external/anthropic-cybersecurity-skills/skills/auditing-cloud-with-cis-benchmarks/SKILL.md
This skill details how to conduct cloud security audits using Center for Internet Security benchmarks for AWS, Azure, and GCP. It covers interpreting CIS Foundations Benchmark controls, running automated assessments with tools like Prowler and ScoutSuite, remediating failed controls, and maintaining continuous compliance monitoring against CIS v5 for AWS, v4 for Azure, and v4 for GCP.
npx skillsauth add seikaikyo/dash-skills auditing-cloud-with-cis-benchmarksInstall this skill globally with one command. Works with Claude Code, Cursor, and Windsurf.
3 of 9 scanners reported clean
Some scanners were skipped, did not run, or reported a non-clean status. Review each row below.
Do not use for runtime threat detection (see detecting-cloud-threats-with-guardduty), for application-level security testing (see conducting-cloud-penetration-testing), or for compliance frameworks not based on CIS (refer to specific regulatory skill files).
Choose the correct benchmark version for each cloud provider. Current versions as of 2025 include CIS AWS Foundations Benchmark v5.0, CIS Azure Foundations Benchmark v4.0, and CIS GCP Foundations Benchmark v4.0.
CIS Benchmark Coverage Areas:
+-------------------+-------------------------+------------------------+
| Section | AWS v5.0 | Azure v4.0 |
+-------------------+-------------------------+------------------------+
| Identity & Access | IAM policies, MFA, root | Azure AD, RBAC, PIM |
| Logging | CloudTrail, Config | Activity Log, Diag |
| Monitoring | CloudWatch alarms | Defender, Sentinel |
| Networking | VPC, SG, NACLs | NSG, ASG, Firewall |
| Storage | S3 encryption, access | Storage encryption |
| Database | RDS encryption | SQL TDE, auditing |
+-------------------+-------------------------+------------------------+
CIS Profile Levels:
Level 1: Practical security settings that can be implemented without significant
performance impact or reduced functionality
Level 2: Defense-in-depth settings that may reduce functionality or require
additional planning for implementation
Execute comprehensive CIS benchmark scans using Prowler for automated control evaluation across AWS, Azure, and GCP.
# AWS CIS v5.0 assessment
prowler aws \
--compliance cis_5.0_aws \
--profile audit-account \
--output-formats json-ocsf,html,csv \
--output-directory ./cis-audit-$(date +%Y%m%d)
# Azure CIS v4.0 assessment
prowler azure \
--compliance cis_4.0_azure \
--subscription-ids "sub-id-1,sub-id-2" \
--output-formats json-ocsf,html,csv \
--output-directory ./cis-audit-azure-$(date +%Y%m%d)
# GCP CIS v4.0 assessment
prowler gcp \
--compliance cis_4.0_gcp \
--project-ids "project-1,project-2" \
--output-formats json-ocsf,html,csv \
--output-directory ./cis-audit-gcp-$(date +%Y%m%d)
# Multi-account AWS scan using ScoutSuite
scout suite aws \
--profile audit-account \
--report-dir ./scout-report \
--ruleset cis-5.0 \
--force
Analyze audit results by section and severity. Prioritize Level 1 controls first as they represent fundamental security hygiene, then address Level 2 controls for defense in depth.
# Parse Prowler results for failed controls
cat ./cis-audit-*/prowler-output-*.json | \
jq '[.[] | select(.StatusExtended == "FAIL")] | group_by(.CheckID) |
map({control: .[0].CheckID, description: .[0].CheckTitle,
failed_resources: length, severity: .[0].Severity}) |
sort_by(-.failed_resources)'
# Generate compliance score by section
cat ./cis-audit-*/prowler-output-*.json | \
jq 'group_by(.Section) | map({
section: .[0].Section,
total: length,
passed: [.[] | select(.StatusExtended == "PASS")] | length,
failed: [.[] | select(.StatusExtended == "FAIL")] | length,
score: (([.[] | select(.StatusExtended == "PASS")] | length) / length * 100 | round)
})'
Address failed controls starting with the highest impact items. Use AWS Config remediation, Azure Policy, or Terraform to apply fixes systematically.
# CIS 1.4: Ensure no root account access key exists
aws iam list-access-keys --user-name root
# If keys exist, delete them
aws iam delete-access-key --user-name root --access-key-id AKIAEXAMPLE
# CIS 2.1.1: Ensure S3 bucket default encryption is enabled
for bucket in $(aws s3api list-buckets --query 'Buckets[*].Name' --output text); do
aws s3api put-bucket-encryption --bucket "$bucket" \
--server-side-encryption-configuration '{
"Rules": [{"ApplyServerSideEncryptionByDefault": {"SSEAlgorithm": "AES256"}}]
}' 2>/dev/null && echo "Encrypted: $bucket" || echo "FAILED: $bucket"
done
# CIS 3.1: Ensure CloudTrail is enabled in all regions
aws cloudtrail create-trail \
--name organization-trail \
--s3-bucket-name cloudtrail-logs-bucket \
--is-multi-region-trail \
--enable-log-file-validation \
--kms-key-id arn:aws:kms:us-east-1:123456789012:key/key-id
aws cloudtrail start-logging --name organization-trail
# CIS 4.x: Configure CloudWatch metric filters and alarms
aws logs put-metric-filter \
--log-group-name CloudTrail/DefaultLogGroup \
--filter-name UnauthorizedAPICalls \
--filter-pattern '{ ($.errorCode = "*UnauthorizedAccess*") || ($.errorCode = "AccessDenied*") }' \
--metric-transformations metricName=UnauthorizedAPICalls,metricNamespace=CISBenchmark,metricValue=1
Deploy automated compliance monitoring to detect configuration drift between periodic audits. Use AWS Security Hub, Azure Policy, or GCP Security Command Center.
# AWS: Enable CIS v5.0 in Security Hub
aws securityhub batch-enable-standards \
--standards-subscription-requests '[
{"StandardsArn": "arn:aws:securityhub:::ruleset/cis-aws-foundations-benchmark/v/5.0.0"}
]'
# Azure: Assign CIS benchmark policy initiative
az policy assignment create \
--name cis-azure-benchmark \
--scope "/subscriptions/<sub-id>" \
--policy-set-definition "1a5bb27d-173f-493e-9568-eb56638dbd0e" \
--params '{"effect": {"value": "AuditIfNotExists"}}'
# Schedule periodic Prowler assessments
# Run weekly via cron or CI/CD pipeline
0 2 * * 1 prowler aws --compliance cis_5.0_aws --output-formats csv --output-directory /opt/audits/weekly-$(date +\%Y\%m\%d)
| Term | Definition | |------|------------| | CIS Benchmark | Prescriptive security configuration guidelines developed by the Center for Internet Security through community consensus | | Level 1 Profile | Practical security controls implementable without significant performance or functionality impact, representing security hygiene | | Level 2 Profile | Defense-in-depth controls that may restrict functionality and require careful planning before implementation | | Foundations Benchmark | CIS benchmark specifically for cloud providers covering IAM, logging, monitoring, networking, and storage security | | Control ID | Unique numerical identifier for each CIS recommendation (e.g., 1.4 for root access key checks, 2.1.1 for S3 encryption) | | Compliance Score | Percentage of CIS controls in a passing state, tracked over time to measure security posture improvement | | Automated Assessment | Tool-driven evaluation of CIS controls using cloud provider APIs to check resource configurations against benchmark requirements | | Remediation Runbook | Documented step-by-step procedure for fixing a specific failed CIS control, including pre-checks and validation |
Context: A SaaS company pursuing SOC 2 Type II certification needs to demonstrate cloud security controls aligned to CIS benchmarks. The auditor requires evidence of continuous compliance monitoring across 45 AWS accounts.
Approach:
Pitfalls: Remediating controls without testing in a staging environment first can break production workloads. Ignoring Level 2 controls entirely weakens the audit narrative even if they are not strictly required.
CIS Benchmark Audit Report
============================
Cloud Provider: AWS
Benchmark Version: CIS AWS Foundations Benchmark v5.0
Accounts Assessed: 45
Assessment Date: 2025-02-23
Tool: Prowler v4.3.0
OVERALL COMPLIANCE SCORE: 74%
COMPLIANCE BY SECTION:
1. Identity and Access Management: 68% (41/60 controls passed)
2. Storage: 82% (28/34 controls passed)
3. Logging: 91% (20/22 controls passed)
4. Monitoring: 55% (18/33 controls passed)
5. Networking: 78% (32/41 controls passed)
TOP FAILED CONTROLS (by affected accounts):
[1.4] Root account has active access keys - 3/45 accounts
[1.5] MFA not enabled for root account - 2/45 accounts
[2.1.1] S3 default encryption not enabled - 12/45 accounts
[3.1] CloudTrail not multi-region - 8/45 accounts
[4.3] No alarm for root account usage - 28/45 accounts
[5.1] VPC flow logs not enabled - 15/45 accounts
[5.4] Security groups allow 0.0.0.0/0 ingress - 22/45 accounts
REMEDIATION PRIORITY:
Critical (Fix within 7 days): Root access keys, missing root MFA
High (Fix within 30 days): S3 encryption, CloudTrail, VPC flow logs
Medium (Fix within 60 days): CloudWatch alarms, security group restrictions
Low (Fix within 90 days): Level 2 controls, informational items
development
拋棄式 HTML mockup 比稿:產出 2 到 3 個設計立場不同的變體(密度 / 版式 / 強調軸,不是換色),各附取捨說明,最後給有立場的對比結論。適用:「畫個草圖」「比較 A 版 B 版」「先看方向再做」「給我看幾種做法」。要 production 元件或設計已定案時不適用。
tools
需求不明時的意圖萃取訪談:一次一題、每題附上自己的猜測、聽出「真正想要 vs 覺得應該要」,直到能預測使用者反應(約 95% 信心)才動工。適用:需求缺少對象 / 動機 / 成功標準 / 約束,或使用者點名「訪談我」「先確認一下」「我們確定嗎」。明確自足的指示、純資訊查詢、機械性操作不適用。
development
對非平凡決策啟動新鮮 context 對抗審查(找碴不背書),在修正還便宜的時候抓出錯誤方向。適用:高風險改動(production、資安敏感邏輯、不可逆操作)、不熟的程式碼、要宣稱「這樣是安全的 / 可行的」之前。機械性操作與一行修改不適用。
testing
Reference for writing and editing agent skills well — the vocabulary and principles that make a skill predictable. Consult when authoring, reviewing, or pruning a SKILL.md.