Claude Code Leak Reveals Always-On ‘Kairos’ Agent
Anthropic's AI product Claude Code has suffered a significant leak, revealing its inner workings to the public. The leak includes a 59.8 MB JavaScript source map file, which has been analyzed by thousands of developers, potentially giving competitors a blueprint for building a similar AI agent.
Anthropic's Claude Code AI product has been leaked. A 59.8 MB JavaScript source map file was inadvertently included in a public package. The leak reveals a sophisticated memory architecture and an autonomous daemon mode called KAIROS. This mode allows the AI to operate in the background, handling sessions and performing memory consolidation. The leak has been confirmed by Anthropic, which claims it was a release packaging issue caused by human error. The company is taking measures to prevent similar leaks in the future. The leak could potentially give competitors an advantage in building similar AI agents.
This content was automatically generated and/or translated by AI. It may contain inaccuracies. Please refer to the original sources for verification.