How to use the Python Testing Patterns agent skill

What does the skill make the agent do?

Once installed, the agent follows a structured guide for writing Python tests with pytest. It knows how to organize test folders into unit, integration and end-to-end directories, name tests clearly, use the Arrange-Act-Assert pattern, mock retries and time-dependent behavior, mark slow or integration tests, and measure coverage.

The skill is a reference of conventions and examples. It does not run tests or code by itself.

• Organizes tests into folders such as test_unit, test_integration and test_e2e with a shared conftest.py.
• Suggests test names in the form test_<unit>_<scenario>_<expected_outcome>.
• Shows how to mock retries with unittest.mock and control time with freezegun.
• Explains pytest markers such as slow, integration, skip, skipif and xfail.
• Shows coverage commands using pytest-cov.

How do you install the Python Testing Patterns skill?

Download the ZIP file from this page and unzip it. For a personal setup, place the unzipped folder in ~/.claude/skills/. For a single project, place it in that project's .claude/skills/ folder.

In the Claude apps you can instead upload the skill under Settings → Capabilities → Skills.

• Personal install: ~/.claude/skills/
• Project install: .claude/skills/ inside your project
• Claude apps: Settings → Capabilities → Skills, then upload the skill

What do you need before using it?

The skill supplies guidance only. You need Python and pytest installed on your machine. Install pytest-cov if you want coverage reports, and freezegun only if you plan to use the time-mocking examples. The software used in examples is installed by you; no API keys or accounts are needed.

The skill does not run scripts and does not use the network.

• Python
• pytest
• pytest-cov (for coverage reporting)
• freezegun (only for time-mocking examples)

How do you trigger the skill with a natural request?

Ask for Python tests in plain language and mention pytest, the kind of test, or the goal. The skill's description covers writing Python tests, setting up test suites and testing best practices, so requests like these should match it.

Be specific about your code and your goal. The agent will then use the patterns in the guide to shape its answer.

• Write pytest unit tests for my parse_price function, covering invalid input.
• Set up a pytest test folder structure for a FastAPI project with unit and integration tests.
• Add a test that checks my token expires after one hour, using freezegun.
• Help me measure coverage for my myapp package and fail CI below 80 percent.

What does an example look like?

The guide includes a retry example that uses a Mock whose side_effect fails twice and then succeeds. The test then checks the returned value and that the call count is 3. A second example shows that a ValueError is not retried, so the call count is 1.

Coverage commands from the guide look like this:

What safety points should you know?

The skill runs no scripts and uses no network access, so it does not execute code on your behalf. Any tests the agent writes are code you should review and run yourself in your own environment. Check the examples before using them against real services, and do not paste secrets into prompts.

• No scripts are run by the skill.
• No network access is used by the skill.
• Review generated tests before committing them.

What tips help you get better results?

Give the agent your existing test layout and pytest configuration so it matches your conventions. Tell it when your team uses a different naming style, since the guide itself says to adapt to your team's preferences. Ask for coverage targets that mean something, because the guide recommends aiming for meaningful coverage rather than just high percentages.

• Share your existing tests/ folder layout first.
• Name the libraries you already use, such as freezegun or pytest-cov.
• Ask for both passing and failing-case tests for each function.