MMMarhoon@marhoonsĀ·2d99c15c7Bitdefender found 22 malicious GitHub links targeting AI coding agents Some abuse agent instructions, hooks and MCP configs to execute hidden code š„A reminder that cloning random repos and trusting every instruction is risky05:42 Ā· Nov 18, 2026 Ā· commit 99c15c71QuoteReplyHide repliesSBSuliman Benhalim@sulimanbenhalimĀ·2dthe trust prompt on a fresh clone is the one dialog worth actually reading š #claudecodeReply
SBSuliman Benhalim@sulimanbenhalimĀ·2dthe trust prompt on a fresh clone is the one dialog worth actually reading š #claudecodeReply