EA Codex Schema
This page lists the canonical object kinds exposed in the public EA Codex v1.1.0 schema catalogue. Each card links to the corresponding schema file embedded in this site. The Atlas remains the navigable view of relationships; this page is the schema object catalogue.
Intent and Meaning
EnterpriseIntent
Structured statement of business outcome and constraints. The EnterpriseIntent is the entry point of the ea.codex/v1 family. Every other artifact ultimately resolves to one or more intents.
BusinessObject
Typed enterprise data object: the canonical definition of what a piece of business data means, who governs it, and which semantic rules any instance must satisfy. BusinessObject sits...
ReferenceArchitecture
Typed reference architecture with layers, required components, relationships, invariants, and rendering metadata. Introduced in Chapter 6. The diagram is a rendered view of this declaration, not a…
Decision and Authority
ArchitecturePrinciple
Stable architectural anchor that downstream standards, decisions, and fitness functions reference. Introduced in Chapter 14. Principles are deliberately stable: when a vendor policy or regulation is…
TechnologyStandard
Catalogue of integration patterns or technology choices that realize an ArchitecturePrinciple, classified along a deliberately coarse axis (preferred, acceptable, acceptable-legacy, prohibited) so that the…
DecisionRecord
Typed architectural decision with explicit provenance. Captures the choice forced by an internal or external trigger and binds it to dependent kinds (ArchitecturePrinciple, TechnologyStandard,…
DesignAuthorityBody
Models the governing body that holds architectural authority. Covers the spectrum from a single human ARB chair with one agent assistant to a multi-agent council with cross-cutting members. Carries members…
DelegationPolicy
Authority model that defines how architectural decisions are delegated between humans and agents. A DelegationPolicy declares a set of named levels (typically L1 through L4, but the schema does not hardcode…
Architecture Work
ArchitecturePackage
BMAD-staged unit of architectural change. Each instance carries the artifact's content across the four BMAD stages: Brief (intent), Map (architecture), Act (executable artifacts), Double-check (evidence).…
ArchitectureChangePacket
TOGAF ADM-aligned change packet. Carries the architectural content of one change end-to-end across Architecture Vision, Business Architecture, Architecture Decisions, and downstream phases. Introduced in…
ContinuousArchitectureContract
Living contract that binds intent, capability, policies, quality attributes, design decisions, specifications, controls, and feedback for a single architectural unit. Introduced in Chapter 2. All sub-fields…
ProductLineSpecification
Software product line family definition. Captures the stable common core, named variation points with binding times, and the rules that link variant selection to policy consequences. Introduced in Chapter…
VariabilitySpecification
Permitted variation envelope for a product, platform, or program family. Used in Chapter 5 to govern SAP S/4HANA country variants under clean-core constraints. Consumed at every transition gate as…
SovereigntySpecification
Domain of architectural control over meaning, location, access, execution, evidence, and exit. Introduced in Chapter 13. The SovereigntySpec is the seed for sovereignty governance: it drives architecture…
Agentic Execution
AgentContract
Bounded contract for an enterprise AI agent. Binds business intent, semantic context, source policy, design decisions, controls, and feedback into one governed unit. Introduced in Chapter 10. All sub-fields…
AgentInteractionContract
Inter-agent communication contract: defines which agent may communicate with which other agent, for which purpose, with which message types, under which data-minimization rule, through which routing…
AgentMemoryPolicy
Agent memory retention and jurisdiction policy. Decides what an agent may remember, for how long, in which jurisdiction, under which deletion rule, and with which inspection mechanism. In regulated domains,…
ToolAccessPolicy
Tool catalog binding for an agent. Defines which tools an agent may call, under which purpose, with which secret-handling pattern, with which prohibited actions, and through which runtime enforcement point.…
Data and Contracts
DataContract
Consumer-facing data contract between a data product and one or more consumers. A DataContract operationalizes a single channel of access governed by a DataProductContract: it carries the technical schema,…
DataProductContract
Federated data product contract: a contracted unit of meaning, access, quality, policy, and evidence. Extends a federated dataset with purpose, jurisdictional constraints, allowed transformations, and…
Execution and Control
PolicyConstraint
Typed enterprise or regulatory policy that constrains decisions and specifications. Distinct from ArchitecturePrinciple (which guides) and TechnologyStandard (which catalogues approved patterns): a…
RegoPackage
Typed reference to a Rego policy package that mechanically enforces architectural decisions, principles, or standards. The Rego content itself lives in a separate policy repository; this artifact links the…
FitnessFunction
Typed architectural control that proves whether a decision, principle, standard, or policy remains respected. A FitnessFunction is a named, identifiable check (FF-SAP-001, FF-PV-002) implemented in a policy…
EdgeEvaluationPrompt
An LLM-evaluated transition condition between governance states. Used at edges in a graph where deterministic Rego rules cannot decide alone. Introduced in Chapter 5 for SAP transition gates and reused in…
ScenarioPack
Reusable bundle of validation scenarios used by an attractor's validation harness. ScenarioPacks are the human-curated content that BMAD's Double-check stage and the StrongDM-style attractor consume. They…
Feedback and Evidence
ObservabilityProfile
Required observability and evidence trace for an artifact. Defines which signals must be captured, where they are stored, and how long they are retained. Named in Chapter 13 as the observability counterpart…
EvidenceRecord
Typed evidence: the actual output produced when a control ran. Distinct from ObservabilityProfile (which describes what to collect) and from FitnessFunction (which is the control itself): an EvidenceRecord…