Claude Code leak exposes a Tamagotchi-style ‘pet’ and an always-on agent
A 512,000-line source code leak exposed unreleased features including an AI pet and background agent capabilities.
Anthropic's Claude Code, the AI-powered coding assistant launched in February 2025, experienced a significant leak when its 2.1.88 update inadvertently included a source map file containing the entire TypeScript codebase. The leak exposed more than 512,000 lines of internal code, which users quickly copied to GitHub where it garnered over 50,000 forks before Anthropic could contain the situation. According to company spokesperson Christopher Nulty, this was a "release packaging issue caused by human error" rather than a security breach, with no customer data or credentials exposed.
Developers digging through the leaked code discovered several unreleased features that provide insight into Anthropic's roadmap. Most notably, users found references to a Tamagotchi-style virtual pet designed to "sit beside your input box and react to your coding," and a feature called "KAIROS" that could enable always-on background agent capabilities. The code also revealed internal instructions for the AI bot, details about its memory architecture, and even candid developer comments questioning implementation decisions. AI analyst Arun Chandrasekaran noted that while the leak poses risks of bad actors bypassing guardrails, it primarily serves as a "call for action for Anthropic to invest more in processes and tools for better operational maturity."
- 512,000+ lines of Claude Code's TypeScript source code leaked in update 2.1.88
- Revealed Tamagotchi-style AI pet and 'KAIROS' always-on agent capabilities
- Code was forked 50,000+ times on GitHub before Anthropic contained the situation
Why It Matters
Exposes competitive AI features in development and highlights security risks when internal code becomes public.