CyberKit

Slopsquatting

Also called AI Package Hallucination Squatting.

Registering a malicious package under a name an AI coding tool hallucinates.

What it is

A supply-chain technique where an attacker registers a package under a name that large language models commonly hallucinate when generating code. AI-assisted or agentic coding tools that trust the model's suggested import then unknowingly install the attacker's malicious package instead of a real dependency.

Why it matters

It shifts the trust boundary from a human choosing a dependency to a model confidently inventing one. As AI-assisted and agentic coding tools install packages with less human review, it becomes a viable, scalable path into a build pipeline.

In practice

An AI coding assistant suggests importing a plausible-sounding but nonexistent library; an attacker had already registered that exact name, so the auto-installed package is malicious.

What to look for

  • Packages added from AI suggestions without review.
  • New packages with few downloads and no history.
  • Install scripts in packages you have never used.

Common mistake

Installing whatever the assistant suggests. Check the package exists, its age and its maintainer first.

Practice it nearby

CyberKit has no tool for this term itself. These are the closest places to practice it.

In the field

Well-known tools where this shows up on the job. They are listed to show the landscape, not as a recommendation.

  • Trivy Open source

    Scans images, file systems and code for known vulnerabilities and secrets.

  • Falco Open source

    Detects suspicious behavior inside running containers and Kubernetes.

  • Sigstore Open source

    Signs and verifies software artifacts so you can prove where a build came from.

  • CrowdStrike Falcon Cloud Security Commercial

    Protects cloud workloads and containers, from build to runtime.

Read next

Sources