Assessment instrument

Control catalog

A minimum control baseline for a bounded V.E.N.K.A.T implementation. Extend it for sector, jurisdiction, and system risk; preserve IDs and applicability decisions for traceability.

Tracking schema

Fields for every assessment record

IdentityAssessment ID, control ID/version, system/scope, layer, requirement, objective, criticality, mandatory-gate flag

AccountabilityControl owner, operator, tester, approver, business owner, assessor, segregation-of-duties check

ApplicabilityApplicable/not applicable, rationale, tailoring, inherited/shared control, provider, related obligation and risk IDs

TestingDesign result, operating result, method, population, sample, period, test date, next test, score, confidence

EvidenceEvidence IDs/links, source, owner, collection time, period covered, integrity/hash, retention, sensitivity

ActionFinding ID/severity, root cause, remediation owner/due date, roadmap item, exception ID/expiry, residual risk, status

Record integrity: evidence must be attributable, relevant to the assessment period, protected from unauthorized change, and reproducible by an independent assessor. A policy alone proves design—not operation.

Verified Data

Trusted inputs, lineage, quality, and lawful use.

ID	Control requirement	Criticality	Minimum evidence	Certification test
V-01	Decision-critical data has an accountable owner, approved business definition, semantic and structural contract, classification, and permitted-use statement.	Critical · Gate	Inventory, glossary/taxonomy links, contracts, approvals, classifications	Sample decisions; trace every critical input to its definition, semantics, current contract, source, and owner.
V-02	Automated quality and freshness checks block, quarantine, or safely degrade on threshold breach.	Critical · Gate	Rules, SLOs, run results, alerts, disposition logs	Inject invalid and stale records; observe enforced response.
V-03	End-to-end lineage and provenance connect source, transformations, model context, decision, and retained outcome.	High	Lineage graph, trace samples, change history	Reconstruct sampled outcomes without implementer assistance.
V-04	Access, minimization, retention, deletion, and bias/privacy monitoring are risk-based and periodically reviewed.	High	Access reviews, DPIA, retention runs, bias results	Test unauthorized access and sample retention/deletion execution.

Event-Driven Architecture

Reliable, attributable, recoverable operational signals.

ID	Control requirement	Criticality	Minimum evidence	Certification test
E-01	Critical events have owned, versioned schemas, semantics, keys, compatibility policy, and producer/consumer inventory.	Critical · Gate	AsyncAPI/schema records, compatibility results, ownership	Attempt incompatible change and confirm prevention or managed migration.
E-02	Consumers safely handle duplicate, late, missing, malformed, and out-of-order events.	Critical · Gate	Idempotency design, test runs, DLQ records	Inject each failure mode and verify no unsafe duplicate action.
E-03	Critical event paths have latency, availability, lag, loss, and recovery SLOs with actionable monitoring.	High	Dashboards, alerts, incidents, SLO reviews	Trace event latency end to end and observe threshold alert.
E-04	Retention, replay, recovery, and dead-letter operations are authorized, auditable, and exercised.	High	Runbooks, access logs, recovery exercise	Perform controlled replay and reconcile resulting actions.

Native Spatial Intelligence

Valid location, topology, routing, and jurisdictional context.

ID	Control requirement	Criticality	Minimum evidence	Certification test
N-01	Spatial data records authoritative source, timestamp, CRS, units, precision, uncertainty, and permitted use.	Critical · Gate	Metadata, contracts, catalog and sample payloads	Sample spatial decisions and verify complete, correct metadata.
N-02	Routes and location actions enforce physical, safety, temporal, legal, and jurisdictional constraints.	Critical · Gate	Constraint registry, route tests, approvals	Attempt restricted-zone, over-limit, and infeasible routes.
N-03	Topology, geofence, boundary, CRS transformation, and edge-case accuracy are validated before release.	High	Benchmark suite, defect log, release results	Run known boundary/CRS cases and compare expected outputs.
N-04	Location privacy and degraded-position behavior are risk-based, transparent, and tested.	High	Privacy assessment, masking tests, fallback runbook	Lower confidence and verify safe fallback; test unauthorized precision.

Knowledge Graphs

Governed semantics, claims, relationships, and inference.

ID	Control requirement	Criticality	Minimum evidence	Certification test
K-01	Ontology/schema, competency questions, identifiers, change authority, and versioning are approved and owned.	Critical · Gate	Ontology, decision log, ownership, versions	Trace a schema change through review, migration, and consumers.
K-02	Material claims and relationships retain source, time, confidence, and inference status.	Critical · Gate	Provenance model, query samples, validation reports	Trace sampled asserted and inferred claims to sources.
K-03	Graph constraints, identity resolution, conflicts, temporal validity, and inference rules are continuously tested.	High	SHACL/rule tests, conflict queue, accuracy sample	Insert conflicting identities/claims and verify detection.
K-04	Authorization applies to nodes, edges, attributes, queries, exports, and derived sensitive knowledge.	High	Policy, entitlement review, query audit	Attempt cross-boundary query and inference; confirm denial and log.

AI Orchestration

Bounded, observable, interruptible agent behavior.

ID	Control requirement	Criticality	Minimum evidence	Certification test
A-01	Every agent, model, prompt, tool, owner, version, intended use, limit, and dependency is registered before use.	High	Registry, approvals, deployment reconciliation	Reconcile runtime inventory to approved registry.
A-02	Tool use is least-privileged, allow-listed, policy-checked, scoped per task, and denied by default.	Critical · Gate	Tool policy, IAM, authorization logs, negative tests	Attempt prohibited, excessive, and cross-tenant tool calls.
A-03	Consequential actions require risk-based human approval or an explicitly approved autonomous boundary.	Critical · Gate	Decision matrix, approval logs, autonomy authorization	Execute threshold boundary cases and verify routing/approval.
A-04	Emergency stop, timeout, rollback/compensation, safe state, and human escalation operate end to end.	Critical · Gate	Runbooks, drills, rollback records, escalation SLAs	Trigger kill switch and failed tool; verify containment and recovery.
A-05	Pre-release and ongoing evaluations cover quality, security, prompt injection, excessive agency, drift, cost, and harmful outcomes.	High	Eval suite/results, red-team findings, release gate	Rerun representative/adversarial tests and reconcile decision.

Trust & Governance

Accountability, assurance, security, recourse, and resilience.

ID	Control requirement	Criticality	Minimum evidence	Certification test
T-01	Named business and technical owners accept purpose, risk tier, decision rights, autonomy, and residual risk.	Critical · Gate	RACI, charter, risk acceptance, governance minutes	Interview owners; reconcile authority to signed records.
T-02	Security, privacy, safety, human-rights, legal, and supplier impact assessments precede deployment and material change.	Critical · Gate	Assessments, threat model, legal/supplier reviews	Trace identified harms to controls, tests, and accepted residuals.
T-03	Immutable-enough logs reconstruct identity, inputs, context, policy, reasoning record, approval, action, and outcome.	Critical · Gate	Audit schema, retention/integrity controls, trace samples	Reconstruct selected consequential actions independently.
T-04	Affected users receive appropriate notice, explanation, human review, correction, and recourse within defined SLAs.	High	Notices, explanation samples, appeal records, SLA dashboard	Submit a challenge and trace it through closure.
T-05	AI incidents are detected, contained, reported, learned from, and tested with business continuity and recovery.	Critical · Gate	Plan, drills, incidents, after-action improvements	Run a severe scenario and verify notification, stop, and recovery.
T-06	Exceptions are approved by authorized risk owners, narrowly scoped, compensated, monitored, and automatically expire.	High	Exception register, approvals, expiry alerts, closure evidence	Sample open/closed exceptions and test expired access/control.

Assessment rule

Score design and operation separately

0 · Absent

No credible design, or the requirement is contradicted in practice.

1 · Designed

Approved design exists, but operation is unproven or too recent.

2 · Partial

Operating evidence exists but has gaps, exceptions, limited coverage, or inconsistent results.

3 · Effective

Designed appropriately and operating consistently for the assessment period with reproducible evidence.

Weights: Critical = 3, High = 2, Standard = 1. Pass requires ≥85% overall and per critical layer; every “Gate” row must score 3. See the full decision workflow and assessment record template.