Testing Skill

This skill provides guidance for writing unit tests and checking coverage for Canvas plugins.

When to Use This Skill

Use this skill when:

• •Writing unit tests for plugin handlers
• •Checking test coverage
• •Mocking Canvas SDK components
• •Setting up test fixtures

Opinionated Testing Rules

These rules are non-negotiable. Follow them exactly.

1. Tests at Container Level, Mirroring Source Structure

CRITICAL: The tests/ directory MUST be at the container level (parallel to the inner plugin folder), NOT inside the inner folder.

my-plugin-name/                       # Container folder (kebab-case)
├── pyproject.toml                    # Container level
├── tests/                            # Container level - NOT inside inner folder!
│   ├── conftest.py
│   ├── protocols/
│   │   ├── test_vitals_handler.py    # mirrors inner/protocols/vitals_handler.py
│   │   └── test_lab_handler.py       # mirrors inner/protocols/lab_handler.py
│   ├── api/
│   │   └── test_routes.py            # mirrors inner/api/routes.py
│   └── helpers/
│       └── test_utils.py             # mirrors inner/helpers/utils.py
└── my_plugin_name/                   # Inner folder (snake_case)
    ├── CANVAS_MANIFEST.json
    ├── README.md
    ├── protocols/
    │   ├── vitals_handler.py
    │   └── lab_handler.py
    ├── api/
    │   └── routes.py
    └── helpers/
        └── utils.py

Key points:

• •tests/ is a sibling to the inner plugin folder, not a child
• •Test file structure mirrors the inner folder's source structure
• •Test file name = test_ + source file name

WRONG structure (DO NOT DO THIS):

my-plugin-name/
└── my_plugin_name/
    ├── protocols/
    │   └── handler.py
    └── tests/                        # WRONG - tests inside inner folder!
        └── test_handler.py

If you find tests inside the inner folder, move them:

mv my_plugin_name/tests ./tests

2. One Test File Per Source File

• •Each source file gets exactly ONE test file
• •Test file name = test_ + source file name
• •Example: protocols/handler.py → tests/protocols/test_handler.py

3. Use Mocks for Isolation

• •Every external dependency MUST be mocked
• •Tests must never make real API calls or database queries
• •Use unittest.mock.MagicMock and patch for all mocking

4. Always Verify Mock Calls

Every mock MUST have its calls verified. Do not just mock and forget.

# REQUIRED - verify mock was called correctly, with the right arguments, for all calls
with patch("canvas_sdk.v1.data.patient.Patient.objects") as mock_objects:
    mock_objects.get.return_value = mock_patient

    handler.compute()

    calls = [call.get(id="patient-123"), call.get(id="patient-456")]
    assert mock_objects.mock_calls == calls  # REQUIRED

5. Exclude Tests from Coverage

The pyproject.toml MUST exclude test files from coverage measurement:

[tool.coverage.run]
source = ["plugin_name"]
omit = ["tests/*", "*/tests/*"]

[tool.coverage.report]
omit = ["tests/*", "*/tests/*"]

Quick Commands

# Run tests
uv run pytest

# Run tests with coverage
uv run pytest --cov=plugin_name --cov-report=term-missing --cov-branch

# Run tests with HTML coverage report
uv run pytest --cov=plugin_name --cov-report=html

Coverage Target

Aim for 90% coverage on plugin code. If coverage is below 90%, offer to write additional tests.

Test Data Approach

Use manual mocking with MagicMock, NOT factory libraries. Factories don't exist for Canvas plugins. Create test data directly in fixtures and customize per-test. See testing_context.txt for patterns.

Documentation

Reference the testing_context.txt file in this skill directory for:

• •Mocking patterns for Canvas SDK
• •Test structure guidelines
• •Coverage checking procedures
• •Common test scenarios

After Tests Are Complete

Do NOT stop after writing tests. Once tests pass with adequate coverage:

1. •Tell the user tests are complete and coverage target is met
2. •Immediately offer to deploy for UAT:

"Tests are passing with X% coverage. The next step is to deploy the plugin to a Canvas instance for user acceptance testing. Ready to deploy?"

1. •Use the /cpa:deploy command to guide deployment and log monitoring

Always drive toward the next step. The workflow is: Implement → Test → Deploy → UAT