Glossary

Core terms

• Intelligence Control Plane: the decision and governance layer above execution infrastructure.
• Eligible path: a model-provider pair that satisfies current hard constraints.
• Objective: the optimization posture used after eligibility.
• Route evidence: attribution for the selected and attempted paths.
• Evaluation: a structured judgment of output requirements.
• Agent identity: signed claims describing the acting principal and authority.
• Guard decision: allow, deny, or require approval before a tool action.
• Outcome Ledger: separated potential, modeled, committed, and measured impact evidence.