Small World with High Risks: A Study of Security Threats in the npm Ecosystem
Not on the map. Why?
Cited by 23 in LLM agents and tool use, but placed by its own abstract, which didn’t fit.
Abstract (arXiv)
The popularity of JavaScript has lead to a large ecosystem of third-party packages available via the npm software package registry. The open nature of npm has boosted its growth, providing over 800,000 free and reusable software packages. Unfortunately, this open nature also causes security risks, as evidenced by recent incidents of single packages that broke or attacked software running on millions of computers. This paper studies security risks for users of npm by systematically analyzing dependencies between packages, the maintainers responsible for these packages, and publicly reported security issues. Studying the potential for running vulnerable or malicious code due to third-party dependencies, we find that individual packages could impact large parts of the entire ecosystem. Moreover, a very small number of maintainer accounts could be used to inject malicious code into the majority of all packages, a problem that has been increasing over time. Studying the potential for accidentally using vulnerable code, we find that lack of maintenance causes many packages to depend on vulnerable code, even years after a vulnerability has become public. Our results provide evidence that npm suffers from single points of failure and that unmaintained packages threaten large code bases. We discuss several mitigation techniques, such as trusted maintainers and total first-party security, and analyze their potential effectiveness.
Key facts. Cited by 36 papers in the map (data through 7 Oct 2026). First posted on arXiv 25 Feb 2019.
Builds on
No reference list yet.
Built on by
Citing papers per year: 2022: 2; 2023: 2; 2024: 3; 2025: 4; 2026: 25.
36 papers.
Sort:
By area:
- 2024We Have a Package for You! A Comprehensive Analysis of Package Hallucinations by Code Generating LLMscited by 61
- 2026"Do Not Mention This to the User": Detecting and Understanding Malicious Agent Skills in the Wildcited by 46
- 2023An Empirical Study of Pre-Trained Model Reuse in the Hugging Face Deep Learning Model Registrycited by 45
- 2026How Your Credentials Are Leaked by LLM Agent Skills: An Empirical Studycited by 14
- 2025A Measurement Study of Model Context Protocol Ecosystemcited by 13
- 2023An Empirical Study on Using Large Language Models to Analyze Software Supply Chain Security Failurescited by 11
- 2024What do we know about Hugging Face? A systematic literature review and quantitative validation of qualitative claimscited by 8
- 2026Behavioral Integrity Verification for AI Agent Skillscited by 6
- 2026Exploiting LLM Agent Supply Chains via Payload-less Skillscited by 5
- 2022Malicious Source Code Detection Using Transformercited by 4
- 2026Trust Me, Import This: Dependency Steering Attacks via Malicious Agent Skillscited by 4
- 2026ClawHub Security Signals: When VirusTotal, Static Analysis, and SkillSpector Disagreecited by 4
24 more
- 2025A First Look at the Security Issues in the Model Context Protocol Ecosystemcited by 3
- 2026Multi-Agent Taint Specification Extraction for Vulnerability Detectioncited by 3
- 2026You Told Me to Do It: Measuring Instructional Text-induced Private Data Leakage in LLM Agentscited by 3
- 2022Machine Learning Systems are Bloated and Vulnerablecited by 2
- 2025Automatically Generating Rules of Malicious Software Packages via Large Language Modelcited by 2
- 2026ADK Arena: Evaluating Agent Development Kits via LLM-as-a-Developercited by 2
- 2026SkillGate: Cost Efficient Runtime Malicious Skill File Detection in Coding Agentscited by 2
- 2024Protect Your Secrets: Understanding and Measuring Data Exposure in VSCode Extensionscited by 1
- 2026SEVRA-BENCH: Social Engineering of Vulnerabilities in Review Agentscited by 1
- 2026From Registry to Repository: How AI Agent Skills Are Written, Adapted, and Maintainedcited by 1
- 2026Setup Complete, Now You Are Compromised: Weaponizing Setup Instructions Against AI Coding Agentscited by 1
- 2026Exposed by Design: A Dynamic Security Assessment of Internet-Facing MCP Servers at Scalecited by 1
- 2025AgentHub: A Registry for Discoverable, Verifiable, and Reproducible AI Agentscited by 0
- 2026CHASE: LLM Agents for Dissecting Malicious PyPI Packagescited by 0
- 2026CCCE: A Continuous Code Calibration Engine for Autonomous Enterprise Codebase Maintenance via Knowledge Graph Traversal and Adaptive Decision Gatingcited by 0
- 2026The Impact of Configuring Agentic AI Coding Tools on Build-vs-Buy Decisions: A Study Protocolcited by 0
- 2026Chai: Agentic Discovery of Cryptographic Misuse Vulnerabilitiescited by 0
- 2026Govern the Repository, Not the Agent: Measuring Ecosystem-Level Risk in AI-Native Softwarecited by 0
- 2026LLM-Enhanced Hierarchical Heterogeneous Graph Representation Learning for Malicious Python Package Detectioncited by 0
- 2026From Adoption to Deployment: A Qualitative Study on AI Integration in Software Development Practicecited by 0
- 2026Static analysis-guided agentic AI translation enables Rust as a full stack bioinformatics languagecited by 0
- 2026TrustShiftProbe: Characterizing, Benchmarking, and Defending Staged Trust Attacks on MCP Serverscited by 0
- 2026Can Agents Trust Their Skills? Uncovering Unsafe Chains of Trust in Skill-Based LLM Agentscited by 0
- 2026Code That Works, Environments That Don't: Measuring Environment Reproducibility in AI-Generated Softwarecited by 0
- 2026Code That Works, Environments That Don't: Measuring Environment Reproducibility in AI-Generated Software30 Sep 2026
- 2026Can Agents Trust Their Skills? Uncovering Unsafe Chains of Trust in Skill-Based LLM Agents30 Sep 2026
- 2026TrustShiftProbe: Characterizing, Benchmarking, and Defending Staged Trust Attacks on MCP Servers24 Aug 2026
- 2026Static analysis-guided agentic AI translation enables Rust as a full stack bioinformatics language13 Aug 2026
- 2026Exposed by Design: A Dynamic Security Assessment of Internet-Facing MCP Servers at Scale31 Jul 2026
- 2026SkillGate: Cost Efficient Runtime Malicious Skill File Detection in Coding Agents28 Jul 2026
- 2026From Adoption to Deployment: A Qualitative Study on AI Integration in Software Development Practice18 Jul 2026
- 2026Setup Complete, Now You Are Compromised: Weaponizing Setup Instructions Against AI Coding Agents16 Jul 2026
- 2026LLM-Enhanced Hierarchical Heterogeneous Graph Representation Learning for Malicious Python Package Detection3 Jul 2026
- 2026From Registry to Repository: How AI Agent Skills Are Written, Adapted, and Maintained1 Jul 2026
- 2026Govern the Repository, Not the Agent: Measuring Ecosystem-Level Risk in AI-Native Software26 Jun 2026
- 2026Chai: Agentic Discovery of Cryptographic Misuse Vulnerabilities25 Jun 2026
24 more
- 2026SEVRA-BENCH: Social Engineering of Vulnerabilities in Review Agents11 Jun 2026
- 2026ADK Arena: Evaluating Agent Development Kits via LLM-as-a-Developer4 Jun 2026
- 2026The Impact of Configuring Agentic AI Coding Tools on Build-vs-Buy Decisions: A Study Protocol2 Jun 2026
- 2026ClawHub Security Signals: When VirusTotal, Static Analysis, and SkillSpector Disagree31 May 2026
- 2026Exploiting LLM Agent Supply Chains via Payload-less Skills14 May 2026
- 2026Behavioral Integrity Verification for AI Agent Skills12 May 2026
- 2026Trust Me, Import This: Dependency Steering Attacks via Malicious Agent Skills10 May 2026
- 2026CCCE: A Continuous Code Calibration Engine for Autonomous Enterprise Codebase Maintenance via Knowledge Graph Traversal and Adaptive Decision Gating10 Apr 2026
- 2026How Your Credentials Are Leaked by LLM Agent Skills: An Empirical Study3 Apr 2026
- 2026You Told Me to Do It: Measuring Instructional Text-induced Private Data Leakage in LLM Agents12 Mar 2026
- 2026"Do Not Mention This to the User": Detecting and Understanding Malicious Agent Skills in the Wild6 Feb 2026
- 2026Multi-Agent Taint Specification Extraction for Vulnerability Detection15 Jan 2026
- 2026CHASE: LLM Agents for Dissecting Malicious PyPI Packages11 Jan 2026
- 2025A First Look at the Security Issues in the Model Context Protocol Ecosystem18 Oct 2025
- 2025AgentHub: A Registry for Discoverable, Verifiable, and Reproducible AI Agents3 Oct 2025
- 2025A Measurement Study of Model Context Protocol Ecosystem29 Sep 2025
- 2025Automatically Generating Rules of Malicious Software Packages via Large Language Model24 Apr 2025
- 2024Protect Your Secrets: Understanding and Measuring Data Exposure in VSCode Extensions1 Dec 2024
- 2024We Have a Package for You! A Comprehensive Analysis of Package Hallucinations by Code Generating LLMs12 Jun 2024
- 2024What do we know about Hugging Face? A systematic literature review and quantitative validation of qualitative claims12 Jun 2024
- 2023An Empirical Study on Using Large Language Models to Analyze Software Supply Chain Security Failures9 Aug 2023
- 2023An Empirical Study of Pre-Trained Model Reuse in the Hugging Face Deep Learning Model Registry5 Mar 2023
- 2022Machine Learning Systems are Bloated and Vulnerable16 Dec 2022
- 2022Malicious Source Code Detection Using Transformer16 Sep 2022
- 2026"Do Not Mention This to the User": Detecting and Understanding Malicious Agent Skills in the Wildcited by 46
- 2026How Your Credentials Are Leaked by LLM Agent Skills: An Empirical Studycited by 14
- 2025A Measurement Study of Model Context Protocol Ecosystemcited by 13
- 2026Behavioral Integrity Verification for AI Agent Skillscited by 6
- 2026Exploiting LLM Agent Supply Chains via Payload-less Skillscited by 5
- 2026Trust Me, Import This: Dependency Steering Attacks via Malicious Agent Skillscited by 4
- 2025A First Look at the Security Issues in the Model Context Protocol Ecosystemcited by 3
- 2026Multi-Agent Taint Specification Extraction for Vulnerability Detectioncited by 3
- 2026You Told Me to Do It: Measuring Instructional Text-induced Private Data Leakage in LLM Agentscited by 3
- 2026ADK Arena: Evaluating Agent Development Kits via LLM-as-a-Developercited by 2
- 2026SkillGate: Cost Efficient Runtime Malicious Skill File Detection in Coding Agentscited by 2
- 2026SEVRA-BENCH: Social Engineering of Vulnerabilities in Review Agentscited by 1
12 most cited of 23 All 23 (JSON)
- 2026Code That Works, Environments That Don't: Measuring Environment Reproducibility in AI-Generated Software30 Sep 2026
- 2026Can Agents Trust Their Skills? Uncovering Unsafe Chains of Trust in Skill-Based LLM Agents30 Sep 2026
- 2026TrustShiftProbe: Characterizing, Benchmarking, and Defending Staged Trust Attacks on MCP Servers24 Aug 2026
- 2026Static analysis-guided agentic AI translation enables Rust as a full stack bioinformatics language13 Aug 2026
- 2026SkillGate: Cost Efficient Runtime Malicious Skill File Detection in Coding Agents28 Jul 2026
- 2026Setup Complete, Now You Are Compromised: Weaponizing Setup Instructions Against AI Coding Agents16 Jul 2026
- 2026From Registry to Repository: How AI Agent Skills Are Written, Adapted, and Maintained1 Jul 2026
- 2026Govern the Repository, Not the Agent: Measuring Ecosystem-Level Risk in AI-Native Software26 Jun 2026
- 2026Chai: Agentic Discovery of Cryptographic Misuse Vulnerabilities25 Jun 2026
- 2026SEVRA-BENCH: Social Engineering of Vulnerabilities in Review Agents11 Jun 2026
- 2026ADK Arena: Evaluating Agent Development Kits via LLM-as-a-Developer4 Jun 2026
- 2026The Impact of Configuring Agentic AI Coding Tools on Build-vs-Buy Decisions: A Study Protocol2 Jun 2026
12 newest of 23 All 23 (JSON)
- 2024We Have a Package for You! A Comprehensive Analysis of Package Hallucinations by Code Generating LLMscited by 61
- 2023An Empirical Study of Pre-Trained Model Reuse in the Hugging Face Deep Learning Model Registrycited by 45
- 2023An Empirical Study on Using Large Language Models to Analyze Software Supply Chain Security Failurescited by 11
- 2024What do we know about Hugging Face? A systematic literature review and quantitative validation of qualitative claimscited by 8
- 2022Malicious Source Code Detection Using Transformercited by 4
- 2026ClawHub Security Signals: When VirusTotal, Static Analysis, and SkillSpector Disagreecited by 4
- 2022Machine Learning Systems are Bloated and Vulnerablecited by 2
- 2025Automatically Generating Rules of Malicious Software Packages via Large Language Modelcited by 2
- 2024Protect Your Secrets: Understanding and Measuring Data Exposure in VSCode Extensionscited by 1
- 2026Exposed by Design: A Dynamic Security Assessment of Internet-Facing MCP Servers at Scalecited by 1
- 2026CCCE: A Continuous Code Calibration Engine for Autonomous Enterprise Codebase Maintenance via Knowledge Graph Traversal and Adaptive Decision Gatingcited by 0
- 2026LLM-Enhanced Hierarchical Heterogeneous Graph Representation Learning for Malicious Python Package Detectioncited by 0
12 most cited of 13 All 13 (JSON)
- 2026Exposed by Design: A Dynamic Security Assessment of Internet-Facing MCP Servers at Scale31 Jul 2026
- 2026From Adoption to Deployment: A Qualitative Study on AI Integration in Software Development Practice18 Jul 2026
- 2026LLM-Enhanced Hierarchical Heterogeneous Graph Representation Learning for Malicious Python Package Detection3 Jul 2026
- 2026ClawHub Security Signals: When VirusTotal, Static Analysis, and SkillSpector Disagree31 May 2026
- 2026CCCE: A Continuous Code Calibration Engine for Autonomous Enterprise Codebase Maintenance via Knowledge Graph Traversal and Adaptive Decision Gating10 Apr 2026
- 2025Automatically Generating Rules of Malicious Software Packages via Large Language Model24 Apr 2025
- 2024Protect Your Secrets: Understanding and Measuring Data Exposure in VSCode Extensions1 Dec 2024
- 2024We Have a Package for You! A Comprehensive Analysis of Package Hallucinations by Code Generating LLMs12 Jun 2024
- 2024What do we know about Hugging Face? A systematic literature review and quantitative validation of qualitative claims12 Jun 2024
- 2023An Empirical Study on Using Large Language Models to Analyze Software Supply Chain Security Failures9 Aug 2023
- 2023An Empirical Study of Pre-Trained Model Reuse in the Hugging Face Deep Learning Model Registry5 Mar 2023
- 2022Machine Learning Systems are Bloated and Vulnerable16 Dec 2022
12 newest of 13 All 13 (JSON)