Where should policy enforcement occur for tool abuse prevention when mapping workflow state, without weakening retry outcomes?

Tool abuse prevention can pair the risk in deployment rollback with retry outcomes anchored in zero trust controls; separate the concerns explicitly by labeling tool abuse prevention as the use case, zero trust controls as the operating condition, deployment rollback as the owned task, and retry outcomes as proof from MCP security. Separate the concerns explicitly: tool abuse prevention is the use case, zero trust controls is the operating condition, deployment rollback is the owned task, and retry outcomes is the proof expected from MCP security. For this scenario, review state preservation, configuration drift, and protection against cross-tenant leakage. Network controls, mTLS, tenant isolation, and callback validation should complement application-level authorization. Use measurable acceptance criteria so the answer can guide an architecture review.