# EM-WRK-03 provider comparison

Claude and Grok agree on the primary boundary: reuse WM-ACT-029 through a mandatory Program discriminator, retain Portfolio as a disjoint sibling profile, keep ProgramComponent, Tranche and BenefitDependency as addressable program-owned assertions, profile Transition Plan on WM-ACT-008, preserve every component master and allocate no new identifier.

Both reviews separate Program from Portfolio, Project, Initiative and an arbitrary tracker container. They also agree that membership never transfers mastership; a tranche is not a project phase; a transition plan is not operational acceptance or sustainment; outputs, outcomes, benefits and attribution are different assertions; and component closure, program closure and benefit realization are independent lifecycle facts.

Grok sharpens two constraints. First, the WM-ACT-029 to WM-ACT-005 edge must be membership through ProgramComponent, never child-identity containment. A profile cannot cancel inherited containment, so the frozen registry contradiction remains a hard canonical hold. Second, BenefitDependency cannot substitute for the missing durable Outcome / Benefit Realization master. The prior EM-STR-02 checkpoint supplies an identifier-unassigned candidate, but EM-WRK-03 must treat it as an external unresolved dependency and must not allocate or master it.

The reconciled profile therefore requires a non-null immutable discriminator, at least one intended outcome or benefit target, non-cascading membership, historical assertion retention, independent versioning of program assertions and transition plans, and explicit references rather than duplicated plan or component facts. Readiness, acceptance and realization judgments reuse WM-ACT-034; role assignments reuse WM-ORG-016 without changing authority or mastership.

Publication remains held by the parent/containment contradiction, the unallocated Outcome / Benefit Realization dependency, unapproved relation rows and non-canonical base drafts. The result is a reviewable no-ID profile dossier with no installability or publication-readiness claim.

Third, the reviews diverge on the home of the causal hypothesis: the local synthesis and Claude place it in the program results chain, Grok places causal claims at most with WM-KNW-011 and excludes BenefitDependency. The reconciled profile assigns the hypothesis to the program results chain and the intended aim to WM-KNW-011.