What problem does it solve? Ensuring that skills actually work as claimed is hard: tests often use mock data, only check happy paths, or lack reproducible evidence. This Skill enforces rigorous quality validation by routing checks to two sub-skills — falsification-validation and golden-test-methodology — so every skill claim is tested against real inputs and golden test coverage is measured consistently. ## Core Features & Use Cases - Falsification Validation: Tests skill claims using real task inputs (never mock data), requires counter-evidence tests, archives full evidence chains to test-results/, and maintains a Bayesian trust score per skill via the update_trust.py script. - Golden Test Methodology: Computes golden test coverage (skills with complete GOLDEN_SET.md + cases/ + expected/), grades coverage (Excellent/Adequate/Low/Critical), and defines prioritized expansion order for missing tests. - Input Validation & Routing: Rejects empty or invalid skill_path inputs with contextual error messages and recovery suggestions, then routes valid requests to the appropriate sub-skill. - Use Case: When the evolution engine needs to verify a newly absorbed skill, invoke this Skill with the skill's path to run falsification tests, update its trust grade, and confirm its golden test suite is complete. ## Quick Start Run a quality check on the skill at skills/core/knowledge-extraction and report its trust grade and golden test coverage.
