testing

name	testing
description	Review/generate unit, integration, and regression test expectations

Anomalib Testing Review/Generation

Use this skill when reviewing/generating additions or changes that should be covered by tests.

Purpose and scope

Use this skill to decide what test coverage is required and whether existing tests still prove the intended behavior.

Request changes when

• new behavior ships without tests;
• behavior changes update code but not existing tests;
• a bug fix has no regression test when one is feasible;
• tests are flaky, network-dependent, or placed in the wrong area of the suite.

Test placement and scope

• Tests should live under tests/ and generally mirror the code area they validate.
• Prefer the established split between tests/unit/ and tests/integration/.
• Ask for unit tests for new behavior and for integration coverage when cross-component behavior changes.

Test style

• Follow pytest conventions already used in the repo.
• Prefer fixtures and parametrization where they improve clarity and coverage.
• Keep tests offline and deterministic where practical.
• For tensor-heavy logic, ensure tests assert the properties that matter: shapes, values, reconstruction behavior, errors, or invariants.

Regression mindset

• Bug fixes should include a regression test when feasible.
• Behavior changes should update existing tests, not only add new ones.
• If the change touches CLI/config/model loading or pipeline orchestration, review whether a higher-level test is also needed.

Repo-grounded review anchors

• pyproject.toml for pytest markers and test path configuration

Review prompts

• What test proves the new behavior works?
• What existing behavior could regress because of this change?
• Is the test placed in the right part of the suite?
• Does the change need both unit and integration coverage?

Reviewer checklist

• Check that new behavior has test coverage.
• Check that changed behavior updates old tests too.
• Check placement under tests/unit/ or tests/integration/.
• Check determinism and offline execution.