M02: Inadequate Supply Chain Security - Attack Vectors

Developer → Package Registry → Build System → App Store → User Device

↓ ↓ ↓ ↓ ↓

Trusts Trusts Trusts Trusts Trusts

↓ ↓ ↓ ↓ ↓

Any compromise in this chain affects everyone downstream

Attack Vector Categories

1. Malicious Package Injection

• Exploits package manager behavior with internal vs public packages
• Attacker publishes package with internal name on public registry
• Build system installs malicious public version instead of internal

• Using trusted brand names to create deceptive packages
• "facebook-sdk-unofficial" appearing legitimate
• Developers trust the name association

2. Legitimate Package Compromise

• Attacker compromises maintainer account credentials
• Publishes malicious update to legitimate package
• All apps auto-updating receive infected version

• Compromising source code repositories
• Injecting malicious code through pull requests
• Social engineering maintainers to merge malicious code

• Attacking CI/CD pipelines
• Injecting malicious code during build process
• Published package differs from source code

3. Transitive Dependency Attacks

• Targeting obscure dependencies deep in dependency tree
• Less scrutiny on packages few developers directly interact with
• Example: Popular package depends on 50+ libraries, one is compromised

• Compromising low-level dependencies
• Malicious code propagates up dependency chain
• Affects thousands of apps indirectly

4. Abandoned Package Exploitation

• Identifying abandoned but widely-used packages
• Social engineering to gain maintainer access
• Publishing "update" with malicious payload

• Registering package names before projects claim them
• Waiting for developers to make assumptions about official packages
• Serving malicious code to unsuspecting developers

5. SDK and Framework Attacks

• Analytics, advertising, or social media SDKs infected
• Automatic updates push malicious code
• Widespread impact across many applications

• Compromising development frameworks
• All apps built with framework are vulnerable
• Long-term persistent access

Attack Scenarios

Scenario 1: The Typosquatting Trap

Attack Flow:

1. Attacker researches popular packages (e.g., "axios" - HTTP client)
2. Creates similar package "axois" with malicious code
3. Developer makes typo during installation
4. Package installed in project
5. Malicious code exfiltrates environment variables
6. API keys and secrets stolen
7. Attacker accesses backend systems

Scenario 2: The Dependency Confusion Attack

Attack Flow:

1. Attacker discovers internal package name from error messages/job postings
2. Publishes package with same name to public npm/PyPI
3. Build system checks public registry first
4. Downloads malicious public package instead of internal
5. Malicious code executes in build environment
6. Source code and secrets exfiltrated
7. Supply chain fully compromised

Scenario 3: The Legitimate Package Hijack

Attack Flow:

1. Attacker identifies popular but under-maintained package
2. Social engineers way into maintainer team
3. Publishes malicious "security update"
4. Apps auto-update to compromised version
5. Malicious SDK collects user data
6. Data sent to attacker-controlled servers
7. Millions of users affected

Scenario 4: The Build Tool Compromise

Attack Flow:

1. Attacker compromises developer's machine via phishing
2. Injects malicious plugin into build tool configuration
3. Plugin modifies compiled application during build
4. Backdoor inserted that's not in source code
5. App published with hidden malicious functionality
6. Source code reviews miss the compromise
7. Users install backdoored application

Attack Chain Analysis

Phase 1: Reconnaissance

• Identifying popular packages and their typo variants
• Discovering internal package names
• Finding vulnerable or abandoned projects
• Researching maintainer accounts and security practices

Phase 2: Compromise

• Publishing malicious packages
• Compromising maintainer accounts via credential theft
• Social engineering to gain commit access
• Exploiting build system vulnerabilities

Phase 3: Distribution

• Developers install compromised packages
• CI/CD systems auto-download malicious dependencies
• Package updates push malicious code to existing installations
• App updates distribute compromised code to users

Phase 4: Exploitation

• Exfiltrating credentials and API keys
• Collecting user data (contacts, location, messages)
• Installing additional malware
• Creating backdoors for persistent access
• Cryptocurrency mining

Phase 5: Persistence

• Backdoors in multiple dependencies
• Compromised build pipelines
• Account access for future updates
• Distributed across app ecosystem

Exploitation Techniques

1. Data Exfiltration

Conceptual Approach:

• Malicious dependency collects sensitive data
• Encodes data to avoid detection
• Sends to attacker-controlled endpoint
• Uses legitimate-looking traffic patterns

• Environment variables with API keys
• User authentication tokens
• Personal user data
• Application source code
• Build artifacts and certificates

2. Backdoor Installation

Conceptual Approach:

• Inject remote code execution capability
• Establish command and control channel
• Hide communication in normal traffic
• Enable future arbitrary code execution

• Remote control of application behavior
• Dynamic payload delivery
• Credential harvesting
• Lateral movement to backend systems

3. Supply Chain Poisoning

Conceptual Approach:

• Compromise one package in dependency tree
• Malicious code spreads to dependent packages
• Creates cascading compromise
• Difficult to trace origin