Malicious Package Detection

Detect and block software supply chain attacks

Continuously evaluate open source packages for malicious code and risky behaviors, and block malware before it enters your codebase.

How Endor Labs blocks malicious packages

1

Detect malicious dependencies

Endor Labs scans every new package uploaded to npm, PyPI, and other registries for malicious code, typosquats, and more.

2

Block threats with Package Firewall

Block policy-violating packages in real time using proprietary intelligence and rapid analysis pipelines.

3

Enforce proactive policies

Define proactive policies to automatically block or allow packages based on your risk tolerance.

Endor Labs catches malicious dependencies before we even hear about a CVE. Their security research team goes beyond automated detection to help us verify the threat so we can act early and decisively.”

Aman Sirohi

SVP - Chief Security Officer & Platform, People.ai

Detect

Detect software supply chain attacks early

Endor Labs’ threat intelligence catches malware before it appears in public databases, protecting developer machines and CI pipelines.

  • Catch compromised maintainer accounts, typosquats, and more
  • Real-time analysis of every newly uploaded package to npm, PyPI, and other popular registries
  • Multi-agent behavioral analysis combines agentic reasoning, static analysis, and more than 150+ other signals

BLOCK

Block risky packages 

Package Firewall blocks risky packages from reaching developer machines, agentic cloud environments, or CI pipelines.

  • Prevent developers and AI coding agents from installing risky packages before they reach the sandbox or cloud environments. 
  • Native integration with JFrog, Nexus, AWS CodeArtifact, and package managers — no workflow changes
  • Invisible to developers: safe packages flow through, risky ones get blocked or flagged automatically

pROTECT

Enforce proactive best practices

Proactively enforce policies for developers and AI agents before vulnerabilities enter your environment.

  • Configurable policies for cooldown periods, unpinned dependencies, unused dependencies, non-compliant licenses, and any custom risk signals
  • Govern both developers and AI agents using a unified policy engine
  • Audit and report every download attempt across integrations: package, version, ecosystem, policy triggered, and block status
How does Endor Labs identify malicious packages?

Every newly published package across npm, PyPI, and other registries is analyzed using agentic reasoning, static analysis, behavioral inspection, and 150+ other signals — catching malicious code, hidden payloads, and suspicious maintainer activity that traditional scanners miss.

What types of supply chain attacks does Endor Labs catch?

Compromised maintainer accounts, typosquats, dependency confusion, install-time scripts, obfuscated payloads, credential stealers, and other malicious behaviors across open-source ecosystems.

How is malicious package detection different from traditional SCA?

Traditional SCA flags known vulnerabilities after CVEs are published. Endor Labs detects malicious packages in real time at upload, then blocks them before they reach developer machines or CI pipelines.

Can we block packages that aren't confirmed malware?

Yes. Define policies based on cooldown periods, unpinned dependencies, unused dependencies, license type, maintainer reputation, and other risk signals — not just malware classification. Enforce your own risk standards beyond what's publicly known to be malicious.

How does Endor Labs protect AI coding agents from installing malicious packages?

The same policy engine governs both human developers and AI coding agents, so packages installed by Claude Code, Cursor, or other AI tools face identical guardrails before reaching sandbox or cloud environments.

Will malicious package blocking slow down our developers?

No. Package Firewall runs in the background between your artifact repository and the public registry — safe packages flow through transparently, and developers don't change how they work.

What does a developer see when a package is blocked?

When connected directly to the firewall, they see an HTTP 403 Forbidden error with a clear message. When going through some registries (like Artifactory), that 403 may appear as a 404 Not Found in the CLI.

Does it work with our existing artifact repository?

Yes. Endor Labs’ Package Firewall integrates with JFrog Artifactory today, with support for Nexus and AWS CodeArtifact coming soon. Private packages are fetched from your artifact repository; public packages are checked through the proxy.

AppSec for The Software Development Revolution