Obelsk watches for people using AI tools, agents acting on their own, and outside agents probing in, then scores each one by how much it threatens you.
Use your Obelsk console account.
Five failed attempts lock sign-in for 15 minutes. All sign-ins are audited.