Execution Admissibility Architecture — Canonical Definitions

About these definitions

This page defines the authoritative vocabulary for Execution Admissibility Architecture.

It covers the concepts through which proposed actions are evaluated, governed, evidenced and permitted or prevented from binding institutional consequence.

These definitions are domain-specific. They inherit from the broader Arqua Canonical Architecture and the Enterprise Intelligence Architecture canonical vocabulary.

They do not define the whole Arqua architecture.

Enterprise Intelligence Architecture defines the complete institutional intelligence system. Execution Admissibility Architecture governs the boundary at which understanding, authority and intent become consequence-bearing action.

🧭

Document type: Arqua Canonical Vocabulary

Canonical parent: The Constitutional Architecture of Enterprise Intelligence

Enterprise vocabulary: Enterprise Intelligence Architecture — Canonical Definitions

Architecture domain: Execution Admissibility Architecture

Primary runtime architecture: SCIA Runtime

Primary consequence boundary: T=0

Vocabulary scope: Domain-specific

Architecture version: Arqua Canonical Architecture v1.0

Status: Published canonical reference

Architecture links

The Constitutional Architecture of Enterprise Intelligence · Enterprise Intelligence Architecture · Enterprise Intelligence Architecture — Canonical Definitions · The Enterprise Control Plane · Agent Control Plane · Architecture of Record · SCIA Runtime

Canonical hierarchy

Arqua Canonical Architecture

The constitutional reference architecture for Enterprise Intelligence.

It defines the overall system through which operational reality becomes governed representation, institutional memory, runtime context, coordinated action, consequence and learning.

Enterprise Intelligence Architecture — Canonical Definitions

The enterprise-wide vocabulary inherited by all Arqua architecture domains.

It defines the common meaning of constructs such as:

Operational Reality
Observed Evidence
Enterprise Representation Intelligence
Accepted Institutional Representation
Institutional Memory
Runtime Context Assembly
Enterprise Control Plane
Governed Coordination
Outcome Evidence
Representation Evolution

Execution Admissibility Architecture — Canonical Definitions

The specialised vocabulary governing the consequence boundary.

It defines the constructs through which a proposed action is assessed against authority, state, context, constraints, risk and evidence before execution is permitted to bind at T=0.

A domain definition may specialise an enterprise definition, but it must not silently redefine it.

What this page governs

This page governs terminology relating to:

proposed actions;
authority and authority lineage;
consequence-bearing execution;
runtime state;
execution evidence;
constraints and conditions;
admissibility resolution;
T=0;
SCIA Runtime;
Architecture of Record;
Enterprise Execution Control Plane;
agent participation in execution;
permitted, conditional, escalated, prevented and unresolved outcomes.

What this page does not redefine

This page does not redefine:

Enterprise Intelligence Architecture;
Enterprise Representation Intelligence;
Institutional Memory;
System Model Foundation;
Runtime Context Assembly;
Enterprise Control Plane;
Semantic Governance Operating Model;
Semantic Contract Surface;
Outcome Evidence;
Representation Evolution.

For the authoritative enterprise definitions of these terms, see Enterprise Intelligence Architecture — Canonical Definitions.

Control plane terminology

Arqua uses three related but distinct control-plane concepts. Do not merge or rename these constructs.

Enterprise Control Plane

The distributed architecture that preserves institutional continuity as accepted architecture becomes implementation, runtime operation, decision, action, consequence and revision.

Canonical transformation:

Accepted Architecture → Conformant Operational Implementation

Agent Control Plane

The applied control architecture that governs enterprise agents through identity, registry, ownership, authority, permitted use, runtime enforcement, observability, evaluation, monitoring, security, cost control, lifecycle management and offboarding.

Canonical transformation:

Accepted Agent Architecture → Agent Control Contract → Conformant Operational Agent Runtime

Enterprise Execution Control Plane

The specific consequence-boundary control architecture within Execution Admissibility Architecture that resolves whether a proposed action may bind at T=0.

Canonical transformation:

Proposed Action → Admissibility Resolution → Permitted or Prevented Consequence

Definitions table

Term	Definition	What it is not	Where it sits	Inputs	Outputs	Canonical parent	Related concepts
Arqua	The architectural category owner defining Execution Admissibility Architecture (EAA): the discipline of governing whether consequential execution is permitted to bind at T=0.	Not an implementation platform, systems integrator, or assurance body.	Category definition + reference architectures + diagnostics.	Enterprise operating context; consequence surfaces; governance intent.	Architectural definitions, reference architectures, diagnostics, and mapping patterns.	—	EAA; AoR; SCIA Runtime; Pressure Tests; Authority Pattern Library.
Enterprise Bridge	A translation and alignment construct that maps an established enterprise discipline, framework, standard, regulatory context, platform pattern, or sector practice into Execution Admissibility Architecture, showing where existing governance remains necessary and where consequence-binding execution must be controlled at T=0.	Not an official extension, certification, endorsement, compliance mapping, legal opinion, implementation guide, or formal representation of any external framework, standard, regulator, or vendor platform.	Between existing enterprise governance disciplines and Arqua’s EAA/AoR/SCIA Runtime architecture corpus.	External framework or discipline; governance artefacts; architecture artefacts; controls; workflows; system patterns; sector context; consequence-bearing workflow candidates.	Bridge note; alignment questions; execution-admissibility gap analysis; artefact mapping; Pressure Test candidates; AoR baseline prompts; implementation-boundary guidance.	Arqua	EAA; AoR; SCIA Runtime; Pre-Execution Pressure Test; Authority Pattern Library; execution surface; T=0; evidence at execution.
Execution Admissibility Architecture (EAA)	The architecture discipline governing the boundary between decision, admissibility, and execution, with the invariant: no state transition without proven integrity at the commit boundary.	Not decision governance; not post-hoc audit; not an “AI policy” document.	Category-level architecture above systems and delivery.	Authority models; operating policies; execution surfaces; constraints; context requirements.	Admissibility model, control points, and runtime enforcement patterns.	Arqua	AoR; SCIA Runtime; admissibility vector; institutional commit boundary; evidence at execution.
Enterprise Execution Control Plane	The enterprise-level architectural control boundary, defined by Execution Admissibility Architecture, through which an organisation determines whether a proposed consequence-bearing action is admissible at T=0 before it binds institutional consequence. It resolves authority, state, context, constraints, risk, and evidence at or before the institutional commit boundary.	Not a software platform, orchestration layer, observability tool, policy engine, AI governance dashboard, model evaluation framework, managed service, compliance certification, legal assurance, or implementation product.	Between the category discipline and operational implementation. Execution Admissibility Architecture defines it. Architecture of Record maps where it must exist. SCIA Runtime defines the reference runtime architecture for admissibility resolution at the commit boundary.	Execution surfaces; authority chains; operational state; semantic context; policy constraints; risk conditions; evidence requirements.	Admissibility control model; required control points; admissibility outcomes; evidence at execution; and implementation boundaries for delivery partners.	Execution Admissibility Architecture (EAA)	EAA; AoR; SCIA Runtime; T=0; institutional commit boundary; admissibility vector; evidence at execution; Execution-Bound Enterprise.
Execution Admissibility	Execution Admissibility is the institutional determination that sufficient authority, state, context, constraints, risk resolution and evidence exist for a proposed consequence-bearing action to execute at T=0. It determines whether execution may proceed, must proceed with conditions, requires escalation, must be prevented or cannot yet be resolved because information is insufficient. Canonical statement: No consequence-bearing state transition without proven integrity at the commit boundary.	Not decision approval, retrospective audit, model confidence, human review in isolation, or a best-effort control.	Runtime decision at the commit boundary.	Authority state; constraints; contextual signals; evidence; current system state; risk conditions.	Admissibility outcome: permitted, conditional, escalated, prevented or unresolved, with associated evidence trail.	EAA	T=0; institutional consequence; SCIA Runtime; evidence at execution; context at execution.
T=0	T=0: The institutional commit boundary: the moment a proposed action becomes consequence-binding execution. In distributed systems, a workflow may contain multiple T=0 boundaries, each tied to a distinct consequence class or irreversible state transition.	Not the time a decision is made; not a review meeting; not a policy publication moment.	Commit boundary of consequence-binding execution.	Proposed action; current authoritative state; required proofs.	Consequence-binding state transition (or refusal to transition).	EAA	Institutional commit boundary; execution surface; SCIA Runtime; evidence at execution.
Institutional consequence	The durable organisational outcome produced when an action binds (e.g., funds move, a contract commits, access changes, a record becomes authoritative).	Not “impact” in the abstract; not analytics; not internal intention.	Downstream of execution; the reason admissibility exists.	Execution events; institutional systems of record; obligations.	Committed state, obligation, entitlement, liability, access, or record truth.	EAA	Consequence system; execution surface; commit boundary; AoR.
Institutional commit boundary	The boundary where an execution system transitions a state that binds institutional consequence (the “commit” surface).	Not a business approval workflow; not a governance committee; not a documentation gate.	Runtime boundary inside execution systems.	Action intent; admissibility outcome; proofs; constraints.	Committed state transition (or rejection/hold).	EAA	T=0; execution surface; consequence system; SCIA Runtime.
Execution surface	Any interface, workflow, API, event, or mechanism through which a system can perform a consequence-binding action. Execution surfaces may include agent tool calls, workflow automations, orchestration steps, embedded SaaS agents, API calls, event triggers or multi-agent execution paths where consequence-bearing action may be initiated.	Not a “feature”; not only user interfaces; not only AI models.	Located and mapped by AoR; governed by EAA.	System interfaces; triggers; service boundaries; permissions; agent tool calls; orchestration paths.	Identified control points requiring admissibility enforcement.	AoR	AoR; commit boundary; consequence systems; pressure tests; Execution Agent; Agent-Influenced Action; Agent Control Contract; Runtime Enforcement Gate.
Consequence system	A system whose state transitions create institutional consequence (payments, entitlements, contracts, identity/access, regulatory records, critical infrastructure operations).	Not a dashboard or reporting layer; not a sandbox.	Downstream execution domain where consequence is committed.	Execution requests; admissibility outcomes; authoritative state.	Committed changes and records of consequence.	AoR	Institutional consequence; execution surface; commit boundary; evidence at execution.
Architecture of Record (AoR)	The mapping layer that defines where consequence binds and where execution control must exist, linking enterprise operating intent to concrete execution surfaces.	Not a documentation library; not an enterprise architecture “as-is/to-be” deck; not a CMDB.	Between category discipline (EAA) and runtime enforcement (SCIA Runtime).	Business capabilities; consequence systems; workflows; integration topology; authority chains.	Authoritative map of consequence surfaces and required control points.	EAA	Execution surfaces; commit boundaries; SCIA Runtime placement; pressure tests.
SCIA Runtime	SCIA Runtime — Stateful Contextual Integrity Architecture — is the runtime control architecture within Execution Admissibility Architecture. It determines whether a proposed consequential action is admissible at T=0, before execution binds institutional consequence.	Not “decision governance”; not a model; not a policy document; not an implementation product claim.	Runtime enforcement layer at/near the commit boundary.	Authority state; constraints; context; evidence; current operational state.	Admissibility outcomes + enforced execution allow/deny/hold + provenance.	EAA	Admissibility vector; evidence at execution; constraint compilation; authority lifecycle integrity.
Stateful Contextual Integrity Architecture	The underlying architectural concept (expanded name) for SCIA Runtime: integrity is proven in-state, in-context, at the moment execution would bind consequence.	Not “sovereign coherent intelligence”; not a general AI architecture.	Definition level inside SCIA Runtime.	Operational state; contextual signals; proofs.	Integrity-gated state transition permissions.	SCIA Runtime	State at execution; context at execution; evidence at execution; T=0.
Admissibility Vector	The defined set of required dimensions that must resolve before a consequence-binding state transition is permitted.	Not a checklist for humans; not an after-action audit framework.	Core construct inside EAA/SCIA Runtime.	Authority; state; constraints; context; evidence.	Vector resolution result (admissible / not admissible / conditional) with justification.	EAA	Authority; constraint compilation; evidence/context/state at execution.
Authority	The legitimate institutional basis by which an actor or system is permitted to cause a specific consequential action. For agent-mediated operation, authority is derived, not inherent. An agent has no independent institutional authority of its own; it exercises authority delegated, configured, approved or derived from a worker, role, process, policy, governance approval, institutional mandate or explicit execution permission. Agents are accountable objects, not accountable parties. Accountability remains with the institution, owner, worker, role, process or approving authority.	Not “role-based access” alone; not informal influence; not system capability; not independent agent accountability.	One axis of the admissibility vector.	Mandates; delegations; policies; approvals; accountabilities; agent authority source; control-contract scope.	Authority claims + required proofs at execution.	EAA	Authority state; authority lifecycle integrity; authority drift; AoR; Effective Actor; Agent Control Contract; Agentic Threshold; Execution Agent; Ghost Agent; Authority Laundering.
Authority State	The current, runtime-resolved status of authority for a proposed action (valid/invalid/expired/superseded/conditional), computed from authoritative sources and context. For agent-mediated execution, authority state includes the validity of the agent’s authority source, Agent Control Contract, lifecycle state, delegated scope and effective actor context.	Not a static permission grant; not an org chart.	Runtime resolution inside SCIA Runtime.	Delegations; entitlements; constraints; context; evidence; agent authority source; lifecycle state; effective actor context.	Resolvable authority outcome used for admissibility at T=0.	SCIA Runtime	Authority; state at execution; context at execution; evidence at execution; Effective Actor; Agent Lifecycle State; Agent Control Contract; Ghost Agent.
Authority Lifecycle Integrity	The end-to-end integrity of authority as it is created, delegated, interpreted, exercised, and evidenced—such that execution can prove legitimacy at T=0. Agent authority lifecycle integrity requires agents to be suspended, retired, transferred or explicitly reapproved when the worker, role, process, policy, project or approval basis from which they derive legitimacy changes.	Not “governance process maturity”; not policy intent without enforceable binding.	Governance + architecture construct supporting SCIA Runtime.	Mandates; delegations; records; constraints; execution evidence patterns; agent lifecycle changes.	Provenance, lineage, and enforceable authority semantics.	EAA	Authority drift; evidence at execution; AoR; constraint compilation; Ghost Agent; Agent Lifecycle State; Agent Control Contract; Authority Laundering.
Authority Drift	The divergence between intended authority and effective authority as systems, teams, vendors, or automation pathways change over time.	Not “bad actors”; not only a training problem; not a monitoring metric by itself.	Risk pattern diagnosed by pressure tests and mitigated by EAA/SCIA Runtime.	System changes; policy changes; automation pathways; vendor layers.	Uncontrolled execution surfaces; misaligned authority states.	Authority Lifecycle Integrity	Pressure tests; AoR; evidence/context at execution.
Constraint Compilation	The process of assembling and resolving the constraints that must hold for a proposed action to be admissible at T=0 (policy, rules, thresholds, prohibitions, conditions). Agent Control Contracts may act as compiled constraint sources for agent runtime behaviour, including permitted data, permitted tools, memory policy, output policy, cost policy, lifecycle policy and execution gates.	Not “policy documentation”; not model prompt rules; not informal guidelines.	Runtime support function inside SCIA Runtime.	Policies; rules; contracts; regulatory constraints; state constraints; Agent Control Contracts; permitted-use controls.	Executable constraint set used in admissibility evaluation.	SCIA Runtime	Constraints; evidence at execution; context/state at execution; Agent Control Contract; Runtime Enforcement Gate; Permitted Use.
Evidence at execution	The proofs produced and captured at T=0 demonstrating that admissibility conditions were met (or not met) for a specific execution attempt. Where an agent materially influenced the proposed action, evidence at execution should reference the relevant Agent Run Record or Agent Evidence Bundle.	Not post-hoc narrative; not a compliance report template.	Output of SCIA Runtime at commit boundary.	Logs, attestations, signatures, lineage, resolved context/state snapshots; Agent Run Records; Agent Evidence Bundles.	Audit-grade trace of admissibility evaluation and enforcement outcome.	SCIA Runtime	Execution admissibility assurance; institutional commit boundary; context/state at execution; Agent Run Record; Agent Evidence Bundle; Agent-Influenced Action; Execution Agent.
Context at execution	The runtime contextual inputs required to evaluate admissibility (who/what/where/when, jurisdiction, purpose, current conditions, surrounding system state). For agent-mediated execution, context at execution includes the effective actor: agent identity, control-contract version, initiating user or process, authority context, approved purpose, task, environment and lifecycle state.	Not “background information” for a decision meeting; not a static form field.	One axis supporting admissibility at T=0.	Signals, environment, jurisdictional parameters, workflow context; effective actor context.	Resolved context snapshot used in admissibility evaluation.	SCIA Runtime	Constraints; state at execution; evidence at execution; Effective Actor; Agent Run Record; Agent Control Contract.
State at execution	The authoritative operational state relevant to the proposed action at the moment of commit (including prior transitions and current entitlements/obligations).	Not a database record in isolation; not a plan; not a future intention.	One axis supporting admissibility at T=0.	System-of-record state; workflow state; authority state; constraint state.	Resolved state snapshot used for admissibility evaluation.	SCIA Runtime	Operational state architecture; evidence at execution; institutional consequence.
Pre-Execution Pressure Test	A diagnostic that reveals where execution is currently uncontrolled by mapping the surfaces where automated or human decisions can bind consequence without an admissibility boundary. For agent-mediated environments, the pressure test should identify whether agents can retrieve, recommend, route, trigger tools or influence execution without a registered identity, valid authority source, Agent Control Contract or admissibility boundary.	Not penetration testing; not an implementation audit; not assurance certification.	Diagnostic entry point into AoR/EAA work.	A selected high-consequence workflow; stakeholders; system topology; agent registry and runtime evidence.	Execution topology + identified uncontrolled execution surfaces + control-point candidates.	Arqua	AoR; execution surface; authority drift; consequence systems; Agentic Threshold; Execution Agent; Agent Control Plane; Ghost Agent.
Execution Admissibility Assurance	The ability to demonstrate, with evidence generated at T=0, that consequence-binding actions only executed when admissibility resolved.	Not a regulatory certification; not a claim that systems are compliant by definition.	Outcome posture enabled by EAA/SCIA Runtime patterns.	Evidence at execution; admissibility outcomes; control-point mapping.	Demonstrable execution integrity trail suitable for oversight and review.	EAA	Evidence at execution; authority lifecycle integrity; AoR.
Authority Pattern Library	A library of recurring authority and consequence patterns observed across sectors, used to accelerate diagnosis and architectural design.	Not a control catalogue; not a compliance standard.	Supporting knowledge base for Arqua diagnostics and architecture.	Sector patterns; authority archetypes; consequence-binding motifs.	Reusable patterns and diagnostic lenses.	Arqua	Authority pattern; pressure tests; AoR.
Authority pattern	A recurring structural configuration of mandate, delegation, accountability, and execution surfaces that determines how authority actually binds in an operating environment.	Not an org chart; not a job description.	Pattern unit within the Authority Pattern Library.	Observed workflows; governance artefacts; system boundaries.	Pattern description usable for diagnosis and design.	Authority Pattern Library	Authority lifecycle integrity; authority drift; consequence systems.
AI-Ready Enterprise Semantics	The semantic preparation of enterprise meaning (definitions, types, relationships, decision vocabularies) so automated and human systems share stable interpretation.	Not prompt engineering; not a data lake; not a taxonomy project alone.	Upstream prerequisite feeding EAA/SCIA Runtime.	Business definitions; policy intent; domain models; metadata.	Shared semantic layer that stabilises meaning across systems and time.	Arqua	Meaning and coherence; operational state architecture; constraint compilation.
Operational State Architecture	The architecture of how operational states are represented, transitioned, and proven across systems—so state at execution can be resolved reliably at T=0.	Not a single database schema; not a workflow diagram.	Foundational architecture supporting admissibility evaluation.	State models; transitions; system-of-record design; provenance.	Resolvable, authoritative state snapshots and transition semantics.	EAA	State at execution; evidence at execution; AoR.
Execution-Bound Enterprise	An enterprise posture where decisions may be distributed and exploratory, but execution is governed: consequence binds only through admissible state transitions at T=0.	Not centralised decision-making; not “slower governance”.	Operating model outcome enabled by EAA + AoR + SCIA Runtime.	Mapped execution surfaces; admissibility controls; semantic prerequisites.	Reduced uncontrolled execution; durable accountability; evidence at execution.	EAA	T=0; institutional consequence; AoR; SCIA Runtime.
Sovereign AI	AI governance framing concerned with jurisdictional control, public-sector authority, and national sovereignty constraints on AI-driven decisions and execution.	Not a claim of sovereign coherent intelligence; not a guarantee of national control by default.	Context framing that may shape constraints, authority, and context at execution.	Jurisdictional policies; security/assurance requirements; cross-border constraints.	Sovereignty-aligned constraint and governance requirements.	Arqua	Jurisdiction-aware controls; constraints; authority; government contexts.

Enterprise Bridges are not canonical definitions of external frameworks. They are Arqua-authored translation and alignment notes that help established enterprise disciplines identify where execution-admissibility questions arise.

Agent Control Plane definitions

Term	Definition	What it is not	Where it sits	Inputs	Outputs	Canonical parent	Related concepts
Agent Architecture	The applied reference architecture for governing AI-enabled systems as runtime actors when they exercise task-level discretion inside enterprise authority boundaries.	Not a product guide, vendor framework, model-evaluation method, or prompt pattern.	Inside the Enterprise Control Plane, upstream of operational agent runtime.	Enterprise authority boundaries; agent purpose; risk tier; data/tool boundaries; monitoring obligations.	Accepted Agent Architecture and Agent Control Contract requirements.	Enterprise Control Plane	Enterprise Agent; Agentic Threshold; Agent Control Plane; Accepted Agent Architecture.
Enterprise Agent	An AI-enabled runtime actor that can retrieve governed context, interpret intent, select tools, shape recommendations, coordinate workflow or influence operational consequence under delegated, role-derived, process-derived, policy-derived or explicitly granted authority.	Not a human accountable party; not a standalone institutional authority; not every AI feature by default.	Runtime actor governed by the Agent Control Plane.	Delegated authority; governed context; task; tools; policy constraints; lifecycle state.	Recommendations, routed work, tool calls, generated artefacts, or agent-influenced actions.	Agent Control Plane	Effective Actor; Agent Control Contract; Agentic Threshold; Execution Agent.
Agentic Threshold	The point at which an AI-enabled capability must be governed as an enterprise agent because it can exercise task-level discretion inside an enterprise authority boundary.	Not a model-size threshold, autonomy label, or vendor feature classification.	Classification boundary between ordinary AI capability and governed enterprise agent.	Capability behaviour; authority boundary; task discretion; tool/data access; operational consequence potential.	Determination that Agent Control Plane governance is required.	Agent Architecture	Enterprise Agent; Agent Control Plane; Execution Agent; Pre-Execution Pressure Test.
Agent Control Plane	The applied control architecture that governs enterprise agents through registry, identity, ownership, authority, permitted use, runtime enforcement, observability, evals, monitoring, security, cost control, lifecycle management and offboarding.	Not the Enterprise Control Plane as a whole; not EAA; not a product, dashboard, or orchestration layer by itself.	Inside the Enterprise Control Plane; upstream of and adjacent to runtime agent operation.	Accepted Agent Architecture; agent registry; authority sources; policies; tool/data boundaries; lifecycle rules.	Agent Control Contracts, registry state, runtime enforcement requirements, monitoring evidence, lifecycle outcomes.	Enterprise Control Plane	Agent Architecture; Agent Control Contract; Conformant Operational Agent Runtime; Agent Registry / Agent CMDB.
Accepted Agent Architecture	The approved design, purpose, authority source, risk tier, data boundaries, tool permissions, monitoring obligations, evidence requirements and lifecycle conditions for an enterprise agent.	Not an informal design note, prompt, implementation ticket, or runtime log.	Approved architecture baseline for a governed enterprise agent.	Agent purpose; authority source; risk classification; data/tool access; monitoring and evidence obligations.	Approved baseline translated into an Agent Control Contract.	Agent Architecture	Agent Control Contract; Agent Registry / Agent CMDB; Conformant Operational Agent Runtime.
Agent Control Contract	The versioned expression of accepted agent architecture, designed to be machine-readable and enforceable through runtime controls. It binds a deployed runtime to approved identity, purpose, authority, data, tools, memory, monitoring, cost, evidence and lifecycle controls.	Not a legal contract by default; not a prompt alone; not an approval substitute; not an independent authority source.	Control bridge between Accepted Agent Architecture and deployed runtime.	Accepted Agent Architecture; registry state; policies; permitted data/tools; lifecycle conditions.	Enforceable runtime constraints and evidence requirements for a deployed agent.	Agent Control Plane	Conformant Operational Agent Runtime; Effective Actor; Runtime Enforcement Gate; Constraint Compilation.
Conformant Operational Agent Runtime	A deployed agent runtime that preserves the approved identity, authority, purpose, permitted use, monitoring, evidence, enforcement and lifecycle controls defined by accepted agent architecture and the Agent Control Contract.	Not any deployed agent; not conformance by intention; not a guarantee of admissible execution.	Operational runtime governed by the Agent Control Plane.	Agent Control Contract; registry state; runtime telemetry; monitoring controls; lifecycle state.	Controlled agent operation and runtime evidence.	Agent Control Plane	Agent Run Record; Agent Lifecycle State; Agent Evidence Bundle; Execution Agent.
Effective Actor	The compound runtime actor formed by agent identity, control-contract version, initiating user or process, authority context, approved purpose, task, environment and lifecycle state.	Not the agent identity alone; not the user alone; not a static role assignment.	Runtime context used to resolve authority and accountability for agent-mediated action.	Agent identity; contract version; initiating user/process; authority context; task; environment; lifecycle state.	Resolved actor context for authority, context, evidence and admissibility evaluation.	Agent Control Plane	Authority State; Context at execution; Agent Run Record; Execution Agent.
Agent Registry / Agent CMDB	The authoritative control surface for agent identity, ownership, authority source, risk tier, approved purpose, permitted data, tools, actions, monitoring obligations and lifecycle state.	Not a model inventory alone; not a static spreadsheet; not an execution approval mechanism by itself.	System of control record for governed enterprise agents.	Agent architecture approvals; owners; authority sources; risk tier; data/tool permissions; lifecycle events.	Authoritative agent identity, ownership, control and lifecycle records.	Agent Control Plane	Agent Control Contract; Agent Lifecycle State; Agent Run Record; Ghost Agent.
Agent Run Record	The runtime evidence record for a material agent run, including agent identity, version, effective actor, authority source, input, retrieved sources, prompt, model, tools, memory access, output, human review, action, cost, outcome and incident flags.	Not a generic application log; not a post-hoc narrative; not sufficient evidence if material sources are missing.	Runtime evidence layer for agent operation.	Agent run telemetry; prompts; retrieved sources; tool calls; memory access; outputs; review and outcome data.	Run-level evidence for reconstruction, monitoring, incident review and admissibility reference.	Agent Control Plane	Agent Evidence Bundle; Evidence at execution; Effective Actor; Agent-Influenced Action.
Agent Evidence Bundle	The assembled evidence required to reconstruct a material agent run, agent-influenced decision, incident or action.	Not a compliance report by itself; not a substitute for admissibility evidence at T=0.	Evidence assembly layer supporting reconstruction and review.	Agent Run Records; source artefacts; model/tool records; human review; outcomes; incident data.	Reconstructable evidence package for review, investigation or admissibility traceability.	Agent Control Plane	Agent Run Record; Evidence at execution; Agent-Influenced Action; Execution Agent.
Agent Lifecycle State	The enforceable status that determines whether an agent may run, must be restricted, must be reviewed, must be suspended, must be retired or must be archived.	Not a project status label; not a documentation-only field.	Lifecycle enforcement layer inside the Agent Control Plane.	Approval state; owner/role validity; policy changes; monitoring results; incidents; expiry and review dates.	Runtime permission, restriction, suspension, retirement or archival outcome.	Agent Control Plane	Ghost Agent; Authority Lifecycle Integrity; Agent Registry / Agent CMDB; Agent Control Contract.
Ghost Agent	An agent that continues operating after the worker, role, project, process, policy or approval basis from which it derived legitimacy has ended.	Not merely an inactive agent; not a benign orphaned asset by default.	Failure mode in agent authority lifecycle integrity.	Lifecycle drift; expired authority source; unresolved ownership; uncontrolled runtime access.	Uncontrolled or illegitimate agent operation requiring suspension, retirement, transfer or reapproval.	Authority Lifecycle Integrity	Agent Lifecycle State; Agent Registry / Agent CMDB; Authority Drift; Pre-Execution Pressure Test.
Authority Laundering	A failure mode in which one agent, user, workflow or orchestration path routes through another agent to access data, tools or authority it does not legitimately possess.	Not legitimate delegation; not ordinary tool chaining; not a logging issue alone.	Failure mode in authority, permitted use and runtime enforcement.	Agent chains; tool access; delegated authority; workflow routing; control-contract boundaries.	Illegitimate authority expansion or access path requiring enforcement and review.	Authority Lifecycle Integrity	Authority; Agent Control Contract; Effective Actor; Runtime Enforcement Gate.
Execution Agent	An enterprise agent that performs or triggers operational action and therefore requires explicit execution authority and admissibility evaluation before consequence binds.	Not every enterprise agent; not a recommendation-only agent unless it materially shapes execution.	Agent class at or near consequence-bearing execution surfaces.	Execution authority; action type; consequence class; control contract; context; evidence.	Proposed or triggered operational action subject to admissibility evaluation.	Agent Control Plane	Execution surface; Agent-Influenced Action; EAA; SCIA Runtime.
Agent-Influenced Action	A proposed or completed action whose formation, recommendation, routing, prioritisation, evidence selection or execution path was materially shaped by an enterprise agent.	Not every downstream human action after AI use; not proof that the agent executed the action.	Between agent runtime influence and consequence-bearing execution.	Agent output; routing; prioritisation; evidence selection; human review; execution path.	Action requiring traceability to agent influence and, where consequence-bearing, admissibility evaluation.	Agent Control Plane	Agent Run Record; Agent Evidence Bundle; Execution Agent; Evidence at execution.

These definitions describe architectural control concepts. They do not assert that any particular agent, platform, vendor product or implementation is compliant, assured, safe or admissible by default.

Canonical admissibility vector

The canonical execution admissibility vector consists of authority, state, constraints, context, and evidence.

Agent-mediated execution note

The canonical execution admissibility vector remains authority, state, constraints, context and evidence.

Agent-mediated execution does not add a new vector axis by default.

Instead, agent-specific controls resolve into the existing vector:

Agent identity, authority source, lifecycle state and effective actor support authority.
Agent task, purpose, environment and orchestration path support context.
Agent Control Contract, permitted-use policy, tool policy, memory policy and execution gates support constraints.
Agent Run Records and Agent Evidence Bundles support evidence.
Operational state, workflow state and consequence class remain part of state at execution.

Execution Admissibility Architecture still evaluates whether proposed consequence-bearing action may bind at T=0. The Agent Control Plane governs the agent before and during runtime. EAA governs consequence at the commit boundary.

Related paper concepts

Term	Definition
Enterprise Control Plane	The distributed continuity architecture that preserves institutional continuity as accepted architecture becomes implementation, AI execution, decision, action, consequence and revision.
Agent Architecture	The applied reference architecture for governing AI-enabled systems as runtime actors inside enterprise authority boundaries.
Enterprise Agent	An AI-enabled runtime actor operating under delegated, derived or explicitly granted enterprise authority.
Agentic Threshold	The point at which an AI-enabled capability must be governed as an enterprise agent because it exercises task-level discretion inside an enterprise authority boundary.
Agent Control Plane	The applied control architecture governing enterprise agents through identity, ownership, authority, permitted use, enforcement, observability, monitoring, security, cost, lifecycle and offboarding controls.
Accepted Agent Architecture	The approved design, purpose, authority source, risk tier, boundaries, permissions, monitoring obligations, evidence requirements and lifecycle conditions for an enterprise agent.
Agent Control Contract	The versioned, machine-readable and enforceable expression of accepted agent architecture.
Conformant Operational Agent Runtime	A deployed agent runtime that preserves the approved controls defined by accepted agent architecture and the Agent Control Contract.
Effective Actor	The compound runtime actor formed by agent identity, contract version, initiating user or process, authority context, approved purpose, task, environment and lifecycle state.
Agent Registry / Agent CMDB	The authoritative control surface for agent identity, ownership, authority source, approved purpose, permissions, monitoring obligations and lifecycle state.
Agent Run Record	The runtime evidence record for a material agent run.
Agent Evidence Bundle	The assembled evidence required to reconstruct a material agent run, agent-influenced decision, incident or action.
Agent Lifecycle State	The enforceable status that determines whether an agent may run, be restricted, reviewed, suspended, retired or archived.
Ghost Agent	An agent that continues operating after the basis from which it derived legitimacy has ended.
Authority Laundering	A failure mode where one actor or path routes through another agent to access data, tools or authority it does not legitimately possess.
Execution Agent	An enterprise agent that performs or triggers operational action and therefore requires explicit execution authority and admissibility evaluation before consequence binds.
Agent-Influenced Action	A proposed or completed action materially shaped by an enterprise agent.
Execution Attribution Collapse	A failure mode in which a consequential action can be technically executed and procedurally logged, but the institution cannot reconstruct a defensible chain connecting authorised judgment to evidence, interpretation, system transformations, permissions, and execution at T=0.
Execution Passport	A portable authority-state bundle carried with a proposed consequential action. It presents the authority claim, scope, context, constraints, evidence, validity conditions, expiry, and escalation requirements needed for SCIA Runtime to determine execution admissibility at T=0. It is not itself an approval, permission grant, product, or guarantee of admissibility.
Alignment Architecture	A three-layer architectural model connecting Meaning, Execution, and Admissibility to preserve coherent action in AI-mediated systems. It treats alignment as architectural continuity between originating intent and consequential execution, and positions admissibility as the control boundary at T=0 where consequence-bearing action is permitted or refused.
Meeting-to-runtime gap	The gap between a point-in-time human decision or approval and downstream execution under changed authority, evidence, state, context, policy, or consequence conditions.

Note: Authority decay is discussed in Arqua papers as a time-sensitive form of Authority Drift. Use Authority Drift as the canonical site term.

Meaning and coherence note

Meaning and coherence are semantic prerequisites. They support admissibility, but do not replace the execution admissibility vector.

Architecture relationship diagram (text)

Enterprise Intelligence Architecture provides the whole reference architecture.
Enterprise Control Plane preserves institutional continuity.
Agent Control Plane governs AI runtime actors.
Execution Admissibility Architecture governs whether consequence-bearing action may bind.
Architecture of Record maps where consequence binds.
SCIA Runtime resolves admissibility at T=0.
Pre-Execution Pressure Test identifies uncontrolled execution surfaces.

Boundary statement

These definitions describe architectural concepts. They do not assert legal compliance, regulatory certification, assurance, system operation, or implementation.

Related canonical architecture

Execution Admissibility Architecture is the consequence-boundary domain within the broader Arqua Canonical Architecture.

It receives governed context, authority, state and proposed intent from the enterprise runtime architecture and determines whether a proposed action may bind institutional consequence.

Explore:

The Constitutional Architecture of Enterprise Intelligence
Enterprise Intelligence Architecture
Enterprise Intelligence Architecture — Canonical Definitions
The Enterprise Control Plane
Agent Control Plane
Architecture of Record
SCIA Runtime
Enterprise Representation Intelligence
Institutional Memory
Runtime Context Assembly