Claude Code Source Leak: A Timeline

A debugging file accidentally exposed 512,000 lines of Anthropic's Claude Code source code when a routine software update included a source map pointing to an unprotected zip archive anyone could download. The March 31st leak, caused by a known bug in the Bun JavaScript runtime, revealed fascinating details about the AI coding assistant's inner workings: 44 unreleased features including autonomous background agents, a controversial "Undercover Mode" that strips AI attribution from public code contributions, and 187 whimsical loading animation verbs like "hullaballooing" and "razzmatazzing."

The leak sparked intense debate across social media, with over 32 million views on X and thousands of GitHub repositories mirroring the code before Anthropic's DMCA takedowns. Developer Sigrid Jin used OpenAI's Codex to rewrite the entire codebase in Python within hours, creating a project that garnered 105,000 GitHub stars and highlighting what legal experts called an "AI Copyright Paradox" since much of Claude Code was written by AI itself. This marked Anthropic's second data exposure in a week, potentially complicating the company's reported $350 billion IPO plans while demonstrating that in AI coding tools, the orchestration layer matters more than the underlying model.