Claude Code leak exposes a Tamagotchi-style ‘pet’ and an always-on agent

After Anthropic released Claude Code update 2.1.88, users quickly discovered that it contained a bundle with a source map file containing its TypeScript codebase, with one of X calling attention to the leak and posting a file containing the code. The leaked data reportedly contains over 512,000 lines of code and provides insight into the inner workings of the AI-based coding tool, as previously reported by Ars Technica And VentureBeat.

Although Anthropic later fixed the issue, that didn’t stop users from copying the code to a repository on GitHub, which has since accumulated over 50,000 forks (or copies of the repository). Anthropic launched Claude Code in February 2025, and the tool gained momentum after adding agent features that perform tasks on a user’s behalf.

“Earlier today, a release of Claude Code included internal source code. No sensitive customer data or identifiers were involved or exposed,” Anthropic spokesperson Christopher Nulty said in an emailed statement to The edge. “This is a release packaging issue caused by human error, not a security vulnerability. We are deploying measures to prevent this from happening again.”

Arun Chandrasekaran, AI analyst at Gartner, says The edge that while the Code Claude leak presents “risks such as providing possible avenues for bad actors to circumvent guardrails,” its long-term impact may be limited to serving as “a call to action for Anthropic to invest more in processes and tools for greater operational maturity.”