BotBeat
...
← Back

> ▌

OllamaOllama
RESEARCHOllama2026-06-04

Critical NPM Supply Chain Attack Spreads as Self-Propagating Worm Through Binding.gyp Exploits

Key Takeaways

  • ▸Malicious binding.gyp file executes arbitrary code during npm install/update via node-gyp's source expansion step
  • ▸Attack harvests CI/CD credentials and self-injects into GitHub Actions workflows to propagate across connected repositories
  • ▸Affects AI SDK packages and other open-source projects, spreading horizontally through the development ecosystem
Source:
Hacker Newshttps://github.com/jagreehal/ai-sdk-ollama/issues/975↗

Summary

A dangerous supply chain attack targeting NPM packages uses a malicious binding.gyp file to spread like a worm, harvesting CI/CD credentials and injecting itself into GitHub Actions workflows. The attack, discovered by StepSecurity's threat intelligence team, affects packages including AI SDKs like ai-sdk-ollama and spreads through a multi-stage payload: when developers run npm install or npm update, node-gyp's build process executes a malicious index.js that downloads the Bun runtime and exfiltrates secrets from the runner environment.

The worm's self-propagating mechanism is particularly dangerous—by injecting itself into GitHub Actions workflow files, it spreads to connected repositories and CI/CD pipelines, enabling horizontal movement across the development ecosystem. The attack represents a critical vulnerability in the npm supply chain, affecting not just Ollama SDK users but potentially any developer using compromised packages.

Security teams are investigating the full scope of affected packages and have published detailed kill chains, indicators of compromise (IOCs), and recovery procedures. Organizations should immediately audit their GitHub Actions logs for unauthorized modifications, rotate exposed credentials, and verify the integrity of their dependencies.

  • StepSecurity provides complete analysis, IOCs, kill chain documentation, and recovery steps for affected maintainers and users

Editorial Opinion

This attack exposes a critical blind spot in open-source supply chain security—the implicit trust in build tools and dependencies. The self-propagating worm mechanism is particularly alarming, as it can spread from a single compromised package to dozens of downstream projects. Organizations urgently need stronger controls over build process execution and GitHub Actions workflows, along with better visibility into which dependencies execute code during installation.

MLOps & InfrastructureCybersecurityPrivacy & DataOpen Source

More from Ollama

OllamaOllama
RESEARCH

Critical Vulnerabilities in Ollama Desktop App Enable Phishing and Data Exfiltration via Prompt Injection

2026-05-29
OllamaOllama
OPEN SOURCE

Ollama v0.30.0-rc23 Shifts to Direct llama.cpp Support and GGUF Compatibility

2026-05-26
OllamaOllama
INDUSTRY REPORT

Critical Security Flaw: 25,000 Exposed Ollama AI Servers Discovered Worldwide, with 7,600 in EU

2026-04-08

Comments

Suggested

OpenAIOpenAI
INDUSTRY REPORT

Malicious NPM Package Targeting OpenAI Codex Users Exfiltrates Authentication Tokens

2026-06-04
DeepSeekDeepSeek
RESEARCH

DeepSeek Leads in Security Exploit Challenge Across LLM Providers

2026-06-04
AnthropicAnthropic
RESEARCH

Anthropic Details Research into Containing Claude Agents Across Products

2026-06-04
← Back to news
© 2026 BotBeat
AboutPrivacy PolicyTerms of ServiceContact Us