Policy-as-Code with Open Policy Agent
Overview
This skill enables policy-as-code enforcement using Open Policy Agent (OPA) for compliance validation, security policy enforcement, and configuration auditing. OPA provides a unified framework for policy evaluation across cloud-native environments, Kubernetes, CI/CD pipelines, and infrastructure-as-code.
Use OPA to codify security requirements, compliance controls, and organizational standards as executable policies written in Rego. Automatically validate configurations, prevent misconfigurations, and maintain continuous compliance.
Quick Start
Install OPA
bash# macOS brew install opa # Linux curl -L -o opa https://openpolicyagent.org/downloads/latest/opa_linux_amd64 chmod +x opa # Verify installation opa version
Basic Policy Evaluation
bash# Evaluate a policy against input data opa eval --data policy.rego --input input.json 'data.example.allow' # Test policies with unit tests opa test policy.rego policy_test.rego --verbose # Run OPA server for live policy evaluation opa run --server --addr localhost:8181
Core Workflow
Step 1: Define Policy Requirements
Identify compliance requirements and security controls to enforce:
- Compliance frameworks (SOC2, PCI-DSS, GDPR, HIPAA, NIST)
- Kubernetes security policies (pod security, RBAC, network policies)
- Infrastructure-as-code policies (Terraform, CloudFormation)
- Application security policies (API authorization, data access)
- Organizational security standards
Step 2: Write OPA Rego Policies
Create policy files in Rego language. Use the provided templates in assets/ for common patterns:
Example: Kubernetes Pod Security Policy
regopackage kubernetes.admission import future.keywords.contains import future.keywords.if deny[msg] { input.request.kind.kind == "Pod" container := input.request.object.spec.containers[_] container.securityContext.privileged == true msg := sprintf("Privileged containers are not allowed: %v", [container.name]) } deny[msg] { input.request.kind.kind == "Pod" container := input.request.object.spec.containers[_] not container.securityContext.runAsNonRoot msg := sprintf("Container must run as non-root: %v", [container.name]) }
Example: Compliance Control Validation (SOC2)
regopackage compliance.soc2 import future.keywords.if # CC6.1: Logical and physical access controls deny[msg] { input.kind == "Deployment" not input.spec.template.metadata.labels["data-classification"] msg := "SOC2 CC6.1: All deployments must have data-classification label" } # CC6.6: Encryption in transit deny[msg] { input.kind == "Service" input.spec.type == "LoadBalancer" not input.metadata.annotations["service.beta.kubernetes.io/aws-load-balancer-ssl-cert"] msg := "SOC2 CC6.6: LoadBalancer services must use SSL/TLS encryption" }
Step 3: Test Policies with Unit Tests
Write comprehensive tests for policy validation:
regopackage kubernetes.admission_test import data.kubernetes.admission test_deny_privileged_container { input := { "request": { "kind": {"kind": "Pod"}, "object": { "spec": { "containers": [{ "name": "nginx", "securityContext": {"privileged": true} }] } } } } count(admission.deny) > 0 } test_allow_unprivileged_container { input := { "request": { "kind": {"kind": "Pod"}, "object": { "spec": { "containers": [{ "name": "nginx", "securityContext": {"privileged": false, "runAsNonRoot": true} }] } } } } count(admission.deny) == 0 }
Run tests:
bashopa test . --verbose
Step 4: Evaluate Policies Against Configuration
Use the bundled evaluation script for policy validation:
bash# Evaluate single file ./scripts/evaluate_policy.py --policy policies/ --input config.yaml # Evaluate directory of configurations ./scripts/evaluate_policy.py --policy policies/ --input configs/ --recursive # Output results in JSON format for CI/CD integration ./scripts/evaluate_policy.py --policy policies/ --input config.yaml --format json
Or use OPA directly:
bash# Evaluate with formatted output opa eval --data policies/ --input config.yaml --format pretty 'data.compliance.violations' # Bundle evaluation for complex policies opa eval --bundle policies.tar.gz --input config.yaml 'data'
Step 5: Integrate with CI/CD Pipelines
Add policy validation to your CI/CD workflow:
GitHub Actions Example:
yaml- name: Validate Policies uses: open-policy-agent/setup-opa@v2 with: version: latest - name: Run Policy Tests run: opa test policies/ --verbose - name: Evaluate Configuration run: | opa eval --data policies/ --input deployments/ \ --format pretty 'data.compliance.violations' > violations.json if [ $(jq 'length' violations.json) -gt 0 ]; then echo "Policy violations detected!" cat violations.json exit 1 fi
GitLab CI Example:
yamlpolicy-validation: image: openpolicyagent/opa:latest script: - opa test policies/ --verbose - opa eval --data policies/ --input configs/ --format pretty 'data.compliance.violations' artifacts: reports: junit: test-results.xml
Step 6: Deploy as Kubernetes Admission Controller
Enforce policies at cluster level using OPA Gatekeeper:
bash# Install OPA Gatekeeper kubectl apply -f https://raw.githubusercontent.com/open-policy-agent/gatekeeper/master/deploy/gatekeeper.yaml # Apply constraint template kubectl apply -f assets/k8s-constraint-template.yaml # Apply constraint kubectl apply -f assets/k8s-constraint.yaml # Test admission control kubectl apply -f test-pod.yaml # Should be denied if violates policy
Step 7: Monitor Policy Compliance
Generate compliance reports using the bundled reporting script:
bash# Generate compliance report ./scripts/generate_report.py --policy policies/ --audit-logs audit.json --output compliance-report.html # Export violations for SIEM integration ./scripts/generate_report.py --policy policies/ --audit-logs audit.json --format json --output violations.json
Security Considerations
- Policy Versioning: Store policies in version control with change tracking and approval workflows
- Least Privilege: Grant minimal permissions for policy evaluation - OPA should run with read-only access to configurations
- Sensitive Data: Avoid embedding secrets in policies - use external data sources or encrypted configs
- Audit Logging: Log all policy evaluations, violations, and exceptions for compliance auditing
- Policy Testing: Maintain comprehensive test coverage (>80%) for all policy rules
- Separation of Duties: Separate policy authors from policy enforcers; require peer review for policy changes
- Compliance Mapping: Map policies to specific compliance controls (SOC2 CC6.1, PCI-DSS 8.2.1) for audit traceability
Bundled Resources
Scripts (scripts/)
evaluate_policy.py- Evaluate OPA policies against configuration files with formatted outputgenerate_report.py- Generate compliance reports from policy evaluation resultstest_policies.sh- Run OPA policy unit tests with coverage reporting
References (references/)
rego-patterns.md- Common Rego patterns for security and compliance policiescompliance-frameworks.md- Policy templates mapped to SOC2, PCI-DSS, GDPR, HIPAA controlskubernetes-security.md- Kubernetes security policies and admission control patternsiac-policies.md- Infrastructure-as-code policy validation for Terraform, CloudFormation
Assets (assets/)
k8s-pod-security.rego- Kubernetes pod security policy templatek8s-constraint-template.yaml- OPA Gatekeeper constraint templatek8s-constraint.yaml- Example Gatekeeper constraint configurationsoc2-compliance.rego- SOC2 compliance controls as OPA policiespci-dss-compliance.rego- PCI-DSS requirements as OPA policiesgdpr-compliance.rego- GDPR data protection policiesterraform-security.rego- Terraform security best practices policiesci-cd-pipeline.yaml- CI/CD integration examples (GitHub Actions, GitLab CI)
Common Patterns
Pattern 1: Kubernetes Admission Control
Enforce security policies at pod creation time:
regopackage kubernetes.admission deny[msg] { input.request.kind.kind == "Pod" not input.request.object.spec.securityContext.runAsNonRoot msg := "Pods must run as non-root user" }
Pattern 2: Infrastructure-as-Code Validation
Validate Terraform configurations before apply:
regopackage terraform.security deny[msg] { resource := input.resource_changes[_] resource.type == "aws_s3_bucket" not resource.change.after.server_side_encryption_configuration msg := sprintf("S3 bucket %v must have encryption enabled", [resource.name]) }
Pattern 3: Compliance Framework Mapping
Map policies to specific compliance controls:
regopackage compliance.soc2 # SOC2 CC6.1: Logical and physical access controls cc6_1_violations[msg] { input.kind == "RoleBinding" input.roleRef.name == "cluster-admin" msg := sprintf("SOC2 CC6.1 VIOLATION: cluster-admin binding for %v", [input.metadata.name]) }
Pattern 4: Data Classification Enforcement
Enforce data handling policies based on classification:
regopackage data.classification deny[msg] { input.metadata.labels["data-classification"] == "restricted" input.spec.template.spec.volumes[_].hostPath msg := "Restricted data cannot use hostPath volumes" }
Pattern 5: API Authorization Policies
Implement attribute-based access control (ABAC):
regopackage api.authz import future.keywords.if allow if { input.method == "GET" input.path[0] == "public" } allow if { input.method == "GET" input.user.role == "admin" } allow if { input.method == "POST" input.user.role == "editor" input.resource.owner == input.user.id }
Integration Points
- CI/CD Pipelines: GitHub Actions, GitLab CI, Jenkins, CircleCI - validate policies before deployment
- Kubernetes: OPA Gatekeeper admission controller for runtime policy enforcement
- Terraform/IaC: Pre-deployment validation using
conftestor OPA CLI - API Gateways: Kong, Envoy, NGINX - authorize requests using OPA policies
- Monitoring/SIEM: Export policy violations to Splunk, ELK, Datadog for security monitoring
- Compliance Tools: Integrate with compliance platforms for control validation and audit trails
Troubleshooting
Issue: Policy Evaluation Returns Unexpected Results
Solution:
- Enable trace mode:
opa eval --data policy.rego --input input.json --explain full 'data.example.allow' - Validate input data structure matches policy expectations
- Check for typos in policy rules or variable names
- Use
opa fmtto format policies and catch syntax errors
Issue: Kubernetes Admission Control Not Blocking Violations
Solution:
- Verify Gatekeeper is running:
kubectl get pods -n gatekeeper-system - Check constraint status:
kubectl get constraints - Review audit logs:
kubectl logs -n gatekeeper-system -l control-plane=controller-manager - Ensure constraint template is properly defined and matches policy expectations
Issue: Policy Tests Failing
Solution:
- Run tests with verbose output:
opa test . --verbose - Check test input data matches expected format
- Verify policy package names match between policy and test files
- Use
print()statements in Rego for debugging
Issue: Performance Degradation with Large Policy Sets
Solution:
- Use policy bundles:
opa build policies/ -o bundle.tar.gz - Enable partial evaluation for complex policies
- Optimize policy rules to reduce computational complexity
- Index data for faster lookups using
input.keypatterns - Consider splitting large policy sets into separate evaluation domains


