Skip to main content
Global
AIMenta
Blog

APAC Kubernetes DevSecOps Guide 2026: Kyverno, Cosign, and Kubescape for Supply Chain Security and Policy Enforcement

A practitioner guide for APAC DevSecOps and platform engineering teams implementing Kubernetes supply chain security in 2026 — covering Kubescape posture scanning, Cosign container image signing, and Kyverno admission policy enforcement for regulated industries in Singapore, Hong Kong, Japan, and South Korea.

AE By AIMenta Editorial Team ·

The Three-Layer Kubernetes Security Model for APAC Enterprise Teams

Kubernetes security in APAC enterprise environments requires a layered approach that no single tool covers alone. The three-layer model that APAC platform engineering and DevSecOps teams are implementing in 2026:

Layer 1 — Configuration Posture: Is your APAC Kubernetes cluster and application configuration aligned with security hardening benchmarks? Are there containers running as root, missing resource limits, or RBAC paths that enable privilege escalation? Kubescape operates at this layer — auditing APAC cluster and manifest configuration against NSA/CISA, MITRE ATT&CK, and CIS frameworks.

Layer 2 — Supply Chain Integrity: Can you prove that the container images running in your APAC production Kubernetes cluster were built by your APAC CI/CD system from approved source code — and have not been tampered with after build? Cosign operates at this layer — signing images in APAC CI/CD and storing verifiable signatures in OCI registries.

Layer 3 — Runtime Policy Enforcement: Even with good configuration and signed images, can APAC development teams still deploy misconfigured or unsigned images to production? Kyverno operates at this layer — enforcing admission control policies that block unsigned images, flag security misconfigurations, and auto-remediate APAC policy deviations before resources are admitted.

This guide covers how APAC mid-market enterprises across Singapore FSI, Japanese manufacturing, and South Korean technology sectors are implementing this three-layer Kubernetes security model in 2026 — and how Kubescape, Cosign, and Kyverno work together as a coherent APAC DevSecOps toolchain.


Context: APAC Regulatory Pressure on Kubernetes Security

The acceleration of APAC Kubernetes DevSecOps in 2026 is partly driven by regulatory pressure. The joint CISA/ASD/CSA/NISC/KISA AI security guidance published in early 2026 explicitly references Kubernetes-hosted AI systems as targets for adversarial attacks, model poisoning, and supply chain compromise — creating compliance obligations for APAC critical infrastructure operators to implement input validation schemas, model provenance documentation, and AI vendor API supply chain security controls that directly map to Kubernetes supply chain security capabilities.

Beyond AI-specific guidance, MAS TRM (Singapore), HKMA SCR (Hong Kong), FSA FISC (Japan), and FSC (South Korea) operational risk frameworks increasingly reference NIST and CIS security benchmarks as baseline requirements for technology risk management. APAC financial services firms that can demonstrate Kubescape scans against CIS Kubernetes Benchmark and Cosign-signed container images in production have a substantially easier path through APAC technology risk assessments than firms that cannot provide security posture evidence.


Layer 1: Kubescape — APAC Security Posture Auditing

Starting with a cluster scan

The first action for APAC platform engineering teams adopting Kubescape is a baseline cluster scan against NSA/CISA Kubernetes Hardening Guidance — the framework most directly referenced by APAC regulatory guidance:

# Install Kubescape CLI
curl -s https://raw.githubusercontent.com/kubescape/kubescape/master/install.sh | /bin/bash

# Run NSA/CISA framework scan against APAC cluster
kubescape scan framework nsa --kubeconfig ~/.kube/apac-prod.yaml --format json \
  --output apac-cluster-nsa-baseline.json

# Run all frameworks for comprehensive APAC coverage
kubescape scan framework nsa,mitre,cis-v1.23-t1.0.1 \
  --kubeconfig ~/.kube/apac-prod.yaml \
  --format pdf \
  --output apac-security-posture-report.pdf

Typical APAC cluster first-scan findings include:

  • Containers running as root (fails NSA/CISA control C-0013): APAC application containers built without explicit non-root user configuration. Remediation: add runAsNonRoot: true to pod securityContext.

  • Missing resource limits (fails NSA/CISA control C-0004): APAC pods without CPU and memory limits, enabling resource exhaustion attacks. Remediation: add resources.limits to all APAC container specs.

  • Privileged containers (fails MITRE ATT&CK T1611): APAC DaemonSets or system pods running with privileged: true. Remediation: replace with specific Linux capabilities using securityContext.capabilities.add.

  • RBAC over-permission (fails CIS 5.1.x): APAC ServiceAccounts with cluster-admin ClusterRoleBindings or wildcard RBAC permissions. Remediation: replace cluster-admin with scoped role definitions.

CI/CD manifest scanning

The highest-leverage Kubescape deployment for APAC DevSecOps teams is scanning Kubernetes YAML and Helm charts in CI/CD before deployment — catching APAC security issues at the PR level:

# GitHub Actions step: Kubescape scan on APAC manifest changes
- name: Kubescape scan
  uses: kubescape/github-action@main
  with:
    files: "./k8s/**"
    frameworks: "nsa,mitre"
    severityThreshold: high    # fail APAC PR on high-severity findings
    format: "sarif"
    outputFile: "kubescape-results.sarif"

- name: Upload SARIF to GitHub Security
  uses: github/codeql-action/upload-sarif@v3
  with:
    sarif_file: kubescape-results.sarif

For APAC teams using Tekton or Buildkite:

# Tekton Task step / Buildkite command
kubescape scan yaml ./k8s/ \
  --framework nsa,mitre \
  --severity-threshold high \
  --format json \
  --output /workspace/kubescape-results.json

# Check exit code — Kubescape exits non-zero if threshold exceeded
if [ $? -ne 0 ]; then
  echo "APAC Kubernetes security scan failed — blocking deployment"
  exit 1
fi

RBAC visualisation for APAC multi-tenant clusters

APAC enterprises running multi-tenant Kubernetes clusters (multiple APAC business units sharing cluster infrastructure) frequently have RBAC configurations that accumulate permission grants over time, creating privilege escalation paths invisible to manual review:

# Kubescape RBAC visualisation
kubescape scan control C-0036 --format json  # check for cluster-admin usage
kubescape scan control C-0054 --format json  # check for wildcard RBAC

# Export RBAC graph for APAC security team review
kubescape viz rbac --kubeconfig ~/.kube/apac-prod.yaml \
  --output apac-rbac-graph.png

Kubescape's RBAC control checks identify the specific APAC ServiceAccount-to-ClusterRole bindings that create elevation paths — enabling APAC platform security teams to prioritise RBAC remediation by risk rather than auditing every binding manually.


Layer 2: Cosign — APAC Container Supply Chain Security

Keyless signing in APAC CI/CD

Cosign keyless signing uses the APAC CI/CD system's OIDC identity (GitHub Actions, GitLab CI, Google Cloud Build) to obtain a short-lived certificate from Sigstore's Fulcio CA — no long-lived keys to manage, rotate, or secure:

# GitHub Actions: sign APAC container image after Buildah build
- name: Sign APAC image
  if: github.ref == 'refs/heads/main'
  env:
    DIGEST: ${{ steps.build.outputs.image-digest }}
  run: |
    # Keyless signing — uses GitHub Actions OIDC token automatically
    cosign sign \
      --yes \
      apac-registry.example.com/app@${DIGEST}

    echo "APAC image signed: apac-registry.example.com/app@${DIGEST}"

Keyless signing produces a signature that records:

  • The APAC image digest being signed
  • The GitHub Actions workflow identity (repository, workflow, ref)
  • The OIDC token claim showing the signing was from an approved APAC CI/CD workflow
  • A Rekor transparency log entry with timestamp

Any APAC infrastructure team can verify this signature:

# Verify APAC container image signature
cosign verify \
  --certificate-identity-regexp "https://github.com/apac-org/app-repo/.github/workflows/.*" \
  --certificate-oidc-issuer https://token.actions.githubusercontent.com \
  apac-registry.example.com/app@sha256:abc123...

This verification confirms the APAC image was signed by a GitHub Actions workflow in the approved APAC repository — not by a developer's local machine or a compromised build system.

Attaching SBOMs with Cosign

Cosign's SBOM attachment enables APAC DevSecOps teams to store Software Bill of Materials documents alongside APAC container images in OCI registries for supply chain documentation:

# Generate SBOM with Syft
syft apac-registry.example.com/app:v1.2.3 -o spdx-json > apac-app-sbom.spdx.json

# Attach SBOM to APAC image in OCI registry using Cosign
cosign attach sbom \
  --sbom apac-app-sbom.spdx.json \
  --type spdx \
  apac-registry.example.com/app@sha256:abc123...

# Sign the SBOM attestation for tamper evidence
cosign attest \
  --predicate apac-app-sbom.spdx.json \
  --type spdxjson \
  apac-registry.example.com/app@sha256:abc123...

APAC regulators and auditors increasingly request SBOM documentation for software deployed in critical APAC infrastructure — Cosign SBOM attestations stored in OCI registries provide a distribution mechanism that keeps SBOMs co-located with the APAC images they document.

Air-gapped APAC environments with self-hosted Sigstore

APAC organisations with air-gapped Kubernetes deployments (common in APAC FSI, defence-adjacent, and government sectors) cannot reach Sigstore's public Fulcio CA and Rekor transparency log. Self-hosted Sigstore deployment using the sigstore-helm-operator provides the same keyless signing capability within the APAC network perimeter:

# Deploy self-hosted Sigstore components to APAC cluster
helm install fulcio sigstore/fulcio \
  --namespace sigstore-system \
  --values apac-fulcio-values.yaml

helm install rekor sigstore/rekor \
  --namespace sigstore-system \
  --values apac-rekor-values.yaml

# Configure Cosign to use APAC internal Sigstore
export COSIGN_MIRROR=https://fulcio.internal.apac.example.com
export REKOR_URL=https://rekor.internal.apac.example.com
cosign sign --yes apac-internal-registry.example.com/app@${DIGEST}

Layer 3: Kyverno — APAC Admission Policy Enforcement

Image signing verification as admission policy

With Cosign signing images in APAC CI/CD, Kyverno enforces signature verification as a Kubernetes admission control policy — blocking any APAC pod that references an unsigned image from being admitted to production:

# Kyverno ClusterPolicy: require Cosign signature on APAC production images
apiVersion: kyverno.io/v1
kind: ClusterPolicy
metadata:
  name: apac-require-signed-images
spec:
  validationFailureAction: Enforce
  background: false
  rules:
    - name: verify-apac-image-signature
      match:
        any:
          - resources:
              kinds: [Pod]
              namespaces: [apac-production, apac-staging]
      verifyImages:
        - imageReferences:
            - "apac-registry.example.com/app*"
          attestors:
            - count: 1
              entries:
                - keyless:
                    subject: "https://github.com/apac-org/app-repo/.github/workflows/build.yaml@refs/heads/main"
                    issuer: "https://token.actions.githubusercontent.com"
                    rekor:
                      url: https://rekor.sigstore.dev

This Kyverno policy blocks any APAC pod referencing an image from apac-registry.example.com/app* that does not have a valid Cosign keyless signature from the approved GitHub Actions workflow. An APAC developer who bypasses the CI/CD pipeline and manually pushes an unsigned image cannot deploy it to APAC production or staging environments.

Enforcing APAC security baselines

Beyond image signing, Kyverno policies enforce the security baseline controls that Kubescape identifies as misconfigured:

# Kyverno ClusterPolicy: enforce Pod Security Standards for APAC
apiVersion: kyverno.io/v1
kind: ClusterPolicy
metadata:
  name: apac-pod-security-baseline
spec:
  validationFailureAction: Enforce
  rules:
    # Rule 1: containers must run as non-root
    - name: require-non-root
      match:
        resources: {kinds: [Pod], namespaces: [apac-production]}
      validate:
        message: "APAC containers must run as non-root"
        pattern:
          spec:
            containers:
              - securityContext:
                  runAsNonRoot: true
    # Rule 2: containers must have resource limits
    - name: require-resource-limits
      match:
        resources: {kinds: [Pod], namespaces: [apac-production]}
      validate:
        message: "APAC containers must specify CPU and memory limits"
        pattern:
          spec:
            containers:
              - resources:
                  limits:
                    cpu: "?*"
                    memory: "?*"

Auto-generating namespace security resources

Kyverno generation policies create required security resources (NetworkPolicy, ResourceQuota) automatically when APAC teams create new namespaces — ensuring every APAC namespace has baseline isolation without requiring APAC application teams to manage it:

# Kyverno ClusterPolicy: generate default NetworkPolicy for new APAC namespaces
apiVersion: kyverno.io/v1
kind: ClusterPolicy
metadata:
  name: apac-default-network-policy
spec:
  rules:
    - name: generate-apac-namespace-netpol
      match:
        resources:
          kinds: [Namespace]
          selector:
            matchLabels:
              apac-team: "true"    # apply to all APAC team namespaces
      generate:
        kind: NetworkPolicy
        name: default-deny-ingress
        namespace: "{{request.object.metadata.name}}"
        data:
          spec:
            podSelector: {}        # select all pods
            policyTypes: [Ingress] # default deny all ingress

This policy ensures every APAC namespace labelled apac-team: "true" gets a default-deny NetworkPolicy at creation — implementing the network isolation baseline that Kubescape checks for and Cosign supply chain security assumes.


Integrating the Three Tools: APAC DevSecOps Pipeline

The full Kubescape → Cosign → Kyverno workflow for APAC Kubernetes DevSecOps:

Step 1 — Baseline (one-time): Run Kubescape cluster scan against NSA/CISA and CIS frameworks. Remediate high-severity findings (root containers, missing limits, RBAC over-permission). Configure Kyverno policies to enforce the remediated baseline. Accept Kubescape findings as the new policy ground truth.

Step 2 — Build-time (per APAC CI/CD run): Kubescape scans Kubernetes manifests in PR CI. Buildah (or Docker) builds the container image. Cosign signs the image digest after successful build and push to APAC dev registry. Skopeo copies signed image (with signatures) from dev to staging registry.

Step 3 — Deploy-time (per APAC staging/production deployment): Kyverno admission webhook intercepts the APAC pod creation request. Kyverno verifies the container image Cosign signature against approved CI/CD workflow identity. Kyverno validates pod security context (non-root, resource limits, no privileged). Kyverno validates image registry allowlist (only approved APAC registries). If all checks pass, pod is admitted; if any fails, admission is denied with a clear error message.

Step 4 — Continuous (periodic): Kubescape Operator runs periodic cluster scans (daily or weekly) producing updated APAC security posture reports. New Kubescape findings trigger Kyverno policy updates if new controls require enforcement. Gremlin or Chaos Mesh chaos experiments test that Kyverno webhook failures do not cascade to APAC service availability issues.


APAC Compliance Mapping

APAC Framework Kubescape Coverage Cosign Coverage Kyverno Coverage
NSA/CISA Kubernetes Hardening ✓ Full framework ✓ Policy enforcement
CIS Kubernetes Benchmark ✓ v1.23+ ✓ Baseline enforcement
MITRE ATT&CK for Kubernetes ✓ 20 APAC techniques ✓ Supply chain controls
MAS TRM (Singapore) Partial (risk evidence) ✓ Provenance evidence ✓ Access control
HKMA SCR (Hong Kong) Partial (risk evidence) ✓ Supply chain ✓ Change control
CISA/APAC AI Security Guidance ✓ AI system hardening ✓ Model provenance ✓ Supply chain gates

For APAC FSI firms preparing MAS TRM or HKMA SCR submissions, the combination of Kubescape CIS Kubernetes Benchmark reports + Cosign SBOM attestations + Kyverno policy configuration provides the security evidence artifacts that support APAC technology risk assessment.


Related APAC Kubernetes Security Resources

For the runtime and execution infrastructure where these security policies are enforced, see the APAC container image tooling guide covering Podman, Buildah, and Skopeo — covering the daemonless build tooling that integrates with Cosign signing in APAC CI/CD.

For the chaos engineering validation that verifies APAC Kyverno policies do not create Kubernetes availability issues, see the APAC chaos engineering guide covering Chaos Mesh, LitmusChaos, and Gremlin.

For the GitOps deployment framework that deploys Kyverno policies and application manifests to APAC Kubernetes clusters, see the APAC Kubernetes GitOps deployment guide covering Argo CD, Argo Rollouts, and Velero.

Beyond this insight

Cross-reference our practice depth.

If this article matches your stage of thinking, the underlying capabilities ship across all six pillars, ten verticals, and nine Asian markets.

Keep reading

Related reading

Blog

APAC Computer Vision Deployment Guide 2026: Ultralytics, LandingAI, and Roboflow Inference

A practitioner guide for APAC ML and engineering teams building and deploying computer vision systems in 2026 — covering Ultralytics YOLO as the state-of-the-art real-time CV framework for training, fine-tuning, and exporting YOLO models to TensorRT, ONNX, and TFLite for APAC edge and cloud deployment with one Python API; LandingAI as a no-code visual inspection platform enabling APAC factory quality engineers to build defect detection models using active learning with 50-200 labeled images and no ML expertise, with edge deployment for on-premise factory inference; and Roboflow Inference as an open-source CV model serving engine that deploys YOLO, GroundingDINO, and SAM2 as Docker APIs with one command, with Workflows for chaining multi-model CV pipelines into single API calls for APAC engineering teams.

Blog

APAC ML Experiment Tracking and Data Versioning Guide 2026: DagsHub, Aim, and DVC

A practitioner guide for APAC data science teams implementing ML reproducibility through data versioning and experiment tracking in 2026 — covering DVC as a Git-compatible data version control tool that tracks large datasets and model artifacts in APAC cloud storage while storing lightweight metadata in Git, enabling reproducible ML pipelines with pipeline stage caching that skips unchanged preprocessing stages; DagsHub as an integrated ML project collaboration platform combining Git hosting, DVC data versioning, MLflow-compatible experiment tracking, and model registry in a GitHub-like interface; and Aim as an open-source self-hosted ML experiment tracker providing APAC regulated industry teams with complete data sovereignty over training metadata, rich run comparison, and hyperparameter visualization without cloud vendor dependency.

Blog

APAC AI Podcast Production Guide 2026: Podcastle, Cleanvoice AI, and Alitu

A practitioner guide for APAC thought leaders, corporate communicators, and content teams launching AI-assisted podcast production workflows in 2026 — covering Podcastle as an AI podcast recording platform with remote multi-track recording for distributed APAC guest networks, AI audio enhancement for non-studio recordings, and transcript-based text editing that removes audio mistakes by deleting transcript text; Cleanvoice AI as a specialized audio cleanup service that automatically removes filler words, mouth noises, dead air, and stutters from APAC podcast recordings via API, with a case study showing 54 hours of editor time saved on 12 back episodes; and Alitu as an all-in-one podcast production and hosting platform where non-technical APAC creators record, clean, assemble, and publish to Apple Podcasts and Spotify in under 90 minutes total without audio engineering knowledge.

Want this applied to your firm?

We use these frameworks daily in client engagements. Let's see what they look like for your stage and market.