What problem does it solve? Designing test cases from specifications alone is error-prone: testers miss invalid input classes, off-by-one boundary defects, and untested condition combinations. This Skill provides systematic black-box techniques that mechanically derive test cases from input-output specifications without reading implementation code, making it usable even before implementation exists (e.g., for TDD test lists). ## Core Features & Use Cases - Equivalence Partitioning: Divide input domains into valid and invalid equivalence classes and pick one representative per class to compress case counts while preserving coverage. - Boundary Value Analysis: Target values just before, on, and after each boundary to catch < vs <= mistakes and off-by-one errors, with guidance on floating-point boundaries and sharing boundary constants with production code. - Domain Analysis: Extend boundary analysis to multiple variables using on/off/in/out points with the one-point-one-test principle for multi-variable region judgments. - Decision Tables: Enumerate condition combinations as rules and fold impossible combinations with don't-care entries to test business logic like discount or credit rules. - Use Case: Given a function that validates a score in range 0-100, apply boundary value analysis to produce cases at -1, 0, 1, 99, 100, 101, then layer equivalence classes for invalid types, yielding a complete specification-based test suite. ## Quick Start Ask the AI to apply equivalence partitioning and boundary value analysis to derive test cases for your function's input specification.
