Anthropic's DMCA blitz nukes 8,100 repos — then backtracks

Anthropic accidentally exposed Claude Code's full source (~513,000 lines of TypeScript) via an unobfuscated source map bundled in npm package @anthropic-ai/claude-code v2.1.88. The leak spawned one of GitHub's fastest-growing fork networks overnight.

Anthropics's response: mass DMCA takedowns targeting the entire fork tree — over 8,100 repositories disabled. The sweep overreached, pulling down heavily modified forks and even forks of Anthropic's own public Claude Code repo that had nothing to do with the leaked source.

After public backlash, Anthropic's head of Claude Code called it an accident and retracted the bulk of the notices. Only 96 repositories remain suspended; the rest were restored. The leaked source itself is still floating around, and Anthropic is reportedly continuing to search GitHub for infringing copies.

Why it matters: A proprietary AI tool's full internals are now public knowledge — and the ham-fisted takedown attempt only amplified the signal.