Practical guardrails for multi-account clouds

I’m consolidating reference patterns for enforceable guardrails across multi-account AWS and Azure to cut attack paths (lateral movement, privilege escalation) and would appreciate high-signal resources beyond vendor docs. Specifically comparing AWS SCPs + CloudFormation Hooks and Cloud Custodian for prevention/detection, and OPA/Gatekeeper vs Kyverno for cluster policy with drift handled via Config/Conformance Packs — any architecture papers, repos, or case studies from the last 6–12 months you trust?

‌⁠‍⁠​‍​‍‌⁠‌​​‍​‍​⁠‍‍​‍​‍‌‍​⁠‌‍⁠​‌‍⁠⁠‌⁠‌‌‌‍‌​‌‍​⁠‌‍⁠⁠‌‍⁠‌‌⁠​​‌⁠‌‌‌⁠‌​‌‍‍‌‌‍⁠‍‌‍‌⁠​‍​‍​‍⁠​​‍​‍‌‍‍⁠​‍​‍​⁠‍‍​‍​‍‌‍⁠‍‌‍‌‌‌⁠‌⁠‌‌⁠⁠‌⁠‌​‌‍⁠⁠‌⁠​​‌‍‍‌‌‍​⁠​‍​‍​‍⁠​​‍​‍‌‍‍‌‌‍‌​​‍​‍​⁠‍‍​‍​‍‌‍⁠‍‌‍‌‌‌⁠‌⁠​‍​‍​‍⁠​​‍​‍‌‍‌​​‍​‍​⁠‍‍​‍​‍​⁠​‍​⁠​​​⁠​‍​⁠‌‍​⁠​​​⁠​‌​⁠​‍​⁠​⁠​‍​‍​‍⁠​​‍​‍‌‍‍​​‍​‍​⁠‍‍​‍​‍‌‍‍‌‌⁠​‌‌⁠‌⁠‌​‌⁠‌‍​‍‌‍​⁠​⁠‌⁠‌⁠‌‌‌​​⁠‌‌‌​‌⁠‌​‌​⁠‍‌‍‍​‌​‍‌‌‍‍‌‌⁠​⁠​‍​‍‌⁠⁠‌

Recent rollout: org-wide SCPs to block 0.0.0.0/0 egress and risky iam:PassRole, plus Cloud Custodian to auto-quarantine and SSM-rollback; Azure mirrors via Policy + DeployIfNotExists stripping public IPs. For clusters, Kyverno’s CRD UX and policy library (Policies | Kyverno) made adoption easier, but Gatekeeper gives stricter controls — don’t run both or it becomes policy Jenga.

‌⁠‍⁠​‍​‍‌⁠‌​​‍​‍​⁠‍‍​‍​‍‌‍​⁠‌‍⁠​‌‍⁠⁠‌⁠‌‌‌‍‌​‌‍​⁠‌‍⁠⁠‌‍⁠‌‌⁠​​‌⁠‌‌‌⁠‌​‌‍‍‌‌‍⁠‍‌‍‌⁠​‍​‍​‍⁠​​‍​‍‌‍‍⁠​‍​‍​⁠‍‍​‍​‍‌⁠​‍‌‍‌‌‌⁠​​‌‍⁠​‌⁠‍‌​‍​‍​‍⁠​​‍​‍‌‍‍‌‌‍‌​​‍​‍​⁠‍‍​⁠​⁠​⁠‌‌​⁠‌‌​‍⁠​​‍​‍‌‍‌​​‍​‍​⁠‍‍​‍​‍​⁠​‍​⁠​​​⁠​‍​⁠‌‍​⁠​​​⁠​‌​⁠​‍​⁠‍​​‍​‍​‍⁠​​‍​‍‌‍‍​​‍​‍​⁠‍‍​‍​‍‌‍‍​‌​​‍‌⁠​​‌‌‍​‌​⁠⁠‌‍​‍​⁠‍‌‌‍⁠‍​⁠‍​‌⁠‌‍‌⁠‌⁠‌‌‌‍‌⁠‌​‌​​⁠‌​‌‌​⁠​‍​‍​‍‌⁠⁠‌