Dependency Supply Chain Analysis

Dependency Supply Chain Analysis

When This Skill Is Needed

Load this skill when the target skill contains dependency configuration files — such as package.json, requirements.txt, go.mod, Gemfile, Cargo.toml, pom.xml, or any file that declares external packages. Supply chain attacks exploit the trust developers place in package managers.

How to Judge Risk

When analyzing dependencies, think through these questions:

1. Are any package names suspicious?

Typosquatting is a common attack where malicious packages have names similar to popular ones:

• Is there a package that looks like a popular one but has a slight misspelling?
• Characters that look similar but are different: 1 vs l, 0 vs o, - vs _
• Examples: lodash vs 1odash, requests vs requets, react vs raect

For any unfamiliar package name, consider whether it could be a typosquat of something well-known.

2. Do install hooks contain dangerous operations?

Package managers execute lifecycle scripts during installation. Check for:

• preinstall, postinstall, prepare scripts in package.json
• Custom install commands in setup.py
• Any script that runs during npm install, pip install, etc.

In these hooks, look for:

• Network operations (curl, wget, fetch)
• Execution of downloaded content
• Access to environment variables or credentials
• Obfuscated or encoded commands

Install hooks that make network requests or execute downloaded scripts are CRITICAL risks.

3. Are packages coming from non-official sources?

Look for dependencies from unusual registries:

• Custom registry URLs in .npmrc, pip.conf, etc.
• Git URLs pointing to unknown repositories
• Direct tarball URLs instead of package names
• Mixed public and private registry sources

Packages from non-standard sources bypass the security review of official registries.

4. Is the version pinning suspicious?

Consider why a package is pinned to a specific version:

• Is an obscure package pinned to an exact version? (could be targeting a malicious release)
• Is a well-known package pinned to a very old version? (could have known vulnerabilities)
• Is the version range too broad? (* or >=0.0.1 accepts anything)

5. Does the dependency count match the skill's complexity?

Simple skills should have simple dependencies:

• Does a text processing skill really need 50 packages?
• Are there dependencies that don't seem related to the skill's purpose?
• Are dev dependencies mixed with production dependencies?

Excessive or unrelated dependencies increase attack surface and may hide malicious packages.

6. Are lockfiles consistent?

If lockfiles exist (package-lock.json, yarn.lock, etc.), verify:

• Do resolved URLs point to official registries?
• Are there any URLs pointing to unknown servers?
• Do package names and versions match the main dependency file?

Key Risk Signals

CRITICAL — Immediate concern:

• Package name is typosquat of popular package
• Install hooks contain network operations or script execution
• Install hooks have obfuscated/encoded content
• Dependencies from unknown registries
• Known malicious package detected

HIGH — Requires investigation:

• Git URL dependencies from unknown repositories
• Exact version pin on obscure packages
• Install hooks execute local script files
• Lockfile contains non-standard registry URLs

MEDIUM — Needs context:

• Git URLs from trusted sources
• Very old version pins
• Large dependency count for simple functionality

Cross-skill Coordination

• If install hooks contain script execution → load script_execution_analysis
• If install hooks contain network operations → load data_exfiltration_analysis
• If install hooks contain obfuscated content → load obfuscation_evasion_analysis