CVE-2025-59828: high-severity vulnerability in anthropics claude-code
Claude Code Vulnerable to Arbitrary Code Execution via Plugin Autoloading with Specific Yarn Versions
Published
E-mail alert when it lands in KEV, gets a public PoC or its score goes up.
No sign of exploitation. No public exploitation artifact known so far.
Claude Code automatically executed Yarn plugins before asking users to trust a directory, allowing attackers to run malicious code without permission. This vulnerability only affected Yarn 2.0+ users and has been fixed in version 1.0.39.
The vulnerability exploits Yarn 2.0+ plugin autoloading mechanism triggered by yarn --version command execution. Claude Code invoked this command before presenting the directory trust dialog to users, allowing pre-execution of arbitrary plugins stored in untrusted directories. An attacker could craft a malicious Yarn configuration to execute code with the user's privileges before consent is obtained.
In the same product, most dangerous first.