Get started

OpenSourceOM Team

Toxic Combinations in AWS and Azure: When Low-Risk Findings Become Critical

Learn what toxic combination cloud security means in AWS and Azure — misconfiguration pairs that create real attack paths, and how security graphs detect them.

  • toxic combination
  • AWS security
  • Azure security
  • cloud misconfiguration
  • attack path analysis

In cloud security, the dangerous issues are often pairs and chains — not single misconfigurations. A medium-severity setting plus a high-privilege identity plus internet exposure becomes a toxic combination: individually defensible in a ticket comment, collectively exploitable in an afternoon.

Toxic combination cloud security is the named pairs in AWS and Azure (public + admin, and so on). How you rank a full finding queue is how to prioritize cloud vulnerabilities. How the graph represents hops is attack path analysis.

What makes a combination “toxic”?

A combination is toxic when:

  1. Two or more findings are low or medium in isolation
  2. Together they form a complete or partial attack path
  3. Impact reaches sensitive data, admin control, or lateral movement

Example:

Finding alone Severity alone Combined
EC2 with outdated package Medium —
EC2 internet-facing Medium Critical path
Instance profile with S3 read on prod bucket High if seen with EC2 Data exfiltration path

Toxic combinations in AWS

1. Internet exposure + IAM instance profile

Pattern:

  • Security group allows 0.0.0.0/0 on 443 or 22
  • EC2 has IMDSv1 enabled or SSRF-vulnerable app
  • Instance profile can s3:GetObject on * or assume powerful role

Why toxic: Remote code execution or credential theft → immediate cloud API access.

Fix: Restrict SG, enforce IMDSv2, scope instance profile to least privilege.

2. Public S3 + sensitive data classification

Pattern:

  • Bucket ACL/policy allows public read or list
  • Objects tagged or scanned as PII/financial data

Why toxic: Direct data breach without lateral movement.

Fix: Block public access at account level; encrypt and classify data.

3. Lambda URL / API Gateway + over-privileged execution role

Pattern:

  • Public function URL or unauthenticated API stage
  • Execution role includes dynamodb:*, secretsmanager:GetSecretValue, or cross-account assume

Why toxic: Serverless entry points are easy to miss in CSPM sweeps.

Fix: Authenticate ingress; shrink execution role.

4. Cross-account role trust + external compromise

Pattern:

  • Role trusts entire external account or * principal
  • Role has admin or data access in your account

Why toxic: Supply-chain or partner account breach becomes your breach.

Fix: External ID, strict principal ARNs, least privilege.

5. EKS public endpoint + weak RBAC

Pattern:

  • Cluster API publicly reachable
  • system:masters or cluster-admin bound to broad groups

Why toxic: Kubernetes control plane becomes internet attack surface.

Fix: Private endpoint, IP allowlists, tighten RBAC.

Toxic combinations in Azure

1. Storage account public blob + confidential containers

Pattern:

  • Anonymous blob access enabled
  • Containers hold backups or exports with credentials

Fix: Disable anonymous access; use private endpoints.

2. NSG allowing RDP/SSH from Internet + managed identity

Pattern:

  • NSG rule Internet → 3389/22
  • VM managed identity has Key Vault secrets read or SQL admin

Fix: Just-in-time access, bastion hosts, scope identities.

3. Over-permissive service principal + exposed app registration

Pattern:

  • App registration secret in repo or logs
  • SP has Contributor or User Access Administrator on subscription

Fix: Certificate auth, conditional access, role reduction.

4. Azure AD role + weak conditional access

Pattern:

  • Global Reader or Helpdesk admin with password reset rights
  • No MFA for privileged roles from untrusted locations

Fix: Privileged Identity Management (PIM), MFA, CA policies.

5. App Service public + connection string in app settings

Pattern:

  • *.azurewebsites.net reachable
  • SQL connection string with db admin in configuration

Fix: Private Link, Key Vault references, managed identity to SQL.

AWS security misconfiguration vs. toxic combination

Single misconfiguration Often rated Toxic when paired with
Open port 443 to world Medium Known RCE CVE on service
Unused IAM user with keys Low Keys in public GitHub repo
VPC flow logs disabled Low Active data exfil you cannot see
Encryption off on disk Medium Snapshot shared to external account

CSPM flags the left column. Attack path analysis flags the row.

How security graphs detect toxic combinations

Graph platforms model:

[Internet] --REACHABLE--> [VM + CVE] --ASSUMES--> [Role] --CAN_ACCESS--> [Datastore]

Toxic combination cloud security rules are graph patterns:

  • internet_exposed AND critical_cve AND path_to_datastore
  • public_storage AND sensitive_tag
  • admin_role AND assumable_from_external_account

OpenSourceOM targets queryable patterns like these in an open, self-hosted graph.

Prioritization playbook

  1. List all internet-exposed assets (CSPM)
  2. Attach identities each asset can use (CIEM)
  3. Overlay vulnerabilities (CWPP/scanner)
  4. Run path queries to sensitive data
  5. Remediate combinations — break any edge in the path

One broken edge (remove exposure, patch, or shrink role) collapses the chain.

Prevention: shift-left and guardrails

  • IaC policies — block 0.0.0.0/0 on admin ports in CI
  • Account guardrails — S3 Block Public Access, Azure Policy
  • Identity boundaries — permission sets per environment
  • Graph continuous monitoring — drift breaks old assumptions daily

Key takeaways

  • Toxic combinations turn medium findings into critical breach paths
  • AWS security misconfiguration and Azure posture issues matter most in pairs
  • CNAPP and security graph tools exist to find combinations, not just singles
  • Fix paths, not just tickets — break reachability, identity, or vulnerability links

Related: How to prioritize cloud vulnerabilities · CIEM explained