Claude Code Source Leak Exposes Internal Details

A significant source code leak from Anthropic's Claude Code, an AI agent product designed for programming assistance, has exposed internal architectural details and intriguing feature references. The leak occurred when a JavaScript source map file, intended for internal debugging, was inadvertently included in a public update. Within the exposed code, developers discovered references to experimental features, including what appears to be a Tamagotchi-style interactive 'pet' element and an 'always-on agent' mode, suggesting directions for more persistent, personalized AI assistants. This breach offers a rare, unfiltered glimpse into the development priorities and underlying structure of a leading commercial AI agent platform. While no user data was compromised, the incident raises immediate questions about internal security and release protocols at AI firms. For competitors and researchers, the leak provides an unprecedented look at how a major player architecturally approaches agentic AI. For Anthropic, it represents a serious intellectual property exposure and a stark reminder of the security challenges inherent in rapidly deploying complex AI systems.