| name | mocking-strategies |
| description | Mock, patch, fake, and stub patterns for isolating dependencies in Python and JavaScript test suites.
|
Mocking strategies for isolating external dependencies in tests.
When to activate
Use this skill when tests need to isolate external dependencies such as
databases, APIs, filesystems, time, or randomness.
Python mocking
unittest.mock.patch("module.path.ClassName") -- replace an object for
the duration of a test. Use as a decorator or context manager.patch.object(obj, "attr") -- patch a specific attribute on a live
object.MagicMock(spec=RealClass) -- create a mock that mirrors the real class
API; accessing non-existent attributes raises AttributeError.side_effect -- set to a function for dynamic return values, or to an
exception class to simulate errors.return_value -- set the fixed return value of the mock.monkeypatch (pytest built-in) -- set attributes, environment variables,
or dict items for the duration of a test without import-path gymnastics.
JavaScript mocking
jest.mock("module") / vi.mock("module") -- auto-mock an entire
module. Provide a factory function for custom behavior.- Manual mocks in
__mocks__/ -- place a file next to node_modules or
the real module to override it project-wide.
- Dependency injection -- pass dependencies as constructor or function
arguments to make them trivially replaceable in tests.
msw (Mock Service Worker) -- intercept HTTP at the network level for
realistic API mocking without changing application code.
When to mock
- External HTTP services and APIs.
- Time and dates (
datetime.now, Date.now).
- Filesystem reads and writes.
- Network sockets and connections.
- Databases and data stores.
- Random number generators and entropy sources.
When NOT to mock
- Pure functions with no side effects -- just call them directly.
- Data transformations and mapping logic.
- Simple utility functions.
- The system under test itself -- that makes the test tautological.
Mock verification
- Python:
mock.assert_called_once_with(arg1, arg2),
mock.call_count, mock.call_args_list.
- JavaScript:
expect(fn).toHaveBeenCalledTimes(1),
expect(fn).toHaveBeenCalledWith(arg1, arg2).
MUST
- Mock at the module boundary -- patch where the dependency is imported,
not where it is defined.
- Use
spec=True (Python) so the mock rejects calls to methods that do not
exist on the real object.
- Reset or restore mocks between tests to prevent leakage.
MUST NOT
- Do not mock the system under test -- only mock its collaborators.
- Do not over-mock to the point where the test only verifies mock wiring
and proves nothing about real behavior.
- Do not mock standard library builtins without a clear reason (e.g.
mocking
open is fine for filesystem isolation, but mocking len is
not).
