What problem does it solve? Writing tests after the implementation only proves the code does what the code does. This Skill enforces a disciplined red-green-refactor workflow for Rust so every test fails first for a predicted reason, and it provides concrete patterns for fakes, fault injection, async tests, and golden contracts that teams usually improvise badly. ## Core Features & Use Cases - Red-Green-Refactor Discipline: Step-by-step cycle with a triage table for reading RED results, plus verification gates using cargo fmt, clippy, nextest, and doctests. - Hand-Written Fakes with Fault Injection: Conventions for Fake* test doubles and a typed FaultQueue with OneShot/Persistent scopes to test error paths, retries, and cancellation without cfg(test) branches in production code. - Async and Golden Test Patterns: Rules for #[tokio::test] with paused time, gated fakes for in-flight assertions, JoinHandle awaiting, and blessing discipline for golden contract fixtures. - Use Case: When fixing a reported bug, write a failing reproduction test through the public API first, fix the root cause at the shared choke point, and keep the test as a permanent regression guard in the same commit. ## Quick Start Ask the agent to write a failing Rust test for the behavior you want, confirm it fails for the predicted reason, then implement the minimum code to make it pass.