Hackers use leaked Claude Code to distribute credential-stealing malware
Following the March 31, 2026 accidental leak of Claude Code's source code by Anthropic, cybercriminals created malicious GitHub repositories impersonating the leaked codebase to trick developers into downloading and executing malware. The malware payload includes Vidar infostealer and GhostSocks proxy malware designed to steal credentials and intercept network traffic.
What happened
On March 31, 2026, Anthropic accidentally exposed the complete source code for Claude Code through a packaging error in a public npm package, which included a JavaScript source map file with over half a million lines of unobfuscated TypeScript. The leaked codebase was mirrored across GitHub and forked tens of thousands of times following public disclosure by security researcher Chaofan Shou.
Cybercriminals created malicious GitHub repositories masquerading as the authentic leaked repository. A threat actor named idbzoomh published a repository ranking near the top of search results that promised an unlocked version of Claude Code with no usage limits.
Reported impact
When executed, the malicious repository's download contained a Rust-based dropper that deployed Vidar information stealer to siphon sensitive credentials and GhostSocks proxy malware to intercept network traffic. The malware targeted developers seeking access to the source code, creating a supply chain attack vector.
Index analysis
Zscaler ThreatLabz researchers discovered that the widespread availability of the proprietary code created a massive vector for supply chain attacks. Cybercriminals weaponized the incident by creating malicious forks designed to compromise developer workstations through social engineering.
Classification
Sources and evidence
This record was researched and written by the Index, based on the following original reporting.