Agent Responsibility Engineering
The discipline of controlling autonomous systems at the moment of action, where authority must be proven before execution.
Glossary
A precise vocabulary for governed autonomous systems.
ARE uses terms from control, governance, evidence, and production operations. The vocabulary is intentionally narrow because governed systems need exact language.
Terms
Definitions for the execution boundary.
These definitions are public-facing and conceptual. They describe the control model without exposing implementation detail.
Authority
The governed right to perform a specific action under a specific identity, scope, delegation, and context.
Capability
What an agent can do. Capability is not authority and does not create permission to act.
Execution Boundary
The control point where intent, plan, or output becomes action in the world or in an operational system.
Passport
A verifiable authority token binding identity, scope, and delegated right to act.
Spawn Chain
The authority chain created when agents create, delegate to, or activate other agents.
Deny-Path
The governed ability to refuse action, not merely observe or report it after execution.
Evidence Bundle
The verifiable record attached to a decision or action, including the basis for permit, modification, or denial.
Operations Ledger
Immutable operational memory for decisions, actions, outcomes, and feedback.
Governance Coprocessor
An interposed control function between agent intent and execution.
SDAVL Loop
Signals to Decisions to Actions to Verification to Learning, closed as a governing cycle.
Legibility
The property that an agent, its authority, and its action path can be inspected and explained.