Anthropic releases open framework for AI-driven code vulnerability detection

Anthropic published a reference implementation for using Claude to identify security vulnerabilities in code. The framework is designed to be practical for teams integrating AI into their security workflows, offering a concrete starting point rather than abstract guidance for vulnerability discovery at scale.