Anthropic Accidentally Leaks 500,000 Lines of Claude Code Source Code via npm Registry
Anthropic accidentally published Claude Code's entire source code (512,000 lines, 1,906 files) to the public npm registry, exposing its three-layer memory architecture.
Anthropic has suffered a major incident after accidentally publishing the entire source code of its popular AI coding tool Claude Code to the public npm registry. The leak comprised 512,000 lines of TypeScript across 1,906 files. According to The Register, the leaked code revealed Claude Code's internal architecture in detail, most notably a three-layer memory system consisting of memory.md files, grep-based search functionality, and an unshipped 'Chyros' daemon. Anthropic described the incident as 'a human error, not a hack' and attempted to remove thousands of GitHub repositories containing the leaked code. However, this large-scale takedown effort itself drew criticism. Claude Code is a $2.5 billion coding tool, and concerns have been raised about competitive intelligence exposure and security vulnerabilities. Some developers have speculated whether the leak was actually an intentional marketing strategy.
Tools Mentioned in This Article
AI Newsletter
Get the latest AI tools and news delivered daily
Related Articles
Mustafa Suleyman: AI Development Won't Hit a Wall — 1,000x Compute by End of 2028
Microsoft AI CEO Mustafa Suleyman writes in MIT Technology Review that AI training data has grown 1 trillion times since 2010, predicting another 1,000x in effective compute by end of 2028.
OpenAI, Anthropic, and Google Unite to Combat Unauthorized Chinese AI Model Copying
Three major US AI companies collaborate through the Frontier Model Forum to combat Chinese 'adversarial distillation,' sharing intelligence to prevent billions in annual losses.
Uber Adopts Amazon's Custom AI Chips (Trainium & Graviton) to Move Beyond GPU Dependency
Uber adopts AWS Graviton4 and Trainium3 custom chips for driver matching optimization and AI model training, accelerating the shift from standard GPUs to specialized hardware.