PipGuard

A security tool to prevent accidental installation of malicious AI-generated packages.

Problem

AI coding assistants (ChatGPT, Claude, Copilot) sometimes invent nonexistent Python packages. Developers might unknowingly run pip install on these hallucinated names, installing malware or typosquat packages that can compromise development environments.

Solution

PipGuard intercepts package installations and provides real-time risk assessment:

• Typosquatting detection - Identifies packages mimicking popular libraries
• Malicious package database - Checks against known malicious packages from OSV, GitHub Advisory Database
• Suspicious pattern detection - Flags new packages with low download counts, obfuscated code
• Warn-by-default approach - Never blocks workflows, builds trust through informative warnings

Installation

pip install pipguard

Usage

# Install a package with safety check
pipguard install requests

# Scan requirements file
pipguard scan requirements.txt

# Audit current environment
pipguard audit .

Risk Assessment

PipGuard analyzes packages for:

• HIGH: Known malicious packages or clear typosquats
• MEDIUM: Suspicious patterns (new package, low downloads, suspicious metadata)
• LOW: Minor concerns worth noting

Example warning:

⚠️  Suspicious package detected: "reqeusts"
Reasons:
- Similar to popular package "requests"
- Published 2 days ago
- Low download count
Risk score: HIGH

Continue anyway? (y/N)

Data Sources

• OpenSSF Malicious Packages
• PyPA Advisory Database
• OSV.dev
• PyPI JSON API

License

MIT License - see LICENSE file for details.