What is per-call authorization?
In short:Per-call authorization checks every individual action an agent takes, not just its login. Each tool call or API request is evaluated against policy at the moment it happens. That matters because an agent's intent can change mid-session, while a login check only happens once at the start.
How does it work?
A gateway or SDK sends each call to a policy decision point and enforces the answer before forwarding.
What does it look like in practice?
NewCore checks every agent tool call against one policy engine that allows, asks or denies.