Policy and Access Decisions

Policy evaluation is the process of checking an access request against a defined set of rules to determine whether the request should be allowed or denied.

Intent evaluation is the optional process of analyzing the context, goal, or semantic meaning of an agent's tool call request to inform the access decision, beyond just checking the tool name and parameters.

Intent-aware authorization is an access control approach that considers the semantic intent of an agent's action—what it is trying to accomplish—as part of the access decision, not just the technical attributes of the request.

Policy and intent-driven access is the combination of formal access policy and semantic intent evaluation to produce access decisions for AI agent tool calls—considering both what the agent is permitted to do and what it is trying to accomplish.

Default deny is an access control principle where all actions are denied unless explicitly permitted by policy—as opposed to default allow, where actions proceed unless explicitly blocked.

Human-in-the-loop approval is a policy outcome where a tool call that exceeds automated policy thresholds is paused and routed to a human reviewer before execution proceeds.

Step-up authorization is an access control mechanism that requires additional verification or approval before a high-risk or sensitive action is permitted, even when a base level of access has already been granted.