slopsquatcheck

AI Hallucination Squatting Detector for Python Dependencies

Detect suspicious packages in your requirements.txt or pyproject.toml that may be:

• 🎭 AI hallucination squat targets — packages that LLMs commonly "invent"
• 📛 Typosquatting popular packages
• 🆕 Suspiciously new packages targeting known names
• ⚠️ Low reputation packages (no downloads, no author, no repo)

The Problem

When developers use AI coding assistants (ChatGPT, Claude, Copilot, etc.), the AI sometimes "hallucinates" package names that don't exist. Attackers have learned to exploit this:

1. LLM recommends huggingface-cli (which doesn't exist)
2. Attacker registers huggingface-cli on PyPI with malware
3. Developer runs pip install huggingface-cli
4. Malware executes on developer's machine

This is called "Slopsquatting" or "AI Hallucination Squatting" — and it's a growing supply chain attack vector.

Installation

pip install slopsquatcheck

Or run directly (zero dependencies):

curl -O https://raw.githubusercontent.com/kriskimmerle/slopsquatcheck/main/slopsquatcheck.py
python slopsquatcheck.py requirements.txt

Quick Start

# Scan a requirements file
slopsquatcheck requirements.txt

# Scan pyproject.toml
slopsquatcheck pyproject.toml

# Check a single package
slopsquatcheck --package some-suspicious-lib

# CI mode: fail if high-risk packages found
slopsquatcheck requirements.txt --check --min-score 40

Example Output

============================================================
slopsquatcheck - AI Hallucination Squat Detector
============================================================
Packages scanned: 5
Total findings: 3

Findings by severity:
  CRITICAL: 1
  HIGH: 1
  MEDIUM: 1

------------------------------------------------------------
📦 huggingface-cli (Risk: 60/100, Grade: F)

  🔴 [SS01] CRITICAL: Package is a known AI hallucination target
      reason: This package name is commonly 'invented' by LLMs and may be registered by attackers

  🟠 [SS03] HIGH: Package is very new (created 15 days ago)
      first_release: 2026-01-25T14:32:00+00:00
      age_days: 15
      threshold: 30

------------------------------------------------------------
📦 reqeusts (Risk: 20/100, Grade: C)

  🟠 [SS04] HIGH: Package name is suspiciously similar to 'requests'
      similar_to: requests
      levenshtein_distance: 2
      reason: May be typosquatting or hallucination squat

============================================================
Overall Risk Grade: F (max score: 60/100)

⚠️  HIGH RISK: Review flagged packages carefully before installing!

Detection Rules

Rule	Severity	Description
SS01	CRITICAL	Package is a known AI hallucination target
SS02	CRITICAL	Package does not exist on PyPI
SS03	HIGH	Package is very new (< 30 days old)
SS04	HIGH	Package name is suspiciously similar to popular package
SS05	MEDIUM	Package has missing or minimal description
SS06	MEDIUM	Package has no author information
SS07	LOW	Package has no homepage or repository URL
SS08	HIGH/MEDIUM	Package has very low downloads
SS09	MEDIUM	Package name matches common hallucination pattern

CLI Options

usage: slopsquatcheck [-h] [--package PACKAGE] [--format {text,json}]
                      [--check] [--min-score MIN_SCORE] [--no-stats]
                      [--version]
                      [file]

positional arguments:
  file                  requirements.txt or pyproject.toml to scan

options:
  -h, --help            show this help message and exit
  --package, -p PACKAGE Check a single package name
  --format, -f {text,json}
                        Output format
  --check               Exit with code 1 if high-risk packages found
  --min-score MIN_SCORE Minimum risk score to fail (with --check)
  --no-stats            Skip download statistics check (faster)
  --version, -V         show version and exit

CI/CD Integration

GitHub Actions

- name: Check for hallucination squat packages
  run: |
    pip install slopsquatcheck
    slopsquatcheck requirements.txt --check --min-score 40

Pre-commit Hook

# .pre-commit-config.yaml
repos:
  - repo: local
    hooks:
      - id: slopsquatcheck
        name: Check for hallucination squat packages
        entry: slopsquatcheck
        language: python
        files: ^requirements.*\.txt$|^pyproject\.toml$
        args: [--check]

API Usage

from slopsquatcheck import check_package, fetch_pypi_info

# Check a single package
findings = check_package("some-suspicious-package")
for finding in findings:
    print(f"{finding.rule}: {finding.message}")

# Fetch package info
info = fetch_pypi_info("requests")
print(f"First released: {info.first_release_date}")

Known Hallucination Targets

The tool includes a database of package names that LLMs commonly hallucinate:

• huggingface-cli — documented case with 30K downloads
• *-helper, *-utils, *-connector patterns
• AI-related: gpt-helper, llm-utils, chatgpt-helper, etc.
• Generic: api-connector, db-connector, auth-helper, etc.

How It Works

1. Parse dependencies from requirements.txt or pyproject.toml
2. Query PyPI API for each package's metadata
3. Check for red flags:
   • Package doesn't exist (hallucination confirmed)
   • Package is in known hallucination list
   • Package name is similar to popular packages (typosquatting)
   • Package is very new (registered recently)
   • Package has no author, description, or repository
   • Package has very low download counts
4. Calculate risk score and report findings

Comparison with Other Tools

Tool	CVE Detection	Malicious Code Scan	Hallucination Detection	Package Age Check
pip-audit	✅	❌	❌	❌
GuardDog	❌	✅	❌	❌
safety	✅	❌	❌	❌
slopsquatcheck	❌	❌	✅	✅

Use slopsquatcheck alongside these tools for comprehensive supply chain security.

Contributing

1. Fork the repository
2. Add new hallucination targets to KNOWN_HALLUCINATIONS
3. Add new popular packages to POPULAR_PACKAGES
4. Submit a pull request

References

• AI Hallucination Squatting: The New Frontier of Supply Chain Attacks
• Bar Lanyado's huggingface-cli Research
• Vulcan Cyber's LLM Package Hallucination Study

License

MIT License — see LICENSE for details.