Zero Trust lab in three layers, all as code: Microsoft Entra ID identity and Tailscale networking (ACLs, SSH, high-availability subnet routing) in Terraform, plus TypeScript AI agents (Scarlet coordinator, JML, GTM) with per-agent identity, human approval and full traces. Threat modeled against the OWASP LLM and Agentic Top 10. Mutation-tested CI.
zerotrust agentic-workflow agentic-ai agent-orchestration zerotrustapplication zerotrustcybersecurity agentidentityprotocol
-
Updated
Oct 3, 2026 - TypeScript