Appearance
Glossary
| Term | Meaning |
|---|---|
| Agent Policy | Portable decision and trace facts for policy around agent actions. |
| Policy decision | Result of evaluating a proposed action. |
| Subject | Actor requesting an action: user, agent, service, subagent, or session. |
| Action | Proposed operation: tool call, model call, export, memory write, delegation, and similar. |
| Resource | Target of an action: file, URL, artifact, source, model, tool, credential, or peer. |
| Context | Facts used to evaluate policy: mode, trust boundary, sensitivity, time, cost, tenant, and refs. |
| Risk scope | Typed boundary affected by the action. |
| Obligation | Enforceable duty attached to a decision. |
| Advice | Non-binding guidance attached to a decision. |
| Approval request | Structured request for approval or escalation. |
| Permission grant | Bounded capability issued after policy evaluation or approval. |
| Waiver | Explicit bounded override of a policy block. |
| Policy trace | Correlation graph for policy inputs, rules, decisions, approvals, grants, waivers, runtime, telemetry, and evidence. |