You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Security gap: no AWS GuardDuty (cloud-level behavioral threat detection). From the Security Model gaps register — part of the security posture gap register epic #1152.
The gap
The platform has strong cloud-level isolation (separate accounts, SCPs, least-privilege IAM, KMS) but no cloud-native threat detection. GuardDuty is the AWS-native behavioral detector — it analyzes CloudTrail management events, VPC flow logs, and DNS logs (plus optional EKS Audit Log, S3, Malware, and RDS protection) for anomalous/malicious patterns (credential exfil, crypto-mining, recon, impossible-travel API calls, known-bad IPs).
This is one leg of the platform's biggest structural weakness: prevention-strong, detection-weak. The others being addressed in parallel — Falco (#149, cluster runtime detection), EKS audit logging (#816, forensics), SIEM (#1150, correlation).
Why deferred (cost)
GuardDuty pricing is usage-based and scales with event/log volume across all accounts; with EKS Runtime Monitoring and the org-wide multi-account footprint it can reach hundreds of $/month with no fixed cap. That conflicts with the platform FinOps posture (ADR-092: free OSS/native only, no spend budget; paid tiers documented-but-deferred).
Partial mitigation that reduces the urgency: Falco (#149, free/OSS) covers a meaningful slice of the runtime detection need on-cluster, and CloudTrail already provides the audit substrate. GuardDuty remains the right long-term cloud-detection layer but is deferred until there's a security budget (or a compliance driver that justifies the spend).
Scope when picked up
Org-level GuardDuty via the management/delegated-admin account, auto-enrolling all member accounts.
Decide protection plans by cost/value: management-event + flow-log + DNS baseline first; EKS Runtime Monitoring, Malware, S3, RDS protection as explicit cost-gated add-ons.
Security gap: no AWS GuardDuty (cloud-level behavioral threat detection). From the Security Model gaps register — part of the security posture gap register epic #1152.
The gap
The platform has strong cloud-level isolation (separate accounts, SCPs, least-privilege IAM, KMS) but no cloud-native threat detection. GuardDuty is the AWS-native behavioral detector — it analyzes CloudTrail management events, VPC flow logs, and DNS logs (plus optional EKS Audit Log, S3, Malware, and RDS protection) for anomalous/malicious patterns (credential exfil, crypto-mining, recon, impossible-travel API calls, known-bad IPs).
This is one leg of the platform's biggest structural weakness: prevention-strong, detection-weak. The others being addressed in parallel — Falco (#149, cluster runtime detection), EKS audit logging (#816, forensics), SIEM (#1150, correlation).
Why deferred (cost)
GuardDuty pricing is usage-based and scales with event/log volume across all accounts; with EKS Runtime Monitoring and the org-wide multi-account footprint it can reach hundreds of $/month with no fixed cap. That conflicts with the platform FinOps posture (ADR-092: free OSS/native only, no spend budget; paid tiers documented-but-deferred).
Partial mitigation that reduces the urgency: Falco (#149, free/OSS) covers a meaningful slice of the runtime detection need on-cluster, and CloudTrail already provides the audit substrate. GuardDuty remains the right long-term cloud-detection layer but is deferred until there's a security budget (or a compliance driver that justifies the spend).
Scope when picked up
Blocked on: a security spend budget. Tracking: #1152.