Cynative blog https://cynative.com/blog/ Notes from the Cynative team on security agents, trust boundaries and sovereign models. en Tue, 08 Sep 2026 00:00:00 GMT Securing AI Agents on AWS: One HTTP Tool, One Action Gateway https://cynative.com/blog/aws-agent-action-gateway/ https://cynative.com/blog/aws-agent-action-gateway/ Tue, 08 Sep 2026 00:00:00 GMT contact@cynative.com (Yuri Shapira) Why Cynative gives security agents one HTTP tool, uses AWS-maintained models to follow API changes, and checks IAM actions before signing requests. Building Security Agents That Cannot Escape Their Trust Boundary https://cynative.com/blog/agent-trust-boundaries/ https://cynative.com/blog/agent-trust-boundaries/ Wed, 12 Aug 2026 00:00:00 GMT contact@cynative.com (Shaked Zin) Live access to production infrastructure, read-only by construction, with no path for the agent to widen it at runtime.