A security flaw dubbed Plugin4Shell impacts four widely used AI coding agents, allowing repository owners to swap pinned plugin versions for malicious code despite strict version locking. Vendors including Anthropic and OpenAI have patched the issue in Claude Code and Codex respectively. Security teams and developers must update their AI development tools immediately to mitigate potential software supply-chain compromises.
READ THE ORIGINAL ARTICLE AT THEHACKERNEWS.COM →