What problem does it solve? It prevents untested or over-engineered code from shipping by enforcing a strict red-green-refactor loop: one failing test per behavior, minimal implementation, and disciplined refactoring, with interface approval gates before any code is written. ## Core Features & Use Cases - Iron Law TDD Cycle: Write exactly one failing test per behavior, watch it fail, implement the minimal code to pass, then refactor with tagged cleanup categories (delete, stdlib, yagni, shrink). - Scope Classification: Classifies tasks as Spike, Bounded, or Architectural before starting, routing architectural changes to a design roundtable instead of the TDD loop. - Stack-Specific Conventions: Loads reference guides for Java/Spring Boot (JUnit 5, Gradle), Angular (Jasmine/Karma/Jest), React/Next.js (Vitest, RTL), and Python (pytest). - Testing Anti-Pattern Guards: Blocks mock-heavy tests, test-only production methods, and non-falsifiable assertions. - Use Case: When adding a new endpoint to a Spring Boot service, the skill requires confirming the interface with the user, listing behaviors including edge cases, then cycling one test at a time using ./gradlew test --tests commands. ## Quick Start Ask the AI to implement a new feature or bug fix using test-driven development with the red-green-refactor cycle for your project's stack.
