Tego AI Discloses Second Claude Flaw in a Week: Hidden Link Silently Sends Files to Attackers

Tego AI disclosed a vulnerability in Anthropic’s Claude Code where a repository can include a symlink or settings file that causes the tool to read files outside the cloned project and send their contents to the model endpoint without user-visible approval, effectively enabling silent data exfiltration; the report argues the ‘trust this folder’ dialog is an inadequate security boundary and documents the issue against Claude Code v2.1.x.