Lesson 6 of 6 · 17 min · Intermediate

Automated assessment, benchmarks and reporting

CIS Benchmarks, Prowler and ScoutSuite, provider security posture tools, triaging results, attack paths and writing a cloud assessment report.

In this lesson you will

  • Use CIS Benchmarks and open-source scanners such as Prowler and ScoutSuite
  • Interpret provider posture tools: Security Hub, Defender for Cloud and Security Command Center
  • Triage automated results into prioritised, verified findings
  • Write a cloud assessment report with attack paths and remediation

A medium-sized cloud estate can contain thousands of resources across dozens of accounts. No tester reviews them all by hand. Professionals use automated tools to cover breadth, then apply judgement to verify, prioritise and connect the results into attack paths. This final lesson shows you how.

Benchmarks: the baseline

The Center for Internet Security (CIS) publishes free, consensus-based benchmarks for AWS, Azure and Google Cloud (and Kubernetes). They define hundreds of specific checks, for example:

  • MFA is enabled for the root or owner account.
  • No access keys exist for the root account.
  • CloudTrail is enabled in all regions.
  • No security groups allow ingress from 0.0.0.0/0 to port 22 or 3389.
  • Storage buckets are not publicly accessible.

Benchmarks give clients an objective, repeatable baseline, and many compliance frameworks map to them. Providers also publish their own frameworks: the AWS Foundational Security Best Practices, the Microsoft cloud security benchmark and Google's security foundations guidance.

Open-source scanners

Tool Providers Notes
Prowler AWS, Azure, GCP, Kubernetes Hundreds of checks mapped to CIS and compliance frameworks; HTML, CSV and JSON reports
ScoutSuite (NCC Group) AWS, Azure, GCP and others Collects configuration, produces an offline HTML report to explore
Steampipe with compliance mods AWS, Azure, GCP Query cloud configuration with SQL
CloudFox AWS, Azure Attack-path focused enumeration
PMapper / AzureHound AWS / Entra ID and Azure Identity attack-path graphs
Trivy / kube-bench Containers, Kubernetes Images, clusters and CIS Kubernetes checks

All of these run with read-only credentials.

bash
# Prowler against an authorised AWS account, read-only profile, CIS benchmark
prowler aws --profile audit --compliance cis_3.0_aws --output-formats html csv

Provider posture tools

Clients often already have native tools. Review them, don't duplicate them:

  • AWS Security Hub: runs CIS and Foundational Security Best Practices checks across accounts, and aggregates GuardDuty, Inspector and Macie findings.
  • Microsoft Defender for Cloud: posture recommendations and a secure score, plus workload protection plans.
  • Google Security Command Center: misconfigurations, vulnerabilities and threats across the organisation.

Check whether they're enabled organisation-wide, whether findings are triaged, and how many critical findings are old. A secure score that hasn't moved in a year tells you about process, not just configuration.

From tool output to findings

A scanner run produces hundreds of results. Your value is turning them into a short list of real, prioritised findings.

1. Verify

  • Confirm the result is real and current. Tools can misread exceptions, compensating controls and resources that are public by design (such as a website bucket).
  • Check context: a public bucket containing marketing images is very different from one containing database backups.

2. Prioritise by real risk

Weigh each issue by exposure (internet-facing or internal), privilege (what an attacker gains), data sensitivity and exploitability. A "medium" benchmark failure on an internet-facing production database may be your top finding; a "high" on an empty test account may be informational.

3. Group

Report one finding per root cause, with every affected resource listed, rather than 200 separate findings. "MFA is not enforced for 37 IAM users" is one finding, not 37.

4. Connect into attack paths

The most powerful findings combine several issues into a story:

text
1. Web application on EC2 has an SSRF vulnerability          (web test)
2. The instance allows IMDSv1                                  (compute config)
3. The instance role can read every S3 bucket                  (IAM)
4. One bucket holds nightly database backups                   (storage)
5. Data access logging is disabled on that bucket              (logging)
→ An unauthenticated attacker can steal the customer database without leaving a data-access trace.

Each item alone might be rated medium. Together they're critical, and the path shows the client exactly which link to break first.

Writing the cloud assessment report

Follow the reporting structure from the Foundations track, with cloud-specific additions:

  • Scope: accounts, subscriptions and projects; regions; access level; dates; and what was not reviewed.
  • Methodology: configuration review against CIS (state the version), tools used, manual verification, external testing.
  • Executive summary: overall posture, the top attack paths, and three to five strategic recommendations.
  • Attack paths: a diagram or numbered narrative for each.
  • Findings: for each, the affected resources (by ARN or resource ID), risk, evidence (redacted configuration or CLI output), and remediation with the exact setting, policy or command.
  • Benchmark summary: pass/fail counts per CIS section, as an appendix.
  • Strategic recommendations: guardrails (SCPs, Azure Policy, organisation policies), central logging, identity federation, posture management processes.

Remediation that sticks

Recommend preventive guardrails alongside fixes: closing one public bucket fixes today's finding, but an organisation-wide public access block prevents the next hundred. Where possible, give the infrastructure-as-code change (Terraform, CloudFormation, Bicep) so fixes survive the next deployment.

Lab

  1. Run Prowler against your AWS lab account with a read-only profile and the CIS benchmark. Open the HTML report.
  2. Run ScoutSuite against the same account and compare the results. What did each tool find that the other didn't?
  3. Pick the five most important results and verify each manually.
  4. Write a one-page mini report: scope, executive summary, one attack path built from your lab findings, and three findings with remediation.
  5. Clean up: destroy every lab resource and check your billing dashboard.

Track wrap-up

You've completed Cloud Security Assessment. You understand shared responsibility and provider testing rules, can assess IAM, storage, logging, networks, compute and containers across AWS, Azure and Google Cloud, and can turn automated results into verified findings and attack paths.

Where next? Practise with CloudGoat, flAWS2 and AzureGoat, and read our cloud research such as AiTM session theft in Microsoft 365 and agentic AI intrusions in AWS. Then continue with the Active Directory Security track: in hybrid environments, on-premises identity and cloud identity are tightly connected.