Evaluation
The implemented evaluation assets include:
- at least 100 competency questions;
- executable Explorer journeys for overview, search, graph, links, timeline, type, resources and narrative views;
- four-nation and local-jurisdiction comparison cases;
- ordinary, exception, appeal and degraded-source paths;
- accessibility and plain-language review; and
- provenance checks that distinguish official, normalised, inferred and editorial-example assertions.
The first executable boundary is the set of three vertical-slice fixture contracts. They define synthetic ordinary and exception journeys and are validated locally against the draft domain profile. The owner approved this boundary on 2026-08-07.
The integrated three-slice Explorer review was rerun locally on 2026-08-07 against the exact consumer recorded under compatibility/. Reader, query, relationship and node provenance, browser source and licence/notice journeys pass; all four original findings are closed. The large-corpus review also passes 293-family search and seven colour facets locally.
The Pack 1 family-beginnings review adds 13 questions and proves alias search, per-record Narrative, family-centred Graph and active official-source handoffs across the first three domains. Pack completion remains local and does not make the corpus release-grade or public.
The Pack 2 learning-and-transition review adds 13 questions across school, transition and post-school study/research. It accounts for all 36 Pack 2 families while preserving the deeper learning-to-drive vertical slice and adds 44 current body-free link receipts.
The Pack 3 work-and-money review adds 36 families, 45 body-free official links and 13 questions across finding work, employment and money/tax/benefits, including explicit Great Britain and Northern Ireland employment boundaries.
The Pack 4 home/place/transport review accounts for all 36 families while preserving the deeper missed-rubbish slice, and adds 44 body-free official links plus 13 questions across housing, local environmental services and accessible or private transport.
The Pack 5 enforcement/consumer/justice review accounts for all 36 families while preserving the reviewed speeding-notice slice, and adds 44 body-free official links plus 13 questions across transport penalties, consumer remedies and police or adjudication routes.
The Pack 6 family/health/care review adds 36 dossiers, 45 body-free official links and 13 questions across family change, health throughout life and disability/care/support. It proves that manual health references remain typed, link-only resources and that clinical, capacity, safeguarding and legal decisions remain outside the bundle.
The Pack 7 civic/enterprise/creativity review adds 36 dossiers, 45 body-free official links and 13 questions across civic participation, organisation operation and ideas/research. It proves alias search, enclosing-process Narrative, family-centred Graph and typed official source access while retaining immigration, legal, tax, electoral and intellectual-property review boundaries.
The Pack 8 mobility/later-life/death review accounts for the final 41 approved identities through 38 compact dossiers and the three preserved death/notification/estate slices. It adds 47 body-free official links and 13 questions, bringing acceptance to 293 families, 104 questions and all 24 life-course domains.
The population assurance review reconciles the whole authored and generated corpus, freezes exact artefact hashes and records six representative visible-Explorer journeys. It establishes the local population-complete gate with zero blocking omissions while retaining the separate specialist-review and publication gates.
The population-preview publication authorisation records the explicit owner request, reconciled Explorer deployment, frozen candidate identity and non-release-grade publication scope. Deployment and public-browser verification require a separate evidence report.
Contract validation is not domain validation, and a local Explorer result is not publication authorisation.