Yesterday afternoon, a significant source code leak took place, involving the renowned AI programming tool, Claude Code. Public information reveals that the leak was not the result of an operational mistake by Anthropic's engineers. Instead, it stemmed from a flaw in the build tool, which inadvertently incorporated Source Map files containing the entire original code during the packaging process for the production environment. These files encompassed 512,000 lines of unobfuscated TypeScript code, more than 40 tool modules, and several unreleased features, thereby laying bare the internal architecture logic and system design of Claude Code. Following the incident, Anthropic promptly removed the relevant files. However, by that time, the code had already proliferated widely within the developer community. Although the leak did not compromise user data or core model weights, it did bring to light security vulnerabilities in the software supply chain and could potentially lower the barriers to technical research and development in the AI Agent field.