Lean and efficient — it assumes Claude's competence, never explains what PBT or pytest is, and each step's prose earns its place beside the code. It does not pad with library background or basic concepts.

Every step ships executable, copy-paste-ready code: strategies catalog, composite draw(), .filter()/assume(), @settings, round-trip, metamorphic, pytest fixture, and CI seed — none are pseudocode.

A clearly sequenced Step 1–9 progression with a real feedback loop in Step 9 (copy the falsifying example into @example to lock in the regression) and a reproducibility checkpoint in Step 6/9 (derandomize/--hypothesis-seed).

The body is an overview that defers the full strategies catalog, @settings table, anti-patterns, and limitations to the one-level-deep, clearly signaled references/hypothesis-reference.md, which exists and matches the body's references.

Lists multiple concrete, API-level actions: "wires `@given` with `strategies`", "uses `assume()` / `.filter()` for preconditions", "configures via `@settings(max_examples=..., deadline=...)`", and "exploits Hypothesis's automatic shrinking".

Explicitly answers both what (authors PBT, wires @given with strategies, configures @settings) and when ("Use when a Python project needs PBT to catch edge cases the example-based tests miss").

Covers the natural terms a PBT user would actually say — "property-based tests", "Hypothesis", "PBT", "edge cases", "boundary values", "pytest fixtures" — rather than generic jargon.

Occupies a clear niche (Hypothesis PBT in Python) with distinct triggers and is unlikely to overlap with sibling skills (fast-check, proptest, jqwik, quickcheck) named in the body.