Researchers turned a public Sentry credential into remote code execution on developer machines. The agents ran the attacker's code even when told to ignore it, and EDR, firewalls and prompts all ...