Conciseness

The body is lean and assumes Claude's competence — terse principle bullets ("Use fixtures for setup/teardown, never class-level mutable state"), lookup tables, copy-paste commands, and a routing table, with zero space spent explaining what pytest is or how it works. Every section earns its tokens; there is no padded explanation to trim, so it sits at anchor 5 rather than the 'minor over-explanation' of anchor 4.

Actionability

Guidance is fully executable: exact commands ("pytest -x", "pytest --lf", "pytest -k 'pattern'", "pytest -n auto"), a copy-paste "pip install pytest pytest-asyncio pytest-mock pytest-cov pytest-xdist" line, a concrete directory tree, exact naming patterns, and a pattern table mapping every common case (`@pytest.fixture`, `@pytest.mark.parametrize`, `pytest.raises`, `mocker.patch`, `pytest.approx`) with its use case. Not a 4: the specific examples cover the common cases, with deeper executable detail correctly delegated to the routed references.

Workflow Clarity

This is a single-purpose conventions skill with no multi-step process, so the simple-skill guidance applies: the single action (apply these conventions, then route to the matching reference by task) is unambiguous via the task-to-reference routing table. No destructive or batch operations exist that would require validation checkpoints or cap the score, and the routing table plus 'reach for these before writing your own' built-in fixtures note give a clear decision structure.

Progressive Disclosure

The body is an overview plus quick reference that appropriately splits detail into four one-level-deep reference files (fixtures.md, parametrization.md, mocking.md, patterns.md), each well-signaled in a routing table keyed by task. All four referenced paths exist in the bundle and the reference files contain no further nested references, matching the anchor-5 structure (clear overview, one-level-deep references, easy navigation) rather than anchor 4's organization gaps.

Specificity

The description lists multiple concrete capabilities — "fixtures and scopes, parametrization, mocking with pytest-mock, built-in fixtures (`tmp_path`, `monkeypatch`, `capsys`, `caplog`), markers, exception testing, and async tests" — naming specific fixtures, plugins, and config files, which comprehensively covers the pytest domain with no vague filler. It is not a 4 because coverage has no minor gaps: configuration (`pytest.ini` / `pyproject.toml`) is included alongside every core pytest feature area.

Completeness

Both questions are answered explicitly: the "what" is the enumerated pytest conventions and built-in fixtures, and the "when" is a concrete trigger clause — "Use when writing, reviewing, or refactoring Python tests, `conftest.py` fixtures, or pytest configuration in `pytest.ini` / `pyproject.toml`". This matches the anchor-5 pattern (explicit what AND when with concrete trigger phrases), not anchor 4 where the 'when' could be more specific.

Trigger Term Quality

Natural trigger terms include "Python test suites", "writing, reviewing, or refactoring Python tests", "`conftest.py` fixtures", "pytest configuration", and file extensions/names ("conftest.py", "pytest.ini", "pyproject.toml") — comprehensive coverage including synonyms and file names that users would actually say. Not a 4: the missing-term allowance of anchor 4 does not apply, since the phrases users naturally use when needing this skill (pytest, tests, conftest, fixtures, mocking) are all present.

Distinctiveness Conflict Risk

The niche is unambiguous (pytest-specific, with named pytest APIs), and it adds explicit negative boundaries — "Does not cover general Python syntax or typing — see modern-python-development — nor FastAPI `TestClient` and app-level test setup — see fastapi-best-practices" — which actively routes away overlapping use cases. Minimal conflict risk; clearly the anchor-5 case rather than the 'minor overlap risk' of anchor 4.