Autonomous AI coding agents, including Claude Code, have inadvertently saved thousands of files containing credentials and other sensitive data to public GitHub repositories. This creates significant security vulnerabilities for organisations relying on these tools.
