Skip to content

Attack Chain Detection

cloud-audit correlates individual findings into exploitable attack paths. Instead of a flat list of 94 findings, you see the 3-5 attack chains that actually matter.

How It Works

Individual findings are correlated using 31 rules based on:

The engine collects resource relationships (EC2 instance roles, Lambda roles, OIDC trust policies) with lightweight API calls and matches them against known attack patterns.

Example

A public security group alone is a finding. IMDSv1 alone is a finding. But together on the same EC2 instance with an admin IAM role, they form an attack chain:

Internet --> Public SG --> EC2 (IMDSv1) --> Admin IAM Creds --> Account Takeover aws-vpc-002 aws-ec2-004 Detected: AC-01, AC-02

All 31 Rules

Tier 1: Internet Exposure + Privilege

ID Name Severity Component Checks
AC-01 Internet-Exposed Admin Instance CRITICAL aws-vpc-002 + EC2 admin IAM role
AC-02 SSRF to Credential Theft CRITICAL aws-vpc-002 + aws-ec2-004 (same instance)
AC-05 Public Lambda with Admin Access CRITICAL aws-lambda-001 + Lambda admin role
AC-07 CI/CD to Admin Takeover CRITICAL aws-iam-007 + admin policy on role

Tier 2: Missing Controls

ID Name Severity Component Checks
AC-09 Unmonitored Admin Access CRITICAL aws-iam-001 + aws-ct-001
AC-10 Completely Blind Admin CRITICAL aws-iam-001 + aws-ct-001 + aws-gd-001
AC-11 Zero Security Visibility HIGH aws-ct-001 + aws-gd-001 + aws-cfg-001
AC-12 Admin Without MFA CRITICAL aws-iam-005 + aws-iam-002
AC-13 Wide Open and Unmonitored Network HIGH aws-vpc-002 + aws-vpc-003
AC-14 No Network Security Layers HIGH aws-vpc-004 + aws-vpc-002 + aws-vpc-003

Tier 3: Data Protection

ID Name Severity Component Checks
AC-17 Exposed Database Without Audit Trail CRITICAL aws-rds-001 + aws-rds-002 + aws-ct-001

Tier 4: Container & Secrets

ID Name Severity Component Checks
AC-19 Container Breakout Path CRITICAL aws-ecs-001 + aws-ecs-003
AC-20 Unmonitored Container Access HIGH aws-ecs-002 + aws-ecs-003
AC-21 Secrets in Plaintext Across Services HIGH aws-ssm-002 + aws-lambda-003

Tier 5: CI/CD

ID Name Severity Component Checks
AC-23 CI/CD Data Exfiltration HIGH aws-iam-007 + S3 policy on role
AC-24 CI/CD Lateral Movement HIGH aws-iam-007 + EC2 policy on role

Tier 6: CIS Compliance Chains

ID Name Severity Component Checks
AC-25 Root Access Keys Without Audit Trail CRITICAL aws-iam-008 + aws-ct-001
AC-26 Unmonitored Admin Escalation Path CRITICAL aws-iam-005 + aws-iam-002 + aws-cw-001
AC-27 Default Network Access Without Logging HIGH aws-vpc-005 + aws-vpc-003
AC-28 External Access Without Analysis HIGH aws-iam-007 + aws-iam-012

Tier 7: Infrastructure Hygiene

ID Name Severity Component Checks
AC-29 Unpatched Instance Exposed to Internet CRITICAL aws-ssm-003 + aws-vpc-002
AC-30 Unpatched Without Vulnerability Scanning HIGH aws-ssm-003 + aws-inspector-001
AC-31 Internet-Exposed Without WAF or Flow Logs HIGH aws-waf-001 + aws-vpc-002 + aws-vpc-003
AC-32 CloudTrail Blind Spot — Alarms Non-Functional HIGH aws-ct-008 + aws-cw-001
AC-33 All-Public VPC Without Segmentation HIGH aws-vpc-006 + aws-vpc-002 + aws-vpc-003

Tier 8: IAM Privilege Escalation

ID Name Severity Component Checks
AC-34 PassRole Escalation to Admin CRITICAL aws-iam-018 + PassRole + admin policy on target role
AC-35 Self-Escalation via IAM Policy Modification CRITICAL aws-iam-018 + iam:PutUserPolicy or iam:AttachUserPolicy
AC-36 External Escalation via OIDC + Privilege Escalation CRITICAL aws-iam-007 + aws-iam-018 + escalation path from OIDC role

Tier 9: AI-SPM

ID Name Severity Component Checks
AC-37 AI Model Theft via SageMaker CRITICAL aws-sagemaker-001 + aws-sagemaker-002 + model access
AC-38 LLMjacking - Unauthorized Model Usage HIGH aws-bedrock-001 + missing invocation logging
AC-39 AI Data Poisoning via Unguarded Pipeline HIGH aws-sagemaker-003 + unencrypted training data

Suppression

If AC-10 fires, AC-09 is suppressed (AC-10 is a superset). Similarly, AC-26 suppresses AC-12.

Compliance Integration

Attack chains map to all 6 compliance frameworks. When an attack chain is detected, the compliance report shows which controls are violated. See Compliance Overview for all supported frameworks.