How Worms Exposed the Real Attack Surface of Open Source
summary
Worms propagate through speed, automation and transitive trust: postinstall life cycle scripts and npm auto-updates run code within minutes, GitHub Actions pone request flaws let pull-request code execute in repo context, and credentials become a replication mechanism. Campaigns use tools like TruffleHog and Trivy, republish mutable Docker/npm tags, and push bun runtimes; one wave hits ~180 packages, debug/chalk collect ~2.6 million downloads, 10% of monitored clouds see infected versions in two hours, 153 GB of leaked credentials from LightLLM, and 3,800 GitHub repos get exfiltrated via an open-sourced worm. exfiltration mutilization
AI-written summary · about
chapters (19)
- 0:00Introduction and supply chain threats
- 3:22Team PCP and their impact
- 7:58Real-world supply chain attacks
- 13:15Debug and chalk compromise
- 16:28Shy Hulude worm emerges
- 20:19Larger attacks on vendors
- 24:18Other major attacks of 2023
- 27:46Trivy compromise and lessons
- 30:53Iran-focused Kubernetes worm
- 34:05Axios compromise by North Korea
- 36:16Checkmarx and dependency risks
- 38:03Light LLM and credential leaks
- 39:27Tanstack compromise explained
- 41:53NX Visual Studio Code breach
- 45:13Automatic updates and cooldowns
- 47:23Trusted publishing adoption
- 50:13Team PCP's mysterious motives
- 53:08Arrests and aftermath
- 56:19Trust in open source today
