These patterns explore how to use AGISystem2 as a “verifiable layer” around AI behavior: representing constraints as theories, validating actions or content with proofs or deterministic traces, and producing remediation-oriented reports.