🚀 Submit a Challenge — SecDim AppSec Village CTF at DEF CON 34 and Win a ROG Xbox Ally

Cloud Security for Developers

3h 45mBeginnerKubernetesCloudDevSecOpsAWSAzureGCP

The average time to detect a cloud breach is over 200 days, and most of those breaches started not with a sophisticated exploit, but with a misconfigured permission or an exposed credential. Cloud providers secure the infrastructure. You secure everything you deploy on it. This course covers the insecure configurations that cause real-world cloud breaches -- overprivileged IAM roles, exposed storage, vulnerable serverless functions, and missing audit trails -- across AWS, GCP, and Azure. You will work through seven modules grounded in incidents at Capital One, Uber, and Microsoft, then apply what you have learned in hands-on labs where you exploit live cloud infrastructure and fix the configuration that made the attack possible.

Topics

  1. Introduction

    Lesson5m

    Overview of the course, learning objectives and syllabus.

  2. Shared Responsibility Gap

    Lesson15m

    Where the cloud provider's security responsibility ends and yours begins -- and the incidents that result from misunderstanding the boundary.

  3. Credential Theft and Privilege Abuse

    Lesson20m

    How overprivileged IAM roles amplify small compromises into full account breaches, and why the platform credential chain eliminates the need for hardcoded keys.

  4. Overprivileged Cloud Identity Lab

    Lab30m

    A file processor service passes user input directly to a shell command. Exploit the command injection vulnerability to exfiltrate the workload's cloud identity credentials -- from Lambda environment variables on AWS and from the instance metadata service on Azure and GCP. Use those credentials from your own machine to demonstrate the full scope of access granted by an admin-level role. Then scope the identity to the minimum permissions the application actually requires.

  5. Insecure Infrastructure as Code

    Lesson20m

    The most common insecure patterns in Terraform configuration, why AI code generation reproduces them reliably, and how static analysis catches them before deployment.

  6. Unintended Network Exposure and SSRF

    Lesson20m

    How public subnets, permissive firewall rules, and the misunderstood boundaries of Kubernetes namespaces expose cloud resources -- and how SSRF turns a web vulnerability into credential theft via the instance metadata endpoint.

  7. SSRF and Metadata Credential Theft

    Lab30m

    A link preview service accepts any URL the caller supplies and fetches it server-side. Exploit the SSRF vulnerability to reach the cloud instance metadata endpoint, retrieve the IAM role credentials, and use them outside the cloud environment. Then fix the two infrastructure misconfigurations that make the exploit dangerous: the metadata endpoint accepting unauthenticated requests, and the attached identity holding far more access than the application needs.

  8. Insecure Storage Configuration

    Lesson20m

    The two-layer access control model for object storage on all three platforms, how SAS tokens and presigned URLs are misconfigured, and what the Microsoft AI Research incident reveals about scope.

  9. Public Cloud Storage Access Lab

    Lab30m

    A documentation site was deployed on cloud object storage and access controls were disabled to resolve an Access Denied error during setup. A database credentials file was later uploaded alongside the public docs. Exploit the misconfigured storage to list the bucket contents and retrieve the credentials file without any authentication. Then fix the storage configuration to block all anonymous access.

  10. Insecure Serverless Configuration

    Lesson20m

    Unauthenticated function triggers, overprivileged execution roles shared across multiple functions, event injection, and secrets stored in environment variables.

  11. Cloud Audit Logging and Observability

    Lesson15m

    Which audit trails to enable, the difference between management and data events, how to isolate logs from the account they describe, and what cloud-native detection services add beyond manual alerting.