What problem does it solve? Autonomous coding agents build software with no human backstop, so correctness must be proven through evidence rather than trust. This Skill designs complete assurance strategies that let agents self-verify their own work, choosing the right testing techniques at the right rigor level for each subsystem. ## Core Features & Use Cases - Four-axis diagnosis: Evaluates criticality, volatility, longevity, and accountability per subsystem to shape the verification strategy. - Graded technique prescriptions: Selects from unit, property-based, model-based, differential, metamorphic, and React unit testing, choosing the lightest rung that produces useful evidence and stating upgrade triggers. - Technique reference cards: Loads detailed guidance on oracles, prerequisites, harness changes, acceptance criteria, and failure modes before prescribing any technique. - Use Case: When starting a new agent-built service, ask for a testing strategy and receive a plan specifying which techniques apply to each module, what evidence each produces, and how the harness wires into make verify and CI. ## Quick Start Ask the agent to design an assurance strategy for your project, describing its subsystems and how critical each one is.
