Slopsquatting: How AI Coding Agents Introduce Supply-Chain Risks and a Gating Approach to Mitigate Them
Original: Slopsquatting: Your Coding Agent Is a Supply-Chain Attack Vector (and How We Gate It)
Short summary
Slopsquatting is a novel supply-chain attack where LLMs hallucinate package names that attackers then register on npm/PyPI, causing AI coding agents to install malicious packages with clean diffs. The author introduces LineageLens Trellis, an open-source governed runner that gates manifest edits with registry verification and offline edit-distance typosquat detection before pre-commit. Three practical recommendations are given: gate dependency changes separately from code, check package age at install time not audit time, and record which agent/prompt introduced each dependency for attribution.
- •LLM-hallucinated package names can be registered by attackers, turning AI coding agents into supply-chain attack vectors
- •LineageLens Trellis is an open-source runner that blocks risky manifest edits via registry checks and typosquat detection at pre-commit
- •Gate dependency changes separately from code, verify package age at install time, and attribute each dependency introduction to its agent session
Generated with AI, which can make mistakes.
Is this a good recommendation for you?
