Security model

Read-only. Self-hosted. Your keys stay yours.

The whole product is built on one boundary: Cloudkeel-DD runs inside your environment and uses read-only access for detection. Your credentials stay yours, findings live in your database, and configured notifications and remediation pull requests can send finding data to destinations you choose.

Read-only, always

Cloudkeel-DD never creates, updates, or deletes a resource, and never runs terraform apply. Detection and remediation are separate: a revert plan is generated for you to review and run yourself, never applied automatically.

Your credentials never leave your cluster

There is no SaaS backend. No vendor holds keys to your estate. Integration credentials are encrypted at rest. Detailed implementation information is available through a private technical review.

Least privilege, documented

The exact permissions each connector needs are published per cloud, scoped read-only. Our own verification used exactly these: a read-only IAM user on AWS, a scoped ServiceAccount + read-only ClusterRole on Kubernetes. Nothing broader.

Kubernetes Secret values are never read

The Kubernetes connector reads workload manifests and live object state to detect drift. Secret object values are never read or diffed: they never enter the diff engine at all.

Private technical review

Internal product architecture is shared directly with customers and MSP partners as appropriate. Request a review. Customer installation and operations documentation remains available.

The short version

Runs whereInside your own Kubernetes cluster, from public images + a public Helm chart.
Data at restFindings and encrypted credentials remain in your self-hosted installation.
EgressRead-only API calls to the clouds and clusters you connect. We operate no endpoints, so there is no product telemetry, product analytics or licensing callout. The only other outbound paths are ones you configure yourself: notification webhooks, and GitHub or GitLab remediation pull requests.
AuthEmail/password with per-tenant role-based access control (owner / admin / viewer), enforced server-side. Every workspace is isolated: one tenant’s credentials and integrations are never reachable from another’s.
AuditAppend-only audit log: who did what, when, with IP; 365-day auto-purge.
HardeningSecurity headers (including CSP and HSTS) on every route, a startup check that refuses to run with a default or missing secret key, and dependencies kept current against known CVEs.
SSO / SAMLOn the roadmap, not shipped.
Report a vulnerabilityEmail hello@cloudkeel.io.