Anthropic accidentally made the whole source code for Claude Code, its main terminal-based coding assistant, public. The leak of Claude Code on GitHub has become a major way for attackers to get into supply chains. Zscaler ThreatLabz researchers recently found a fake campaign that uses the leaked code as a social engineering lure to get developers to give it to them.

These campaigns are spreading malicious GitHub repositories that use the leaked Claude Code source. They pretend to be real repositories and trick people into compromising their workstations. Threat actors could take over devices without anyone knowing or steal credentials by getting developers to clone a repository that isn't trusted. To protect development environments from these opportunistic attacks, immediate defensive measures are necessary.

Security teams should strongly warn all developers not to download, build, or run any code that claims to be the leaked Anthropic software. To keep things honest, only use official channels and signed binaries. Set up a Zero Trust architecture and limit access to important apps to keep damage to a minimum if a developer's workstation is hacked.

Check for strange outbound network connections and look for unexpected npm packages in local environments to find early signs of infection. LinkedIn and X send out daily cybersecurity updates, and you can contact us if you want to see featured stories. Follow us on Twitter at @LinkedInX and @Xlinkedin, and we'll put you in our weekly Newsquiz for more news about cybersecurity.

We'd love to hear what you think about our coverage on Facebook and Twitter @linkedin.