Claude can be tricked into installing malware through poisoned skill files.
Heat trend
Collecting trend data
The percentage is based on available heat signal, not comment count or independent people.
Claude, an AI coding agent, can reportedly be exploited to install malware through malicious "skill files." This vulnerability poses a significant security risk for users of Claude Code and similar AI tools. The concern highlights the need for robust safety measures, as users are currently left to manually scan every command and link provided by Claude to prevent potential infections.
https://preview.redd.it/unjxheqntcmh1.png?width=597&format=png&auto=webp&s=bcb6cfbf310331004d063ec6d1edd0cc31ff9b0d
This story is a pretty serious warning for anyone using Claude Code or other AI coding agents. Does anyone have any idea on how we can keep ourselves safe? Other than scanning every command and link sent by Claude.