claude-code-hooks

No clear malware/backdoor behavior is evident in the shown hook recipes. However, the design introduces substantial security/privacy risk: it can persistently log tool commands and file changes to disk, propagate local .env secrets into another env file, and exfiltrate untrusted permission/notification messages to an external Slack webhook via curl. Additionally, because it executes shell commands based on event payload fields, robust allowlisting/escaping is not consistently enforced, leaving room for injection or bypass if upstream inputs are attacker-controlled.