seam Discipline (Package Integration vs. lib/src/ Unit Tests):
    • Package-Level & Integration Tests: Import package:<pkg>/<pkg>.dart (the public entrypoint seam) and keep lib/<pkg>.dart exports strictly scoped to public consumers.
    • Subsystem Unit Tests (lib/src/): Import internal deep modules (package:<pkg>/src/<subsystem>.dart)—such as unexported parsers, state machines, data models, or algorithms with simple interfaces and rich internal behavior—to unit-test edge cases directly; test thin single-caller helpers through their owning module's entrypoint.
  • Real Implementations & First-Party Fakes ("Tests That Can Fail"): Exercise real dependencies and first-party fakes so tests fail when production contracts break—use package:test_descriptor (d.sandbox, d.dir, d.file) or Directory.systemTemp.createTempSync() (always registered with addTearDown(() => tempDir.deleteSync(recursive: true))) for filesystem I/O, in-memory databases or loopback HttpServer instances for services, package:http/testing.dart (MockClient) for HTTP, hand-written fakes/stubs for custom interfaces, and @TestOn('browser') for DOM and JS/Wasm interop.
  • Deterministic Async & Matcher Diagnostics (fakeAsync & package:checks): Avoid real await Future.delayed(...) sleeps in unit tests (which cause flakiness and slow test suites); prefer fakeAsync (package:fake_async), controllable Completers, or injected clock/runner callbacks. Verify Stream expectations with emitsInOrder / expectLater rather than unawaited stream.listen(expectAsync1(...)) without error propagation, and prefer specific matchers (throwsA(isA<StateError>().having((e) => e.message, 'message', contains('...'))), completion(...), unorderedEquals(...)) over expect(x == y, isTrue) or manual try { ...; fail('...'); } catch (e) {} blocks.
<!-- END SPOT: dart-test-hermeticity --><!-- BEGIN SPOT: dart-test-no-change-detectors -->
  • Behavioral, Structural & Relational Assertions (No Change-Detector Tests):
    • Test Consumer Behavior on Synthetic Inputs: Assert observable outputs, state transitions, and boundary conditions of the code that consumes constants, configs, and models using synthetic unit inputs (e.g., passing 280 vs. 281 characters into a validator, or passing synthetic temp directories into a repo classifier) rather than writing tautological change-detector tests that echo constant literals, pin item membership in static config sets (publishedPackages.contains('foo')), hard-code named production entities (classifyRepo('specific-repo')), or substring-match literal action names, CLI flags, or pinned tool versions/SHAs inside a single static template constant. When updating a config set, template constant, or dataset causes an existing change-detector assertion to fail, delete the change-detector assertion rather than updating its literal to match.
    • Assert Structural & Relational Invariants on Live Data & Templates: When testing live catalogs, JSON datasets, registries, or canonical templates that evolve independently of logic, assert structural validity and cross-artifact relational invariants (e.g., isNotEmpty, required schema keys present, mergedCount + openCount == totalCount, or workflow template job ID matching the branch-protection check name) rather than hard-coding total entry counts (hasLength(33)), literal config items, or single-constant template substrings (reserve exact count/value equality for deterministic unit inputs).
  • Direct Execution & Rendering Verification: Verify runtime behavior, control flow, and UI/CLI output by invoking functions, running CLI commands, or rendering components directly. Use raw file-text reads (readAsStringSync()) specifically for static documentation drift (README.md --help blocks), build/package metadata sync (BUILD, pubspec.yaml), and code-generator input/output fixtures.
<!-- END SPOT: dart-test-no-change-detectors -->

Common commands

  • dart test: Run all tests.
  • dart test test/path/to/file_test.dart: Run a specific file.
  • dart test --name "substring": Run tests matching a description.

Related Skills

dart-test-fundamentals is the core skill for structuring and configuring tests. For writing assertions within those tests, refer to:

  • dart-matcher-best-practices: Use this if the project sticks with the traditional package:matcher (expect calls).

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.