# Exact unsent Grok prompt — EM-WRK-06

Independent enterprise metamodel review. Do not browse, invent identifiers or claim publication readiness.

EM-WRK-06 covers Dependency, DependencyType, CriticalityAssessment, ImpactScenario and DependencyEvidence. Complete adjacent drafts cover WM-XCT-037 Dependency / Impact, WM-ACT-034 Assessment / Evaluation, WM-KNW-007 Claim, WM-KNW-008 Evidence Citation, WM-MAT-008 Observation and WM-KNW-015 Risk.

Assess this proposal: narrow WM-XCT-037 to dependency edge declaration, type binding, conditions, epistemics, completeness and declarative propagation licence; create identifier-unassigned Dependency Type Registry and Impact Scenario roots; profile Criticality Assessment on WM-ACT-034; bind WM-KNW-008 and WM-MAT-008 as external evidence masters. Allocate no other identifier.

Separate edge from type vocabulary, endpoint masters, evidence payload, observation, causal claim, risk, assessment and computed impact projection. Define canonical direction, inverse derivation, arity, guards, lag, phases, transitivity, cycle rules, completeness, unknown/absent states, confidence, snapshot pins, propagation limits, criticality schemes and uncertainty.

Test an API change and a resource delay on one frozen graph. Their affected sets must differ by type, phase and propagation licence; incomplete or unknown edges reduce completeness. Reject treating every graph path as a proven causal chain.

Return <=900 words with: Verdict; strongest evidence; strongest counterexample; identity/mastership; edge/type; evidence/epistemics; graph rules; impact scenario; criticality/assessment; causality/risk; time/version/snapshot; scenario; at least 10 invariants; minimum model set; blockers. Explicitly decide which candidates need independent roots and how WM-XCT-037's boundary contradictions should be resolved.